diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,107128 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 20.0,
+  "global_step": 17840,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 4.9838,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 5.2327,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 5.0659,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 5.1148,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 4.9057,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 5.008,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 4.9586,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 4.8863,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 5.0875,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9e-07,
+      "loss": 4.9751,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 5.0414,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1e-06,
+      "loss": 5.0625,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 4.9253,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3e-06,
+      "loss": 4.9931,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3e-06,
+      "loss": 4.9128,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 5.0039,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5e-06,
+      "loss": 5.0044,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 5.0845,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 4.8552,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8e-06,
+      "loss": 4.9856,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9e-06,
+      "loss": 4.8844,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 4.8926,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 4.9609,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.2e-06,
+      "loss": 5.0059,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.3e-06,
+      "loss": 4.8663,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 5.0267,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5e-06,
+      "loss": 4.7886,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6e-06,
+      "loss": 4.7043,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7e-06,
+      "loss": 4.7329,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 4.9152,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9e-06,
+      "loss": 4.9241,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3e-06,
+      "loss": 4.6468,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.1e-06,
+      "loss": 4.7872,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 4.6226,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.3e-06,
+      "loss": 4.8247,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 4.7664,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.5000000000000004e-06,
+      "loss": 4.8099,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6e-06,
+      "loss": 4.7516,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7e-06,
+      "loss": 4.9987,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.8e-06,
+      "loss": 4.7503,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9e-06,
+      "loss": 4.6063,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 4.8159,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.1000000000000006e-06,
+      "loss": 4.6264,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 4.7633,
+      "step": 44
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.2999999999999995e-06,
+      "loss": 4.8731,
+      "step": 45
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.4e-06,
+      "loss": 5.0248,
+      "step": 46
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5e-06,
+      "loss": 4.869,
+      "step": 47
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6e-06,
+      "loss": 5.2137,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7e-06,
+      "loss": 5.3017,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 5.0329,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 4.5108,
+      "step": 51
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5e-06,
+      "loss": 4.6215,
+      "step": 52
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.1e-06,
+      "loss": 4.4346,
+      "step": 53
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.2e-06,
+      "loss": 4.5324,
+      "step": 54
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.3e-06,
+      "loss": 4.5609,
+      "step": 55
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.4e-06,
+      "loss": 4.6176,
+      "step": 56
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 4.6011,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 4.7913,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.7000000000000005e-06,
+      "loss": 4.5426,
+      "step": 59
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.8e-06,
+      "loss": 4.475,
+      "step": 60
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9e-06,
+      "loss": 4.624,
+      "step": 61
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6e-06,
+      "loss": 4.5521,
+      "step": 62
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.1e-06,
+      "loss": 4.3977,
+      "step": 63
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.2e-06,
+      "loss": 4.6049,
+      "step": 64
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 4.3664,
+      "step": 65
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 4.6367,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 4.3834,
+      "step": 67
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.6e-06,
+      "loss": 4.4498,
+      "step": 68
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 4.3949,
+      "step": 69
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 4.5359,
+      "step": 70
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.900000000000001e-06,
+      "loss": 4.6607,
+      "step": 71
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 4.5668,
+      "step": 72
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.1e-06,
+      "loss": 4.5201,
+      "step": 73
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.2e-06,
+      "loss": 4.5518,
+      "step": 74
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.2999999999999996e-06,
+      "loss": 4.3511,
+      "step": 75
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.4e-06,
+      "loss": 4.6297,
+      "step": 76
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.5e-06,
+      "loss": 4.5203,
+      "step": 77
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.6e-06,
+      "loss": 4.4192,
+      "step": 78
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.7e-06,
+      "loss": 4.2852,
+      "step": 79
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.8e-06,
+      "loss": 4.5105,
+      "step": 80
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.9e-06,
+      "loss": 4.5685,
+      "step": 81
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 4.4065,
+      "step": 82
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.1e-06,
+      "loss": 4.5752,
+      "step": 83
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.200000000000001e-06,
+      "loss": 4.5229,
+      "step": 84
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.3e-06,
+      "loss": 4.4704,
+      "step": 85
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 4.4494,
+      "step": 86
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.500000000000002e-06,
+      "loss": 4.4507,
+      "step": 87
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.599999999999999e-06,
+      "loss": 4.4451,
+      "step": 88
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.7e-06,
+      "loss": 4.5326,
+      "step": 89
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.8e-06,
+      "loss": 4.4269,
+      "step": 90
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.9e-06,
+      "loss": 4.6798,
+      "step": 91
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9e-06,
+      "loss": 4.5629,
+      "step": 92
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 4.2875,
+      "step": 93
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.2e-06,
+      "loss": 4.5126,
+      "step": 94
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.3e-06,
+      "loss": 4.7714,
+      "step": 95
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.4e-06,
+      "loss": 4.7541,
+      "step": 96
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.5e-06,
+      "loss": 4.4375,
+      "step": 97
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 4.7905,
+      "step": 98
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.7e-06,
+      "loss": 4.5313,
+      "step": 99
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 4.6112,
+      "step": 100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.900000000000002e-06,
+      "loss": 4.363,
+      "step": 101
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1e-05,
+      "loss": 4.3847,
+      "step": 102
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0100000000000002e-05,
+      "loss": 4.5182,
+      "step": 103
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.02e-05,
+      "loss": 4.4093,
+      "step": 104
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.03e-05,
+      "loss": 4.237,
+      "step": 105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.04e-05,
+      "loss": 4.4695,
+      "step": 106
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.05e-05,
+      "loss": 4.449,
+      "step": 107
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.06e-05,
+      "loss": 4.6313,
+      "step": 108
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0700000000000001e-05,
+      "loss": 4.1594,
+      "step": 109
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.08e-05,
+      "loss": 4.3165,
+      "step": 110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.09e-05,
+      "loss": 4.4963,
+      "step": 111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 4.3661,
+      "step": 112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.11e-05,
+      "loss": 4.3313,
+      "step": 113
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 4.2762,
+      "step": 114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.13e-05,
+      "loss": 4.3184,
+      "step": 115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 4.1543,
+      "step": 116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1500000000000002e-05,
+      "loss": 4.439,
+      "step": 117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.16e-05,
+      "loss": 4.2838,
+      "step": 118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1700000000000001e-05,
+      "loss": 4.4891,
+      "step": 119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.18e-05,
+      "loss": 4.4837,
+      "step": 120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.19e-05,
+      "loss": 4.4199,
+      "step": 121
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.2e-05,
+      "loss": 4.359,
+      "step": 122
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.2100000000000001e-05,
+      "loss": 4.2646,
+      "step": 123
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.22e-05,
+      "loss": 4.4685,
+      "step": 124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.23e-05,
+      "loss": 4.4269,
+      "step": 125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.24e-05,
+      "loss": 4.4925,
+      "step": 126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.25e-05,
+      "loss": 4.554,
+      "step": 127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 4.3696,
+      "step": 128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.27e-05,
+      "loss": 4.303,
+      "step": 129
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 4.5032,
+      "step": 130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.29e-05,
+      "loss": 4.386,
+      "step": 131
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 4.4807,
+      "step": 132
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3100000000000002e-05,
+      "loss": 4.4106,
+      "step": 133
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.32e-05,
+      "loss": 4.4578,
+      "step": 134
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3300000000000001e-05,
+      "loss": 4.3066,
+      "step": 135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 4.3413,
+      "step": 136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 4.7065,
+      "step": 137
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 4.4404,
+      "step": 138
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.3700000000000001e-05,
+      "loss": 4.4617,
+      "step": 139
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 4.246,
+      "step": 140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.3900000000000002e-05,
+      "loss": 4.5735,
+      "step": 141
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 4.6352,
+      "step": 142
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 4.5829,
+      "step": 143
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.42e-05,
+      "loss": 4.4466,
+      "step": 144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.43e-05,
+      "loss": 4.5799,
+      "step": 145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.44e-05,
+      "loss": 4.5775,
+      "step": 146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.45e-05,
+      "loss": 4.5871,
+      "step": 147
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4599999999999999e-05,
+      "loss": 4.3921,
+      "step": 148
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.47e-05,
+      "loss": 4.7372,
+      "step": 149
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.48e-05,
+      "loss": 4.3184,
+      "step": 150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.49e-05,
+      "loss": 4.477,
+      "step": 151
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5e-05,
+      "loss": 4.3332,
+      "step": 152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.51e-05,
+      "loss": 4.4186,
+      "step": 153
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.52e-05,
+      "loss": 4.2746,
+      "step": 154
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.53e-05,
+      "loss": 4.2408,
+      "step": 155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.54e-05,
+      "loss": 4.2963,
+      "step": 156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.55e-05,
+      "loss": 4.4445,
+      "step": 157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.56e-05,
+      "loss": 4.5764,
+      "step": 158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5700000000000002e-05,
+      "loss": 4.593,
+      "step": 159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.58e-05,
+      "loss": 4.3032,
+      "step": 160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.59e-05,
+      "loss": 4.3158,
+      "step": 161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 4.4078,
+      "step": 162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6100000000000002e-05,
+      "loss": 4.3443,
+      "step": 163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.62e-05,
+      "loss": 4.4846,
+      "step": 164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.63e-05,
+      "loss": 4.4033,
+      "step": 165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6400000000000002e-05,
+      "loss": 4.444,
+      "step": 166
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.65e-05,
+      "loss": 4.3711,
+      "step": 167
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.66e-05,
+      "loss": 4.1415,
+      "step": 168
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6700000000000003e-05,
+      "loss": 4.1865,
+      "step": 169
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 4.3506,
+      "step": 170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.69e-05,
+      "loss": 4.5146,
+      "step": 171
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 4.4737,
+      "step": 172
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.7100000000000002e-05,
+      "loss": 4.3658,
+      "step": 173
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7199999999999998e-05,
+      "loss": 4.2463,
+      "step": 174
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.73e-05,
+      "loss": 4.3311,
+      "step": 175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.74e-05,
+      "loss": 4.158,
+      "step": 176
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.75e-05,
+      "loss": 4.236,
+      "step": 177
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.76e-05,
+      "loss": 4.2928,
+      "step": 178
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.77e-05,
+      "loss": 4.383,
+      "step": 179
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.78e-05,
+      "loss": 4.3843,
+      "step": 180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.79e-05,
+      "loss": 4.4321,
+      "step": 181
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8e-05,
+      "loss": 4.2168,
+      "step": 182
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.81e-05,
+      "loss": 4.4573,
+      "step": 183
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 4.3832,
+      "step": 184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83e-05,
+      "loss": 4.3981,
+      "step": 185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.84e-05,
+      "loss": 4.3146,
+      "step": 186
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.85e-05,
+      "loss": 4.5188,
+      "step": 187
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.86e-05,
+      "loss": 4.2181,
+      "step": 188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.87e-05,
+      "loss": 4.3208,
+      "step": 189
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.88e-05,
+      "loss": 4.4359,
+      "step": 190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8900000000000002e-05,
+      "loss": 4.4542,
+      "step": 191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9e-05,
+      "loss": 4.4263,
+      "step": 192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.91e-05,
+      "loss": 4.6343,
+      "step": 193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 4.3283,
+      "step": 194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.93e-05,
+      "loss": 4.3409,
+      "step": 195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.94e-05,
+      "loss": 4.6093,
+      "step": 196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9500000000000003e-05,
+      "loss": 4.6184,
+      "step": 197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 4.5437,
+      "step": 198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.97e-05,
+      "loss": 4.6958,
+      "step": 199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 4.7785,
+      "step": 200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9900000000000003e-05,
+      "loss": 4.1794,
+      "step": 201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 4.3105,
+      "step": 202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.01e-05,
+      "loss": 4.1893,
+      "step": 203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0200000000000003e-05,
+      "loss": 4.2441,
+      "step": 204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0300000000000002e-05,
+      "loss": 4.1452,
+      "step": 205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.04e-05,
+      "loss": 4.1601,
+      "step": 206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.05e-05,
+      "loss": 4.226,
+      "step": 207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.06e-05,
+      "loss": 4.116,
+      "step": 208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.07e-05,
+      "loss": 4.3524,
+      "step": 209
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.08e-05,
+      "loss": 4.2692,
+      "step": 210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.09e-05,
+      "loss": 4.2094,
+      "step": 211
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.1e-05,
+      "loss": 4.3623,
+      "step": 212
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.11e-05,
+      "loss": 4.2965,
+      "step": 213
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.12e-05,
+      "loss": 4.3382,
+      "step": 214
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.13e-05,
+      "loss": 4.2327,
+      "step": 215
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.1400000000000002e-05,
+      "loss": 4.3403,
+      "step": 216
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.15e-05,
+      "loss": 4.3522,
+      "step": 217
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.16e-05,
+      "loss": 4.2744,
+      "step": 218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.1700000000000002e-05,
+      "loss": 4.3475,
+      "step": 219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.18e-05,
+      "loss": 4.3136,
+      "step": 220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.19e-05,
+      "loss": 4.2441,
+      "step": 221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 4.3483,
+      "step": 222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2100000000000002e-05,
+      "loss": 4.4063,
+      "step": 223
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.22e-05,
+      "loss": 4.285,
+      "step": 224
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.23e-05,
+      "loss": 4.4122,
+      "step": 225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 4.5432,
+      "step": 226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.25e-05,
+      "loss": 4.3215,
+      "step": 227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.26e-05,
+      "loss": 4.2291,
+      "step": 228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2700000000000003e-05,
+      "loss": 4.2839,
+      "step": 229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 4.3239,
+      "step": 230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.29e-05,
+      "loss": 4.2715,
+      "step": 231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 4.4176,
+      "step": 232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.3100000000000002e-05,
+      "loss": 4.2666,
+      "step": 233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.32e-05,
+      "loss": 4.3716,
+      "step": 234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.3300000000000004e-05,
+      "loss": 4.453,
+      "step": 235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 4.3051,
+      "step": 236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.35e-05,
+      "loss": 4.3399,
+      "step": 237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.36e-05,
+      "loss": 4.5725,
+      "step": 238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.37e-05,
+      "loss": 4.3398,
+      "step": 239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.38e-05,
+      "loss": 4.4554,
+      "step": 240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.39e-05,
+      "loss": 4.4346,
+      "step": 241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.4e-05,
+      "loss": 4.5572,
+      "step": 242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.41e-05,
+      "loss": 4.4308,
+      "step": 243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.4200000000000002e-05,
+      "loss": 4.3764,
+      "step": 244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.43e-05,
+      "loss": 4.5323,
+      "step": 245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.44e-05,
+      "loss": 4.4945,
+      "step": 246
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.45e-05,
+      "loss": 4.8437,
+      "step": 247
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.46e-05,
+      "loss": 4.8184,
+      "step": 248
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.47e-05,
+      "loss": 4.6514,
+      "step": 249
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.48e-05,
+      "loss": 4.3213,
+      "step": 250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.4900000000000002e-05,
+      "loss": 4.505,
+      "step": 251
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5e-05,
+      "loss": 4.1797,
+      "step": 252
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.51e-05,
+      "loss": 4.2116,
+      "step": 253
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5200000000000003e-05,
+      "loss": 4.3823,
+      "step": 254
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5300000000000002e-05,
+      "loss": 4.2173,
+      "step": 255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.54e-05,
+      "loss": 4.2465,
+      "step": 256
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5500000000000003e-05,
+      "loss": 4.3771,
+      "step": 257
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 4.2258,
+      "step": 258
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.57e-05,
+      "loss": 4.2178,
+      "step": 259
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.58e-05,
+      "loss": 4.3779,
+      "step": 260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5900000000000003e-05,
+      "loss": 4.3978,
+      "step": 261
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 4.1387,
+      "step": 262
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.61e-05,
+      "loss": 4.2491,
+      "step": 263
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6200000000000003e-05,
+      "loss": 4.2279,
+      "step": 264
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6300000000000002e-05,
+      "loss": 4.3271,
+      "step": 265
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.64e-05,
+      "loss": 4.3932,
+      "step": 266
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6500000000000004e-05,
+      "loss": 4.24,
+      "step": 267
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6600000000000003e-05,
+      "loss": 4.5169,
+      "step": 268
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6700000000000002e-05,
+      "loss": 4.3127,
+      "step": 269
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6800000000000004e-05,
+      "loss": 4.4424,
+      "step": 270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6900000000000003e-05,
+      "loss": 4.0541,
+      "step": 271
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 4.2222,
+      "step": 272
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7100000000000005e-05,
+      "loss": 4.2296,
+      "step": 273
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 4.172,
+      "step": 274
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7300000000000003e-05,
+      "loss": 4.404,
+      "step": 275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7400000000000002e-05,
+      "loss": 4.3539,
+      "step": 276
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 4.2925,
+      "step": 277
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 4.2352,
+      "step": 278
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7700000000000002e-05,
+      "loss": 4.2552,
+      "step": 279
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 4.3908,
+      "step": 280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7900000000000004e-05,
+      "loss": 4.1687,
+      "step": 281
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 4.3196,
+      "step": 282
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8100000000000005e-05,
+      "loss": 4.0877,
+      "step": 283
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 4.3293,
+      "step": 284
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.83e-05,
+      "loss": 4.3693,
+      "step": 285
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.84e-05,
+      "loss": 4.2878,
+      "step": 286
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 4.4492,
+      "step": 287
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.86e-05,
+      "loss": 4.3742,
+      "step": 288
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.87e-05,
+      "loss": 4.4633,
+      "step": 289
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.88e-05,
+      "loss": 4.4401,
+      "step": 290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8899999999999998e-05,
+      "loss": 4.4441,
+      "step": 291
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9e-05,
+      "loss": 4.3023,
+      "step": 292
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.91e-05,
+      "loss": 4.462,
+      "step": 293
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9199999999999998e-05,
+      "loss": 4.3695,
+      "step": 294
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.93e-05,
+      "loss": 4.2608,
+      "step": 295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.94e-05,
+      "loss": 4.5505,
+      "step": 296
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.95e-05,
+      "loss": 4.6193,
+      "step": 297
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.96e-05,
+      "loss": 4.4659,
+      "step": 298
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.97e-05,
+      "loss": 4.8052,
+      "step": 299
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.98e-05,
+      "loss": 4.8301,
+      "step": 300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9900000000000002e-05,
+      "loss": 4.1515,
+      "step": 301
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3e-05,
+      "loss": 4.2716,
+      "step": 302
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.01e-05,
+      "loss": 4.2277,
+      "step": 303
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.02e-05,
+      "loss": 4.3596,
+      "step": 304
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.03e-05,
+      "loss": 4.1543,
+      "step": 305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.04e-05,
+      "loss": 4.2093,
+      "step": 306
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.05e-05,
+      "loss": 4.2674,
+      "step": 307
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.06e-05,
+      "loss": 4.3471,
+      "step": 308
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.07e-05,
+      "loss": 4.2533,
+      "step": 309
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.08e-05,
+      "loss": 4.3865,
+      "step": 310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.09e-05,
+      "loss": 4.4996,
+      "step": 311
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.1e-05,
+      "loss": 4.4109,
+      "step": 312
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.1100000000000004e-05,
+      "loss": 4.2948,
+      "step": 313
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.12e-05,
+      "loss": 4.3084,
+      "step": 314
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.13e-05,
+      "loss": 4.175,
+      "step": 315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.1400000000000004e-05,
+      "loss": 4.3797,
+      "step": 316
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.15e-05,
+      "loss": 4.196,
+      "step": 317
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.16e-05,
+      "loss": 4.2061,
+      "step": 318
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1700000000000005e-05,
+      "loss": 4.3233,
+      "step": 319
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.18e-05,
+      "loss": 4.2135,
+      "step": 320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.19e-05,
+      "loss": 4.356,
+      "step": 321
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 4.2659,
+      "step": 322
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.21e-05,
+      "loss": 4.2197,
+      "step": 323
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2200000000000003e-05,
+      "loss": 4.3643,
+      "step": 324
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2300000000000006e-05,
+      "loss": 4.3829,
+      "step": 325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.24e-05,
+      "loss": 4.4073,
+      "step": 326
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 4.2241,
+      "step": 327
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.26e-05,
+      "loss": 4.2472,
+      "step": 328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.27e-05,
+      "loss": 4.4192,
+      "step": 329
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.2800000000000004e-05,
+      "loss": 4.5124,
+      "step": 330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.29e-05,
+      "loss": 4.2705,
+      "step": 331
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.3e-05,
+      "loss": 4.3056,
+      "step": 332
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.3100000000000005e-05,
+      "loss": 4.2735,
+      "step": 333
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.32e-05,
+      "loss": 4.3437,
+      "step": 334
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.33e-05,
+      "loss": 4.4882,
+      "step": 335
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3400000000000005e-05,
+      "loss": 4.2491,
+      "step": 336
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.35e-05,
+      "loss": 4.1632,
+      "step": 337
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 4.417,
+      "step": 338
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3700000000000006e-05,
+      "loss": 4.2131,
+      "step": 339
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.38e-05,
+      "loss": 4.3315,
+      "step": 340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3900000000000004e-05,
+      "loss": 4.2924,
+      "step": 341
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 4.422,
+      "step": 342
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.41e-05,
+      "loss": 4.3567,
+      "step": 343
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4200000000000005e-05,
+      "loss": 4.4224,
+      "step": 344
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.430000000000001e-05,
+      "loss": 4.4397,
+      "step": 345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 4.6309,
+      "step": 346
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.45e-05,
+      "loss": 4.5586,
+      "step": 347
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.46e-05,
+      "loss": 4.5247,
+      "step": 348
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4699999999999996e-05,
+      "loss": 4.6826,
+      "step": 349
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.48e-05,
+      "loss": 4.6195,
+      "step": 350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.49e-05,
+      "loss": 4.2817,
+      "step": 351
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5e-05,
+      "loss": 4.2991,
+      "step": 352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.51e-05,
+      "loss": 4.2362,
+      "step": 353
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.52e-05,
+      "loss": 4.252,
+      "step": 354
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.53e-05,
+      "loss": 4.4083,
+      "step": 355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.54e-05,
+      "loss": 4.1247,
+      "step": 356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.55e-05,
+      "loss": 4.1842,
+      "step": 357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.56e-05,
+      "loss": 4.257,
+      "step": 358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.57e-05,
+      "loss": 4.2429,
+      "step": 359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.58e-05,
+      "loss": 4.4186,
+      "step": 360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.59e-05,
+      "loss": 4.2737,
+      "step": 361
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6e-05,
+      "loss": 4.2433,
+      "step": 362
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.61e-05,
+      "loss": 4.1903,
+      "step": 363
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.62e-05,
+      "loss": 4.3808,
+      "step": 364
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.63e-05,
+      "loss": 4.1352,
+      "step": 365
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6400000000000004e-05,
+      "loss": 4.2693,
+      "step": 366
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.65e-05,
+      "loss": 4.211,
+      "step": 367
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.66e-05,
+      "loss": 4.3071,
+      "step": 368
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6700000000000004e-05,
+      "loss": 4.4047,
+      "step": 369
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.68e-05,
+      "loss": 4.3532,
+      "step": 370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.69e-05,
+      "loss": 4.2028,
+      "step": 371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.7e-05,
+      "loss": 4.4709,
+      "step": 372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.71e-05,
+      "loss": 4.2867,
+      "step": 373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.72e-05,
+      "loss": 4.3559,
+      "step": 374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.73e-05,
+      "loss": 4.1859,
+      "step": 375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.74e-05,
+      "loss": 4.3583,
+      "step": 376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 4.1307,
+      "step": 377
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.76e-05,
+      "loss": 4.4598,
+      "step": 378
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.77e-05,
+      "loss": 4.3065,
+      "step": 379
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 4.4477,
+      "step": 380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.79e-05,
+      "loss": 4.3751,
+      "step": 381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.8e-05,
+      "loss": 4.2969,
+      "step": 382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.8100000000000005e-05,
+      "loss": 4.3815,
+      "step": 383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.82e-05,
+      "loss": 4.3073,
+      "step": 384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.83e-05,
+      "loss": 4.276,
+      "step": 385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 4.3101,
+      "step": 386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.85e-05,
+      "loss": 4.2662,
+      "step": 387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.86e-05,
+      "loss": 4.1635,
+      "step": 388
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.8700000000000006e-05,
+      "loss": 4.2496,
+      "step": 389
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.88e-05,
+      "loss": 4.4039,
+      "step": 390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.8900000000000004e-05,
+      "loss": 4.4787,
+      "step": 391
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 4.1902,
+      "step": 392
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.91e-05,
+      "loss": 4.3323,
+      "step": 393
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9200000000000004e-05,
+      "loss": 4.4308,
+      "step": 394
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9300000000000007e-05,
+      "loss": 4.6842,
+      "step": 395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.94e-05,
+      "loss": 4.3799,
+      "step": 396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9500000000000005e-05,
+      "loss": 4.3979,
+      "step": 397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.960000000000001e-05,
+      "loss": 4.7533,
+      "step": 398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.97e-05,
+      "loss": 4.5113,
+      "step": 399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 4.6379,
+      "step": 400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.99e-05,
+      "loss": 4.4328,
+      "step": 401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4e-05,
+      "loss": 4.2678,
+      "step": 402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.0100000000000006e-05,
+      "loss": 4.3898,
+      "step": 403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.02e-05,
+      "loss": 4.339,
+      "step": 404
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.0300000000000004e-05,
+      "loss": 4.3729,
+      "step": 405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.0400000000000006e-05,
+      "loss": 4.3148,
+      "step": 406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.05e-05,
+      "loss": 4.2553,
+      "step": 407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.0600000000000004e-05,
+      "loss": 4.2059,
+      "step": 408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.07e-05,
+      "loss": 4.4559,
+      "step": 409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.08e-05,
+      "loss": 4.3311,
+      "step": 410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.09e-05,
+      "loss": 4.1914,
+      "step": 411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.1e-05,
+      "loss": 4.3019,
+      "step": 412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.11e-05,
+      "loss": 4.3914,
+      "step": 413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.12e-05,
+      "loss": 4.1375,
+      "step": 414
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.13e-05,
+      "loss": 4.1963,
+      "step": 415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.14e-05,
+      "loss": 4.4056,
+      "step": 416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.15e-05,
+      "loss": 4.2072,
+      "step": 417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.16e-05,
+      "loss": 4.3547,
+      "step": 418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.17e-05,
+      "loss": 4.2649,
+      "step": 419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.18e-05,
+      "loss": 4.2984,
+      "step": 420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.19e-05,
+      "loss": 4.3149,
+      "step": 421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2e-05,
+      "loss": 4.3728,
+      "step": 422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.21e-05,
+      "loss": 4.333,
+      "step": 423
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.22e-05,
+      "loss": 4.363,
+      "step": 424
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.23e-05,
+      "loss": 4.3812,
+      "step": 425
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.24e-05,
+      "loss": 4.4311,
+      "step": 426
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.25e-05,
+      "loss": 4.2849,
+      "step": 427
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.26e-05,
+      "loss": 4.4876,
+      "step": 428
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.27e-05,
+      "loss": 4.2629,
+      "step": 429
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2800000000000004e-05,
+      "loss": 4.392,
+      "step": 430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.29e-05,
+      "loss": 4.539,
+      "step": 431
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.3e-05,
+      "loss": 4.3672,
+      "step": 432
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.3100000000000004e-05,
+      "loss": 4.0878,
+      "step": 433
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.32e-05,
+      "loss": 4.3161,
+      "step": 434
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.33e-05,
+      "loss": 4.2642,
+      "step": 435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.3400000000000005e-05,
+      "loss": 4.5003,
+      "step": 436
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.35e-05,
+      "loss": 4.4681,
+      "step": 437
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.36e-05,
+      "loss": 4.378,
+      "step": 438
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.3700000000000005e-05,
+      "loss": 4.2554,
+      "step": 439
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.38e-05,
+      "loss": 4.3479,
+      "step": 440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.39e-05,
+      "loss": 4.3922,
+      "step": 441
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 4.2615,
+      "step": 442
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.41e-05,
+      "loss": 4.337,
+      "step": 443
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.4200000000000004e-05,
+      "loss": 4.5572,
+      "step": 444
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.43e-05,
+      "loss": 4.3559,
+      "step": 445
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.44e-05,
+      "loss": 4.1737,
+      "step": 446
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.4500000000000004e-05,
+      "loss": 4.3339,
+      "step": 447
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.46e-05,
+      "loss": 4.5056,
+      "step": 448
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.47e-05,
+      "loss": 4.5106,
+      "step": 449
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.4800000000000005e-05,
+      "loss": 5.0764,
+      "step": 450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.49e-05,
+      "loss": 4.3769,
+      "step": 451
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5e-05,
+      "loss": 4.2846,
+      "step": 452
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5100000000000005e-05,
+      "loss": 4.5215,
+      "step": 453
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.52e-05,
+      "loss": 4.1522,
+      "step": 454
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.53e-05,
+      "loss": 4.4702,
+      "step": 455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5400000000000006e-05,
+      "loss": 4.402,
+      "step": 456
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.55e-05,
+      "loss": 4.3347,
+      "step": 457
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 4.3137,
+      "step": 458
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5700000000000006e-05,
+      "loss": 4.3229,
+      "step": 459
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.58e-05,
+      "loss": 4.4458,
+      "step": 460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.5900000000000004e-05,
+      "loss": 4.237,
+      "step": 461
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 4.2437,
+      "step": 462
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.61e-05,
+      "loss": 4.4029,
+      "step": 463
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6200000000000005e-05,
+      "loss": 4.2168,
+      "step": 464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.630000000000001e-05,
+      "loss": 4.3064,
+      "step": 465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.64e-05,
+      "loss": 4.3616,
+      "step": 466
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6500000000000005e-05,
+      "loss": 4.2622,
+      "step": 467
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.660000000000001e-05,
+      "loss": 4.2298,
+      "step": 468
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6700000000000003e-05,
+      "loss": 4.3436,
+      "step": 469
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6800000000000006e-05,
+      "loss": 4.3518,
+      "step": 470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.69e-05,
+      "loss": 4.1647,
+      "step": 471
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7e-05,
+      "loss": 4.2264,
+      "step": 472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.71e-05,
+      "loss": 4.7449,
+      "step": 473
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.72e-05,
+      "loss": 4.4482,
+      "step": 474
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.73e-05,
+      "loss": 4.2871,
+      "step": 475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.74e-05,
+      "loss": 4.5033,
+      "step": 476
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.75e-05,
+      "loss": 4.5866,
+      "step": 477
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.76e-05,
+      "loss": 4.2064,
+      "step": 478
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.77e-05,
+      "loss": 4.1415,
+      "step": 479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.78e-05,
+      "loss": 4.3966,
+      "step": 480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.79e-05,
+      "loss": 4.2096,
+      "step": 481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8e-05,
+      "loss": 4.2843,
+      "step": 482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8100000000000004e-05,
+      "loss": 4.4285,
+      "step": 483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.82e-05,
+      "loss": 4.3456,
+      "step": 484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.83e-05,
+      "loss": 4.4676,
+      "step": 485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8400000000000004e-05,
+      "loss": 4.202,
+      "step": 486
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.85e-05,
+      "loss": 4.1047,
+      "step": 487
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.86e-05,
+      "loss": 4.2552,
+      "step": 488
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.87e-05,
+      "loss": 4.4285,
+      "step": 489
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.88e-05,
+      "loss": 4.4068,
+      "step": 490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.89e-05,
+      "loss": 4.1897,
+      "step": 491
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9e-05,
+      "loss": 4.4892,
+      "step": 492
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.91e-05,
+      "loss": 4.5035,
+      "step": 493
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.92e-05,
+      "loss": 4.3042,
+      "step": 494
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.93e-05,
+      "loss": 4.504,
+      "step": 495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.94e-05,
+      "loss": 4.2335,
+      "step": 496
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9500000000000004e-05,
+      "loss": 4.1033,
+      "step": 497
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.96e-05,
+      "loss": 4.846,
+      "step": 498
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.97e-05,
+      "loss": 4.7581,
+      "step": 499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 4.2613,
+      "step": 500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.99e-05,
+      "loss": 4.6527,
+      "step": 501
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5e-05,
+      "loss": 4.2356,
+      "step": 502
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.0100000000000005e-05,
+      "loss": 4.4912,
+      "step": 503
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.02e-05,
+      "loss": 4.3186,
+      "step": 504
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.03e-05,
+      "loss": 4.2636,
+      "step": 505
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.0400000000000005e-05,
+      "loss": 4.3494,
+      "step": 506
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.05e-05,
+      "loss": 4.4616,
+      "step": 507
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.0600000000000003e-05,
+      "loss": 4.4479,
+      "step": 508
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.0700000000000006e-05,
+      "loss": 4.3084,
+      "step": 509
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.08e-05,
+      "loss": 4.4979,
+      "step": 510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.0900000000000004e-05,
+      "loss": 4.3494,
+      "step": 511
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.1000000000000006e-05,
+      "loss": 4.3937,
+      "step": 512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.11e-05,
+      "loss": 4.5938,
+      "step": 513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.1200000000000004e-05,
+      "loss": 4.3197,
+      "step": 514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.130000000000001e-05,
+      "loss": 4.4052,
+      "step": 515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.14e-05,
+      "loss": 4.1644,
+      "step": 516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.1500000000000005e-05,
+      "loss": 4.4265,
+      "step": 517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.16e-05,
+      "loss": 4.3283,
+      "step": 518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.17e-05,
+      "loss": 4.5096,
+      "step": 519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.1800000000000005e-05,
+      "loss": 4.4203,
+      "step": 520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.19e-05,
+      "loss": 4.346,
+      "step": 521
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 4.2322,
+      "step": 522
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.2100000000000006e-05,
+      "loss": 4.281,
+      "step": 523
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.22e-05,
+      "loss": 4.3095,
+      "step": 524
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.2300000000000004e-05,
+      "loss": 4.5652,
+      "step": 525
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.2400000000000007e-05,
+      "loss": 4.2078,
+      "step": 526
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.25e-05,
+      "loss": 4.3638,
+      "step": 527
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.2600000000000005e-05,
+      "loss": 4.3545,
+      "step": 528
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.270000000000001e-05,
+      "loss": 4.3525,
+      "step": 529
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.28e-05,
+      "loss": 4.5453,
+      "step": 530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.2900000000000005e-05,
+      "loss": 4.4939,
+      "step": 531
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.300000000000001e-05,
+      "loss": 4.0817,
+      "step": 532
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.31e-05,
+      "loss": 4.2966,
+      "step": 533
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.3200000000000006e-05,
+      "loss": 4.1715,
+      "step": 534
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.330000000000001e-05,
+      "loss": 4.3798,
+      "step": 535
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.3400000000000004e-05,
+      "loss": 4.438,
+      "step": 536
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.3500000000000006e-05,
+      "loss": 4.5099,
+      "step": 537
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.360000000000001e-05,
+      "loss": 4.5752,
+      "step": 538
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.3700000000000004e-05,
+      "loss": 4.2976,
+      "step": 539
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.380000000000001e-05,
+      "loss": 4.463,
+      "step": 540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.390000000000001e-05,
+      "loss": 4.429,
+      "step": 541
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 4.4518,
+      "step": 542
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.410000000000001e-05,
+      "loss": 4.5849,
+      "step": 543
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.420000000000001e-05,
+      "loss": 4.5696,
+      "step": 544
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.4300000000000005e-05,
+      "loss": 4.645,
+      "step": 545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.440000000000001e-05,
+      "loss": 4.6736,
+      "step": 546
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.45e-05,
+      "loss": 4.8372,
+      "step": 547
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.4600000000000006e-05,
+      "loss": 4.9467,
+      "step": 548
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.470000000000001e-05,
+      "loss": 4.2428,
+      "step": 549
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.4800000000000004e-05,
+      "loss": 5.8854,
+      "step": 550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.4900000000000006e-05,
+      "loss": 4.2545,
+      "step": 551
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 4.4878,
+      "step": 552
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.5100000000000004e-05,
+      "loss": 4.4589,
+      "step": 553
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.520000000000001e-05,
+      "loss": 4.3166,
+      "step": 554
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.530000000000001e-05,
+      "loss": 4.2724,
+      "step": 555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.5400000000000005e-05,
+      "loss": 4.3117,
+      "step": 556
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.550000000000001e-05,
+      "loss": 4.1779,
+      "step": 557
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.560000000000001e-05,
+      "loss": 4.3345,
+      "step": 558
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.5700000000000005e-05,
+      "loss": 4.2961,
+      "step": 559
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.580000000000001e-05,
+      "loss": 4.2034,
+      "step": 560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.590000000000001e-05,
+      "loss": 4.463,
+      "step": 561
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 4.2215,
+      "step": 562
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.610000000000001e-05,
+      "loss": 4.2534,
+      "step": 563
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.620000000000001e-05,
+      "loss": 4.2875,
+      "step": 564
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.63e-05,
+      "loss": 4.3902,
+      "step": 565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 4.4239,
+      "step": 566
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.65e-05,
+      "loss": 4.3055,
+      "step": 567
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.66e-05,
+      "loss": 4.2984,
+      "step": 568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6699999999999996e-05,
+      "loss": 4.3919,
+      "step": 569
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.68e-05,
+      "loss": 4.3509,
+      "step": 570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.69e-05,
+      "loss": 4.3028,
+      "step": 571
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 4.3882,
+      "step": 572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.71e-05,
+      "loss": 4.2948,
+      "step": 573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.72e-05,
+      "loss": 4.1845,
+      "step": 574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.73e-05,
+      "loss": 4.5676,
+      "step": 575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.74e-05,
+      "loss": 4.2202,
+      "step": 576
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7499999999999995e-05,
+      "loss": 4.3046,
+      "step": 577
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.76e-05,
+      "loss": 4.5347,
+      "step": 578
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.77e-05,
+      "loss": 3.9755,
+      "step": 579
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7799999999999995e-05,
+      "loss": 4.5953,
+      "step": 580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.79e-05,
+      "loss": 4.4368,
+      "step": 581
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8e-05,
+      "loss": 4.2276,
+      "step": 582
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8099999999999996e-05,
+      "loss": 4.3176,
+      "step": 583
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.82e-05,
+      "loss": 4.3604,
+      "step": 584
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.83e-05,
+      "loss": 4.5786,
+      "step": 585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.8399999999999997e-05,
+      "loss": 4.5632,
+      "step": 586
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.85e-05,
+      "loss": 4.5521,
+      "step": 587
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.86e-05,
+      "loss": 4.3586,
+      "step": 588
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.87e-05,
+      "loss": 4.2613,
+      "step": 589
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.88e-05,
+      "loss": 4.2485,
+      "step": 590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.89e-05,
+      "loss": 4.3518,
+      "step": 591
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9e-05,
+      "loss": 4.4674,
+      "step": 592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.91e-05,
+      "loss": 4.8796,
+      "step": 593
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.92e-05,
+      "loss": 4.4078,
+      "step": 594
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.93e-05,
+      "loss": 4.5768,
+      "step": 595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.94e-05,
+      "loss": 4.216,
+      "step": 596
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.95e-05,
+      "loss": 4.3493,
+      "step": 597
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.96e-05,
+      "loss": 4.5981,
+      "step": 598
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.97e-05,
+      "loss": 4.7776,
+      "step": 599
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9800000000000003e-05,
+      "loss": 4.3111,
+      "step": 600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.99e-05,
+      "loss": 4.5018,
+      "step": 601
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6e-05,
+      "loss": 4.3088,
+      "step": 602
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.0100000000000004e-05,
+      "loss": 4.3427,
+      "step": 603
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.02e-05,
+      "loss": 4.3183,
+      "step": 604
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.03e-05,
+      "loss": 4.3999,
+      "step": 605
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.04e-05,
+      "loss": 4.2787,
+      "step": 606
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.05e-05,
+      "loss": 4.4824,
+      "step": 607
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.06e-05,
+      "loss": 4.3904,
+      "step": 608
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.07e-05,
+      "loss": 4.3128,
+      "step": 609
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.08e-05,
+      "loss": 4.4125,
+      "step": 610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.09e-05,
+      "loss": 4.3113,
+      "step": 611
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.1e-05,
+      "loss": 4.47,
+      "step": 612
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.110000000000001e-05,
+      "loss": 4.313,
+      "step": 613
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.12e-05,
+      "loss": 4.7422,
+      "step": 614
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.13e-05,
+      "loss": 4.5166,
+      "step": 615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.14e-05,
+      "loss": 4.3098,
+      "step": 616
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.15e-05,
+      "loss": 4.3229,
+      "step": 617
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.16e-05,
+      "loss": 4.1977,
+      "step": 618
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.170000000000001e-05,
+      "loss": 4.2407,
+      "step": 619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.18e-05,
+      "loss": 4.5572,
+      "step": 620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.19e-05,
+      "loss": 4.4847,
+      "step": 621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.2e-05,
+      "loss": 4.1931,
+      "step": 622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.21e-05,
+      "loss": 4.4773,
+      "step": 623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.220000000000001e-05,
+      "loss": 4.3577,
+      "step": 624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.23e-05,
+      "loss": 4.6452,
+      "step": 625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.24e-05,
+      "loss": 4.4528,
+      "step": 626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.25e-05,
+      "loss": 4.3628,
+      "step": 627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.26e-05,
+      "loss": 4.2851,
+      "step": 628
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.27e-05,
+      "loss": 4.3031,
+      "step": 629
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.280000000000001e-05,
+      "loss": 4.6898,
+      "step": 630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.29e-05,
+      "loss": 4.3564,
+      "step": 631
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.3e-05,
+      "loss": 4.2896,
+      "step": 632
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.31e-05,
+      "loss": 4.4209,
+      "step": 633
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.32e-05,
+      "loss": 4.3634,
+      "step": 634
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.330000000000001e-05,
+      "loss": 4.4859,
+      "step": 635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.340000000000001e-05,
+      "loss": 4.4461,
+      "step": 636
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.35e-05,
+      "loss": 4.2736,
+      "step": 637
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.36e-05,
+      "loss": 4.2203,
+      "step": 638
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.37e-05,
+      "loss": 4.4757,
+      "step": 639
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.38e-05,
+      "loss": 4.4022,
+      "step": 640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.390000000000001e-05,
+      "loss": 4.2927,
+      "step": 641
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 4.7692,
+      "step": 642
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.41e-05,
+      "loss": 4.815,
+      "step": 643
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.42e-05,
+      "loss": 4.3132,
+      "step": 644
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.43e-05,
+      "loss": 4.2775,
+      "step": 645
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.440000000000001e-05,
+      "loss": 4.5218,
+      "step": 646
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.450000000000001e-05,
+      "loss": 4.3002,
+      "step": 647
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.460000000000001e-05,
+      "loss": 4.4016,
+      "step": 648
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.47e-05,
+      "loss": 4.5974,
+      "step": 649
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.48e-05,
+      "loss": 4.3331,
+      "step": 650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.49e-05,
+      "loss": 4.6461,
+      "step": 651
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 4.5024,
+      "step": 652
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.510000000000001e-05,
+      "loss": 4.4683,
+      "step": 653
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.52e-05,
+      "loss": 4.6136,
+      "step": 654
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.53e-05,
+      "loss": 4.3552,
+      "step": 655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.54e-05,
+      "loss": 4.3501,
+      "step": 656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.55e-05,
+      "loss": 4.3682,
+      "step": 657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.560000000000001e-05,
+      "loss": 4.5563,
+      "step": 658
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.570000000000001e-05,
+      "loss": 4.3076,
+      "step": 659
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.58e-05,
+      "loss": 4.5142,
+      "step": 660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.59e-05,
+      "loss": 4.2513,
+      "step": 661
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6e-05,
+      "loss": 4.2477,
+      "step": 662
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.610000000000001e-05,
+      "loss": 4.6804,
+      "step": 663
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.620000000000001e-05,
+      "loss": 4.2947,
+      "step": 664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.630000000000001e-05,
+      "loss": 4.2677,
+      "step": 665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.64e-05,
+      "loss": 4.4713,
+      "step": 666
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.65e-05,
+      "loss": 4.2813,
+      "step": 667
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.66e-05,
+      "loss": 4.4243,
+      "step": 668
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.670000000000001e-05,
+      "loss": 4.4902,
+      "step": 669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.680000000000001e-05,
+      "loss": 4.4925,
+      "step": 670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.690000000000001e-05,
+      "loss": 4.2405,
+      "step": 671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.7e-05,
+      "loss": 4.2419,
+      "step": 672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.71e-05,
+      "loss": 4.184,
+      "step": 673
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.720000000000001e-05,
+      "loss": 4.285,
+      "step": 674
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.730000000000001e-05,
+      "loss": 4.2557,
+      "step": 675
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.740000000000001e-05,
+      "loss": 4.4999,
+      "step": 676
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.750000000000001e-05,
+      "loss": 4.1788,
+      "step": 677
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.76e-05,
+      "loss": 4.1102,
+      "step": 678
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.77e-05,
+      "loss": 4.2047,
+      "step": 679
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.780000000000001e-05,
+      "loss": 4.3496,
+      "step": 680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.790000000000001e-05,
+      "loss": 4.6399,
+      "step": 681
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 4.1935,
+      "step": 682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.81e-05,
+      "loss": 4.4716,
+      "step": 683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.82e-05,
+      "loss": 4.589,
+      "step": 684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.83e-05,
+      "loss": 4.5778,
+      "step": 685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.840000000000001e-05,
+      "loss": 4.3114,
+      "step": 686
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.850000000000001e-05,
+      "loss": 4.362,
+      "step": 687
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.860000000000001e-05,
+      "loss": 4.4691,
+      "step": 688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.87e-05,
+      "loss": 4.5104,
+      "step": 689
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.879999999999999e-05,
+      "loss": 4.6305,
+      "step": 690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.89e-05,
+      "loss": 4.2792,
+      "step": 691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.9e-05,
+      "loss": 4.3662,
+      "step": 692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.91e-05,
+      "loss": 4.2527,
+      "step": 693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.92e-05,
+      "loss": 4.7107,
+      "step": 694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.93e-05,
+      "loss": 4.5945,
+      "step": 695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.939999999999999e-05,
+      "loss": 4.4652,
+      "step": 696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.95e-05,
+      "loss": 4.8063,
+      "step": 697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.96e-05,
+      "loss": 4.3883,
+      "step": 698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.97e-05,
+      "loss": 4.5129,
+      "step": 699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.98e-05,
+      "loss": 4.6013,
+      "step": 700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.99e-05,
+      "loss": 4.2265,
+      "step": 701
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7e-05,
+      "loss": 4.3968,
+      "step": 702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.01e-05,
+      "loss": 4.2033,
+      "step": 703
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.02e-05,
+      "loss": 4.1391,
+      "step": 704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.03e-05,
+      "loss": 4.3174,
+      "step": 705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.04e-05,
+      "loss": 4.6155,
+      "step": 706
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.05e-05,
+      "loss": 4.6319,
+      "step": 707
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.06e-05,
+      "loss": 4.5327,
+      "step": 708
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.07e-05,
+      "loss": 4.448,
+      "step": 709
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.08e-05,
+      "loss": 4.4728,
+      "step": 710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.09e-05,
+      "loss": 4.3432,
+      "step": 711
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.1e-05,
+      "loss": 4.2789,
+      "step": 712
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.11e-05,
+      "loss": 4.2979,
+      "step": 713
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.12e-05,
+      "loss": 4.4785,
+      "step": 714
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.13e-05,
+      "loss": 4.5647,
+      "step": 715
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.14e-05,
+      "loss": 4.295,
+      "step": 716
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.15e-05,
+      "loss": 4.5177,
+      "step": 717
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.16e-05,
+      "loss": 4.3025,
+      "step": 718
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.17e-05,
+      "loss": 4.1011,
+      "step": 719
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.18e-05,
+      "loss": 4.3867,
+      "step": 720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.19e-05,
+      "loss": 4.1649,
+      "step": 721
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.2e-05,
+      "loss": 4.2186,
+      "step": 722
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.21e-05,
+      "loss": 4.3461,
+      "step": 723
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.22e-05,
+      "loss": 4.4331,
+      "step": 724
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.23e-05,
+      "loss": 4.2326,
+      "step": 725
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.24e-05,
+      "loss": 4.4329,
+      "step": 726
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.25e-05,
+      "loss": 4.3711,
+      "step": 727
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.26e-05,
+      "loss": 4.3121,
+      "step": 728
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.27e-05,
+      "loss": 4.1895,
+      "step": 729
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.280000000000001e-05,
+      "loss": 4.4186,
+      "step": 730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.29e-05,
+      "loss": 4.6062,
+      "step": 731
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.3e-05,
+      "loss": 4.4235,
+      "step": 732
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.31e-05,
+      "loss": 4.2582,
+      "step": 733
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.32e-05,
+      "loss": 4.4717,
+      "step": 734
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.33e-05,
+      "loss": 4.1512,
+      "step": 735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.340000000000001e-05,
+      "loss": 4.2936,
+      "step": 736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.35e-05,
+      "loss": 4.4175,
+      "step": 737
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.36e-05,
+      "loss": 4.3174,
+      "step": 738
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.37e-05,
+      "loss": 4.3359,
+      "step": 739
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.38e-05,
+      "loss": 4.5623,
+      "step": 740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.390000000000001e-05,
+      "loss": 4.5692,
+      "step": 741
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.4e-05,
+      "loss": 4.4243,
+      "step": 742
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.41e-05,
+      "loss": 4.1862,
+      "step": 743
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.42e-05,
+      "loss": 4.2999,
+      "step": 744
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.43e-05,
+      "loss": 4.6837,
+      "step": 745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.44e-05,
+      "loss": 4.8498,
+      "step": 746
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.450000000000001e-05,
+      "loss": 4.3745,
+      "step": 747
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.46e-05,
+      "loss": 4.5771,
+      "step": 748
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.47e-05,
+      "loss": 4.8582,
+      "step": 749
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.48e-05,
+      "loss": 4.4838,
+      "step": 750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.49e-05,
+      "loss": 4.4141,
+      "step": 751
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 4.3833,
+      "step": 752
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.510000000000001e-05,
+      "loss": 4.4822,
+      "step": 753
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.52e-05,
+      "loss": 4.361,
+      "step": 754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.53e-05,
+      "loss": 4.4574,
+      "step": 755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.54e-05,
+      "loss": 4.5577,
+      "step": 756
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.55e-05,
+      "loss": 4.6015,
+      "step": 757
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.560000000000001e-05,
+      "loss": 4.4762,
+      "step": 758
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.570000000000001e-05,
+      "loss": 4.2204,
+      "step": 759
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.58e-05,
+      "loss": 4.3177,
+      "step": 760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.59e-05,
+      "loss": 4.4328,
+      "step": 761
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.6e-05,
+      "loss": 4.4035,
+      "step": 762
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.61e-05,
+      "loss": 4.4347,
+      "step": 763
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.620000000000001e-05,
+      "loss": 4.2929,
+      "step": 764
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.630000000000001e-05,
+      "loss": 4.4841,
+      "step": 765
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.64e-05,
+      "loss": 4.4149,
+      "step": 766
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.65e-05,
+      "loss": 4.354,
+      "step": 767
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.66e-05,
+      "loss": 4.3543,
+      "step": 768
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.670000000000001e-05,
+      "loss": 4.4526,
+      "step": 769
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.680000000000001e-05,
+      "loss": 4.4153,
+      "step": 770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.69e-05,
+      "loss": 4.3102,
+      "step": 771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.7e-05,
+      "loss": 4.4538,
+      "step": 772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.71e-05,
+      "loss": 4.671,
+      "step": 773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.72e-05,
+      "loss": 4.1196,
+      "step": 774
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.730000000000001e-05,
+      "loss": 4.1843,
+      "step": 775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.740000000000001e-05,
+      "loss": 4.5961,
+      "step": 776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.75e-05,
+      "loss": 4.2305,
+      "step": 777
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.76e-05,
+      "loss": 4.473,
+      "step": 778
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.77e-05,
+      "loss": 4.3451,
+      "step": 779
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.780000000000001e-05,
+      "loss": 4.2912,
+      "step": 780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.790000000000001e-05,
+      "loss": 4.2049,
+      "step": 781
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 4.5228,
+      "step": 782
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.81e-05,
+      "loss": 4.4774,
+      "step": 783
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.82e-05,
+      "loss": 4.2949,
+      "step": 784
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.83e-05,
+      "loss": 4.504,
+      "step": 785
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.840000000000001e-05,
+      "loss": 4.1903,
+      "step": 786
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.850000000000001e-05,
+      "loss": 4.5175,
+      "step": 787
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.860000000000001e-05,
+      "loss": 4.3908,
+      "step": 788
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.87e-05,
+      "loss": 4.368,
+      "step": 789
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.88e-05,
+      "loss": 4.3476,
+      "step": 790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.890000000000001e-05,
+      "loss": 4.4,
+      "step": 791
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.900000000000001e-05,
+      "loss": 4.1701,
+      "step": 792
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.910000000000001e-05,
+      "loss": 4.3001,
+      "step": 793
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.920000000000001e-05,
+      "loss": 4.6359,
+      "step": 794
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.93e-05,
+      "loss": 4.6467,
+      "step": 795
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.94e-05,
+      "loss": 4.36,
+      "step": 796
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.950000000000001e-05,
+      "loss": 4.8033,
+      "step": 797
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.960000000000001e-05,
+      "loss": 4.4973,
+      "step": 798
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.970000000000001e-05,
+      "loss": 5.2729,
+      "step": 799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.98e-05,
+      "loss": 4.6859,
+      "step": 800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.99e-05,
+      "loss": 4.6159,
+      "step": 801
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8e-05,
+      "loss": 4.3627,
+      "step": 802
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.010000000000001e-05,
+      "loss": 4.4869,
+      "step": 803
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.020000000000001e-05,
+      "loss": 4.3447,
+      "step": 804
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.030000000000001e-05,
+      "loss": 4.1593,
+      "step": 805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.04e-05,
+      "loss": 4.4084,
+      "step": 806
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.05e-05,
+      "loss": 4.4197,
+      "step": 807
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.060000000000001e-05,
+      "loss": 4.256,
+      "step": 808
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.070000000000001e-05,
+      "loss": 4.573,
+      "step": 809
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.080000000000001e-05,
+      "loss": 4.2074,
+      "step": 810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.090000000000001e-05,
+      "loss": 4.6632,
+      "step": 811
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.1e-05,
+      "loss": 4.5951,
+      "step": 812
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.11e-05,
+      "loss": 4.2436,
+      "step": 813
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.120000000000001e-05,
+      "loss": 4.4481,
+      "step": 814
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.13e-05,
+      "loss": 4.4408,
+      "step": 815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.14e-05,
+      "loss": 4.3154,
+      "step": 816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.15e-05,
+      "loss": 4.6263,
+      "step": 817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.16e-05,
+      "loss": 4.4842,
+      "step": 818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.17e-05,
+      "loss": 4.436,
+      "step": 819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.18e-05,
+      "loss": 4.4879,
+      "step": 820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.19e-05,
+      "loss": 4.5195,
+      "step": 821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.2e-05,
+      "loss": 4.3613,
+      "step": 822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.21e-05,
+      "loss": 4.3907,
+      "step": 823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.22e-05,
+      "loss": 4.3614,
+      "step": 824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.23e-05,
+      "loss": 4.5803,
+      "step": 825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.24e-05,
+      "loss": 4.4035,
+      "step": 826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.25e-05,
+      "loss": 4.3269,
+      "step": 827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.26e-05,
+      "loss": 4.182,
+      "step": 828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.27e-05,
+      "loss": 4.2814,
+      "step": 829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.28e-05,
+      "loss": 4.5063,
+      "step": 830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.29e-05,
+      "loss": 4.2673,
+      "step": 831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.3e-05,
+      "loss": 4.3087,
+      "step": 832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.31e-05,
+      "loss": 4.5561,
+      "step": 833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.32e-05,
+      "loss": 4.3948,
+      "step": 834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.33e-05,
+      "loss": 4.366,
+      "step": 835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.34e-05,
+      "loss": 4.4692,
+      "step": 836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.35e-05,
+      "loss": 4.4586,
+      "step": 837
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.36e-05,
+      "loss": 4.8869,
+      "step": 838
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.37e-05,
+      "loss": 4.4208,
+      "step": 839
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.38e-05,
+      "loss": 4.4364,
+      "step": 840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.39e-05,
+      "loss": 4.8223,
+      "step": 841
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.4e-05,
+      "loss": 4.5662,
+      "step": 842
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.41e-05,
+      "loss": 4.7252,
+      "step": 843
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.42e-05,
+      "loss": 4.4637,
+      "step": 844
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.43e-05,
+      "loss": 4.737,
+      "step": 845
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.44e-05,
+      "loss": 5.1351,
+      "step": 846
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.450000000000001e-05,
+      "loss": 4.369,
+      "step": 847
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.46e-05,
+      "loss": 5.4089,
+      "step": 848
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.47e-05,
+      "loss": 4.7256,
+      "step": 849
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.48e-05,
+      "loss": 4.4306,
+      "step": 850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.49e-05,
+      "loss": 4.511,
+      "step": 851
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.5e-05,
+      "loss": 4.4147,
+      "step": 852
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.510000000000001e-05,
+      "loss": 4.6451,
+      "step": 853
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.52e-05,
+      "loss": 4.4067,
+      "step": 854
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.53e-05,
+      "loss": 4.4271,
+      "step": 855
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.54e-05,
+      "loss": 4.3588,
+      "step": 856
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.55e-05,
+      "loss": 4.577,
+      "step": 857
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.560000000000001e-05,
+      "loss": 4.3999,
+      "step": 858
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.57e-05,
+      "loss": 4.1621,
+      "step": 859
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.58e-05,
+      "loss": 4.347,
+      "step": 860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.59e-05,
+      "loss": 4.5538,
+      "step": 861
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.6e-05,
+      "loss": 4.316,
+      "step": 862
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.61e-05,
+      "loss": 4.5773,
+      "step": 863
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.620000000000001e-05,
+      "loss": 4.8271,
+      "step": 864
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.63e-05,
+      "loss": 4.5339,
+      "step": 865
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.64e-05,
+      "loss": 4.3601,
+      "step": 866
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.65e-05,
+      "loss": 4.3586,
+      "step": 867
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.66e-05,
+      "loss": 4.3987,
+      "step": 868
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.67e-05,
+      "loss": 4.5497,
+      "step": 869
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.680000000000001e-05,
+      "loss": 4.6732,
+      "step": 870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.69e-05,
+      "loss": 4.4661,
+      "step": 871
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.7e-05,
+      "loss": 4.5729,
+      "step": 872
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.71e-05,
+      "loss": 4.3671,
+      "step": 873
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.72e-05,
+      "loss": 4.3974,
+      "step": 874
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.730000000000001e-05,
+      "loss": 4.6719,
+      "step": 875
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.740000000000001e-05,
+      "loss": 4.2851,
+      "step": 876
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.75e-05,
+      "loss": 4.3199,
+      "step": 877
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.76e-05,
+      "loss": 4.3031,
+      "step": 878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.77e-05,
+      "loss": 4.4681,
+      "step": 879
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.78e-05,
+      "loss": 4.8406,
+      "step": 880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.790000000000001e-05,
+      "loss": 4.2304,
+      "step": 881
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 4.4267,
+      "step": 882
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.81e-05,
+      "loss": 4.3356,
+      "step": 883
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.82e-05,
+      "loss": 4.4564,
+      "step": 884
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.83e-05,
+      "loss": 4.4367,
+      "step": 885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.840000000000001e-05,
+      "loss": 4.5415,
+      "step": 886
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.850000000000001e-05,
+      "loss": 4.2524,
+      "step": 887
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.86e-05,
+      "loss": 4.6554,
+      "step": 888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.87e-05,
+      "loss": 4.8786,
+      "step": 889
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.88e-05,
+      "loss": 4.488,
+      "step": 890
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.89e-05,
+      "loss": 4.6429,
+      "step": 891
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.900000000000001e-05,
+      "loss": 4.1816,
+      "step": 892
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.910000000000001e-05,
+      "loss": 4.4809,
+      "step": 893
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.92e-05,
+      "loss": 4.6162,
+      "step": 894
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.93e-05,
+      "loss": 4.2593,
+      "step": 895
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.94e-05,
+      "loss": 4.0558,
+      "step": 896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.950000000000001e-05,
+      "loss": 4.1731,
+      "step": 897
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.960000000000001e-05,
+      "loss": 4.4209,
+      "step": 898
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.970000000000001e-05,
+      "loss": 4.0567,
+      "step": 899
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.98e-05,
+      "loss": 4.3808,
+      "step": 900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.99e-05,
+      "loss": 4.1731,
+      "step": 901
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9e-05,
+      "loss": 4.3003,
+      "step": 902
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.010000000000001e-05,
+      "loss": 4.2106,
+      "step": 903
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.020000000000001e-05,
+      "loss": 4.4533,
+      "step": 904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.030000000000001e-05,
+      "loss": 4.0967,
+      "step": 905
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.04e-05,
+      "loss": 4.0577,
+      "step": 906
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.05e-05,
+      "loss": 4.2321,
+      "step": 907
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.06e-05,
+      "loss": 4.302,
+      "step": 908
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.070000000000001e-05,
+      "loss": 4.1338,
+      "step": 909
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.080000000000001e-05,
+      "loss": 4.1139,
+      "step": 910
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.090000000000001e-05,
+      "loss": 4.03,
+      "step": 911
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.1e-05,
+      "loss": 4.1155,
+      "step": 912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.11e-05,
+      "loss": 4.1578,
+      "step": 913
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.120000000000001e-05,
+      "loss": 4.17,
+      "step": 914
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.130000000000001e-05,
+      "loss": 3.9033,
+      "step": 915
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.140000000000001e-05,
+      "loss": 4.1073,
+      "step": 916
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.15e-05,
+      "loss": 4.0899,
+      "step": 917
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.16e-05,
+      "loss": 4.3304,
+      "step": 918
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.17e-05,
+      "loss": 4.278,
+      "step": 919
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.180000000000001e-05,
+      "loss": 4.1,
+      "step": 920
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.190000000000001e-05,
+      "loss": 3.8213,
+      "step": 921
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 4.2282,
+      "step": 922
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.21e-05,
+      "loss": 4.4296,
+      "step": 923
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.22e-05,
+      "loss": 4.0637,
+      "step": 924
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.230000000000001e-05,
+      "loss": 4.2574,
+      "step": 925
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.240000000000001e-05,
+      "loss": 4.1422,
+      "step": 926
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.250000000000001e-05,
+      "loss": 4.2329,
+      "step": 927
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.260000000000001e-05,
+      "loss": 4.4103,
+      "step": 928
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.27e-05,
+      "loss": 4.6301,
+      "step": 929
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.28e-05,
+      "loss": 4.3308,
+      "step": 930
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.290000000000001e-05,
+      "loss": 4.2943,
+      "step": 931
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.300000000000001e-05,
+      "loss": 4.0084,
+      "step": 932
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.310000000000001e-05,
+      "loss": 4.0416,
+      "step": 933
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.320000000000002e-05,
+      "loss": 4.3415,
+      "step": 934
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.33e-05,
+      "loss": 4.33,
+      "step": 935
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.340000000000001e-05,
+      "loss": 4.3291,
+      "step": 936
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.350000000000001e-05,
+      "loss": 4.3445,
+      "step": 937
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.360000000000001e-05,
+      "loss": 4.0164,
+      "step": 938
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.370000000000001e-05,
+      "loss": 3.8924,
+      "step": 939
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.38e-05,
+      "loss": 4.3878,
+      "step": 940
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.39e-05,
+      "loss": 4.5948,
+      "step": 941
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.4e-05,
+      "loss": 4.4983,
+      "step": 942
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.41e-05,
+      "loss": 4.2636,
+      "step": 943
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.42e-05,
+      "loss": 4.2204,
+      "step": 944
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.43e-05,
+      "loss": 4.3158,
+      "step": 945
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.44e-05,
+      "loss": 4.1153,
+      "step": 946
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.449999999999999e-05,
+      "loss": 4.5182,
+      "step": 947
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.46e-05,
+      "loss": 4.408,
+      "step": 948
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.47e-05,
+      "loss": 3.9666,
+      "step": 949
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.48e-05,
+      "loss": 4.04,
+      "step": 950
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.49e-05,
+      "loss": 4.2848,
+      "step": 951
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.5e-05,
+      "loss": 4.8365,
+      "step": 952
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.51e-05,
+      "loss": 4.2834,
+      "step": 953
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.52e-05,
+      "loss": 4.1732,
+      "step": 954
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.53e-05,
+      "loss": 4.2952,
+      "step": 955
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.54e-05,
+      "loss": 4.1687,
+      "step": 956
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.55e-05,
+      "loss": 4.144,
+      "step": 957
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.56e-05,
+      "loss": 4.0318,
+      "step": 958
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.57e-05,
+      "loss": 4.3684,
+      "step": 959
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.58e-05,
+      "loss": 4.0152,
+      "step": 960
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.59e-05,
+      "loss": 4.3435,
+      "step": 961
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.6e-05,
+      "loss": 4.1021,
+      "step": 962
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.61e-05,
+      "loss": 4.3054,
+      "step": 963
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.620000000000001e-05,
+      "loss": 4.4242,
+      "step": 964
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.63e-05,
+      "loss": 4.2225,
+      "step": 965
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.64e-05,
+      "loss": 4.2853,
+      "step": 966
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.65e-05,
+      "loss": 4.297,
+      "step": 967
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.66e-05,
+      "loss": 4.0873,
+      "step": 968
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.67e-05,
+      "loss": 4.1726,
+      "step": 969
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.680000000000001e-05,
+      "loss": 3.9287,
+      "step": 970
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.69e-05,
+      "loss": 4.3089,
+      "step": 971
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.7e-05,
+      "loss": 4.1653,
+      "step": 972
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.71e-05,
+      "loss": 4.3468,
+      "step": 973
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.72e-05,
+      "loss": 4.3807,
+      "step": 974
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.730000000000001e-05,
+      "loss": 4.0688,
+      "step": 975
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.74e-05,
+      "loss": 4.1799,
+      "step": 976
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.75e-05,
+      "loss": 4.4529,
+      "step": 977
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.76e-05,
+      "loss": 4.0878,
+      "step": 978
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.77e-05,
+      "loss": 4.0386,
+      "step": 979
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.78e-05,
+      "loss": 4.2527,
+      "step": 980
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.790000000000001e-05,
+      "loss": 4.329,
+      "step": 981
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.8e-05,
+      "loss": 4.285,
+      "step": 982
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.81e-05,
+      "loss": 4.5728,
+      "step": 983
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.82e-05,
+      "loss": 4.1068,
+      "step": 984
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.83e-05,
+      "loss": 3.9573,
+      "step": 985
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.84e-05,
+      "loss": 4.2259,
+      "step": 986
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.850000000000001e-05,
+      "loss": 4.1895,
+      "step": 987
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.86e-05,
+      "loss": 4.4722,
+      "step": 988
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.87e-05,
+      "loss": 3.9947,
+      "step": 989
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.88e-05,
+      "loss": 5.1361,
+      "step": 990
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.89e-05,
+      "loss": 4.1309,
+      "step": 991
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.900000000000001e-05,
+      "loss": 4.3135,
+      "step": 992
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.910000000000001e-05,
+      "loss": 4.1869,
+      "step": 993
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.92e-05,
+      "loss": 4.2132,
+      "step": 994
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.93e-05,
+      "loss": 4.588,
+      "step": 995
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.94e-05,
+      "loss": 4.2875,
+      "step": 996
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.95e-05,
+      "loss": 4.184,
+      "step": 997
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.960000000000001e-05,
+      "loss": 4.4858,
+      "step": 998
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.970000000000001e-05,
+      "loss": 4.205,
+      "step": 999
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.98e-05,
+      "loss": 4.0157,
+      "step": 1000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.99e-05,
+      "loss": 4.5169,
+      "step": 1001
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001,
+      "loss": 4.2609,
+      "step": 1002
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.999406175771971e-05,
+      "loss": 4.2565,
+      "step": 1003
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998812351543943e-05,
+      "loss": 4.5845,
+      "step": 1004
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998218527315915e-05,
+      "loss": 4.2005,
+      "step": 1005
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.997624703087887e-05,
+      "loss": 4.1753,
+      "step": 1006
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.997030878859858e-05,
+      "loss": 4.502,
+      "step": 1007
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.99643705463183e-05,
+      "loss": 4.3062,
+      "step": 1008
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.9958432304038e-05,
+      "loss": 4.1443,
+      "step": 1009
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.995249406175773e-05,
+      "loss": 4.1294,
+      "step": 1010
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.994655581947743e-05,
+      "loss": 4.1716,
+      "step": 1011
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.994061757719715e-05,
+      "loss": 4.0862,
+      "step": 1012
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.993467933491687e-05,
+      "loss": 4.0391,
+      "step": 1013
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.992874109263658e-05,
+      "loss": 4.0757,
+      "step": 1014
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.99228028503563e-05,
+      "loss": 3.9888,
+      "step": 1015
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.991686460807601e-05,
+      "loss": 4.2584,
+      "step": 1016
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.991092636579573e-05,
+      "loss": 3.94,
+      "step": 1017
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.990498812351544e-05,
+      "loss": 4.1766,
+      "step": 1018
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.989904988123516e-05,
+      "loss": 4.2547,
+      "step": 1019
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.989311163895488e-05,
+      "loss": 5.12,
+      "step": 1020
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.98871733966746e-05,
+      "loss": 4.262,
+      "step": 1021
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.98812351543943e-05,
+      "loss": 4.154,
+      "step": 1022
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.987529691211401e-05,
+      "loss": 4.2827,
+      "step": 1023
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.986935866983373e-05,
+      "loss": 4.2962,
+      "step": 1024
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.986342042755345e-05,
+      "loss": 4.1761,
+      "step": 1025
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.985748218527317e-05,
+      "loss": 4.2523,
+      "step": 1026
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.985154394299288e-05,
+      "loss": 4.4331,
+      "step": 1027
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.98456057007126e-05,
+      "loss": 4.4276,
+      "step": 1028
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.98396674584323e-05,
+      "loss": 4.1404,
+      "step": 1029
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.983372921615203e-05,
+      "loss": 4.4389,
+      "step": 1030
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.982779097387173e-05,
+      "loss": 4.254,
+      "step": 1031
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.982185273159145e-05,
+      "loss": 4.2004,
+      "step": 1032
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.981591448931117e-05,
+      "loss": 4.1713,
+      "step": 1033
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.980997624703088e-05,
+      "loss": 3.9861,
+      "step": 1034
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.98040380047506e-05,
+      "loss": 5.4904,
+      "step": 1035
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.979809976247031e-05,
+      "loss": 4.1472,
+      "step": 1036
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.979216152019003e-05,
+      "loss": 4.4612,
+      "step": 1037
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.978622327790973e-05,
+      "loss": 5.3692,
+      "step": 1038
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.978028503562947e-05,
+      "loss": 4.4212,
+      "step": 1039
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.977434679334918e-05,
+      "loss": 4.5247,
+      "step": 1040
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.97684085510689e-05,
+      "loss": 5.0384,
+      "step": 1041
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.97624703087886e-05,
+      "loss": 4.597,
+      "step": 1042
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.975653206650832e-05,
+      "loss": 4.3015,
+      "step": 1043
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.975059382422803e-05,
+      "loss": 4.3721,
+      "step": 1044
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.974465558194775e-05,
+      "loss": 4.2023,
+      "step": 1045
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.973871733966747e-05,
+      "loss": 4.2111,
+      "step": 1046
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.973277909738718e-05,
+      "loss": 4.0737,
+      "step": 1047
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.97268408551069e-05,
+      "loss": 4.0907,
+      "step": 1048
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.97209026128266e-05,
+      "loss": 4.8861,
+      "step": 1049
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.971496437054632e-05,
+      "loss": 4.5571,
+      "step": 1050
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.970902612826603e-05,
+      "loss": 4.4433,
+      "step": 1051
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.970308788598575e-05,
+      "loss": 4.3186,
+      "step": 1052
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.969714964370547e-05,
+      "loss": 4.2538,
+      "step": 1053
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.969121140142518e-05,
+      "loss": 4.8854,
+      "step": 1054
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.96852731591449e-05,
+      "loss": 4.4031,
+      "step": 1055
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.96793349168646e-05,
+      "loss": 4.3464,
+      "step": 1056
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.967339667458433e-05,
+      "loss": 4.323,
+      "step": 1057
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.966745843230403e-05,
+      "loss": 4.1481,
+      "step": 1058
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.966152019002377e-05,
+      "loss": 4.9973,
+      "step": 1059
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.965558194774347e-05,
+      "loss": 4.1922,
+      "step": 1060
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.96496437054632e-05,
+      "loss": 4.3675,
+      "step": 1061
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.96437054631829e-05,
+      "loss": 4.4189,
+      "step": 1062
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.963776722090262e-05,
+      "loss": 4.3698,
+      "step": 1063
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.963182897862233e-05,
+      "loss": 4.3233,
+      "step": 1064
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.962589073634205e-05,
+      "loss": 4.3355,
+      "step": 1065
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.961995249406177e-05,
+      "loss": 4.6455,
+      "step": 1066
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.961401425178148e-05,
+      "loss": 4.1259,
+      "step": 1067
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.96080760095012e-05,
+      "loss": 4.6023,
+      "step": 1068
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.96021377672209e-05,
+      "loss": 4.2826,
+      "step": 1069
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.959619952494062e-05,
+      "loss": 4.2497,
+      "step": 1070
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.959026128266033e-05,
+      "loss": 4.1776,
+      "step": 1071
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.958432304038005e-05,
+      "loss": 4.2205,
+      "step": 1072
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.957838479809977e-05,
+      "loss": 4.1873,
+      "step": 1073
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.957244655581949e-05,
+      "loss": 4.1678,
+      "step": 1074
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.95665083135392e-05,
+      "loss": 3.9185,
+      "step": 1075
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.95605700712589e-05,
+      "loss": 4.3019,
+      "step": 1076
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.955463182897862e-05,
+      "loss": 4.1339,
+      "step": 1077
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.954869358669833e-05,
+      "loss": 4.3994,
+      "step": 1078
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.954275534441807e-05,
+      "loss": 4.525,
+      "step": 1079
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.953681710213777e-05,
+      "loss": 4.2741,
+      "step": 1080
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.953087885985749e-05,
+      "loss": 4.5573,
+      "step": 1081
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.95249406175772e-05,
+      "loss": 4.2156,
+      "step": 1082
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.951900237529692e-05,
+      "loss": 4.4789,
+      "step": 1083
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.951306413301663e-05,
+      "loss": 4.0092,
+      "step": 1084
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.950712589073635e-05,
+      "loss": 4.1451,
+      "step": 1085
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.950118764845607e-05,
+      "loss": 4.3632,
+      "step": 1086
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.949524940617577e-05,
+      "loss": 4.4124,
+      "step": 1087
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.94893111638955e-05,
+      "loss": 4.2495,
+      "step": 1088
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.94833729216152e-05,
+      "loss": 4.4235,
+      "step": 1089
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.947743467933492e-05,
+      "loss": 4.1811,
+      "step": 1090
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.947149643705463e-05,
+      "loss": 4.8561,
+      "step": 1091
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.946555819477435e-05,
+      "loss": 4.2078,
+      "step": 1092
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.945961995249407e-05,
+      "loss": 4.7112,
+      "step": 1093
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.945368171021379e-05,
+      "loss": 4.0684,
+      "step": 1094
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.94477434679335e-05,
+      "loss": 4.0835,
+      "step": 1095
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.94418052256532e-05,
+      "loss": 4.4166,
+      "step": 1096
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.943586698337292e-05,
+      "loss": 4.2387,
+      "step": 1097
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.942992874109263e-05,
+      "loss": 4.4457,
+      "step": 1098
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.942399049881236e-05,
+      "loss": 4.838,
+      "step": 1099
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.941805225653207e-05,
+      "loss": 4.6989,
+      "step": 1100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.941211401425179e-05,
+      "loss": 4.5283,
+      "step": 1101
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.94061757719715e-05,
+      "loss": 4.079,
+      "step": 1102
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.940023752969122e-05,
+      "loss": 4.379,
+      "step": 1103
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.939429928741093e-05,
+      "loss": 4.2469,
+      "step": 1104
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.938836104513065e-05,
+      "loss": 4.3928,
+      "step": 1105
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.938242280285037e-05,
+      "loss": 4.1058,
+      "step": 1106
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.937648456057007e-05,
+      "loss": 4.7313,
+      "step": 1107
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.93705463182898e-05,
+      "loss": 4.3948,
+      "step": 1108
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.93646080760095e-05,
+      "loss": 4.3309,
+      "step": 1109
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.935866983372922e-05,
+      "loss": 4.4175,
+      "step": 1110
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.935273159144893e-05,
+      "loss": 4.257,
+      "step": 1111
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.934679334916865e-05,
+      "loss": 4.2912,
+      "step": 1112
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.934085510688837e-05,
+      "loss": 4.4846,
+      "step": 1113
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.933491686460809e-05,
+      "loss": 4.4349,
+      "step": 1114
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.93289786223278e-05,
+      "loss": 4.3936,
+      "step": 1115
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.932304038004752e-05,
+      "loss": 4.1756,
+      "step": 1116
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.931710213776722e-05,
+      "loss": 4.2348,
+      "step": 1117
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.931116389548694e-05,
+      "loss": 4.9971,
+      "step": 1118
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.930522565320666e-05,
+      "loss": 4.2987,
+      "step": 1119
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.929928741092637e-05,
+      "loss": 4.0176,
+      "step": 1120
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.929334916864609e-05,
+      "loss": 4.4535,
+      "step": 1121
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.92874109263658e-05,
+      "loss": 4.1817,
+      "step": 1122
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.928147268408552e-05,
+      "loss": 4.3489,
+      "step": 1123
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.927553444180522e-05,
+      "loss": 4.1111,
+      "step": 1124
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.926959619952494e-05,
+      "loss": 4.4833,
+      "step": 1125
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.926365795724466e-05,
+      "loss": 4.0834,
+      "step": 1126
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.925771971496437e-05,
+      "loss": 4.1785,
+      "step": 1127
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.925178147268409e-05,
+      "loss": 4.4717,
+      "step": 1128
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.92458432304038e-05,
+      "loss": 4.3857,
+      "step": 1129
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.923990498812352e-05,
+      "loss": 4.6164,
+      "step": 1130
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.923396674584323e-05,
+      "loss": 4.246,
+      "step": 1131
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.922802850356295e-05,
+      "loss": 4.1461,
+      "step": 1132
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.922209026128267e-05,
+      "loss": 4.3011,
+      "step": 1133
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.921615201900239e-05,
+      "loss": 4.3833,
+      "step": 1134
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.92102137767221e-05,
+      "loss": 3.8892,
+      "step": 1135
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.920427553444181e-05,
+      "loss": 4.2201,
+      "step": 1136
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.919833729216152e-05,
+      "loss": 4.6946,
+      "step": 1137
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.919239904988124e-05,
+      "loss": 4.8812,
+      "step": 1138
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.918646080760096e-05,
+      "loss": 4.0698,
+      "step": 1139
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.918052256532067e-05,
+      "loss": 5.0052,
+      "step": 1140
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.917458432304039e-05,
+      "loss": 4.654,
+      "step": 1141
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.91686460807601e-05,
+      "loss": 4.3306,
+      "step": 1142
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.916270783847982e-05,
+      "loss": 4.5636,
+      "step": 1143
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.915676959619952e-05,
+      "loss": 4.2009,
+      "step": 1144
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.915083135391924e-05,
+      "loss": 4.607,
+      "step": 1145
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.914489311163896e-05,
+      "loss": 4.5729,
+      "step": 1146
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.913895486935868e-05,
+      "loss": 4.2084,
+      "step": 1147
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.913301662707839e-05,
+      "loss": 4.2989,
+      "step": 1148
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.91270783847981e-05,
+      "loss": 4.4703,
+      "step": 1149
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.912114014251782e-05,
+      "loss": 4.2408,
+      "step": 1150
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.911520190023752e-05,
+      "loss": 4.6987,
+      "step": 1151
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.910926365795726e-05,
+      "loss": 4.1526,
+      "step": 1152
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.910332541567697e-05,
+      "loss": 4.3992,
+      "step": 1153
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.909738717339669e-05,
+      "loss": 4.3902,
+      "step": 1154
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.909144893111639e-05,
+      "loss": 4.4643,
+      "step": 1155
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.908551068883611e-05,
+      "loss": 4.0571,
+      "step": 1156
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.907957244655582e-05,
+      "loss": 4.31,
+      "step": 1157
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.907363420427554e-05,
+      "loss": 4.2885,
+      "step": 1158
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.906769596199526e-05,
+      "loss": 4.1804,
+      "step": 1159
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.906175771971497e-05,
+      "loss": 3.9087,
+      "step": 1160
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.905581947743469e-05,
+      "loss": 4.0516,
+      "step": 1161
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.90498812351544e-05,
+      "loss": 4.108,
+      "step": 1162
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.904394299287411e-05,
+      "loss": 4.1929,
+      "step": 1163
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.903800475059382e-05,
+      "loss": 4.3225,
+      "step": 1164
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.903206650831354e-05,
+      "loss": 4.8453,
+      "step": 1165
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.902612826603326e-05,
+      "loss": 4.2769,
+      "step": 1166
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.902019002375298e-05,
+      "loss": 4.2378,
+      "step": 1167
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.901425178147269e-05,
+      "loss": 4.0901,
+      "step": 1168
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.90083135391924e-05,
+      "loss": 4.2744,
+      "step": 1169
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.900237529691212e-05,
+      "loss": 4.4762,
+      "step": 1170
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.899643705463182e-05,
+      "loss": 4.6052,
+      "step": 1171
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.899049881235156e-05,
+      "loss": 4.693,
+      "step": 1172
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.898456057007126e-05,
+      "loss": 4.438,
+      "step": 1173
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.897862232779098e-05,
+      "loss": 4.339,
+      "step": 1174
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.897268408551069e-05,
+      "loss": 4.4128,
+      "step": 1175
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.896674584323041e-05,
+      "loss": 4.3437,
+      "step": 1176
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.896080760095012e-05,
+      "loss": 4.5125,
+      "step": 1177
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.895486935866984e-05,
+      "loss": 4.1326,
+      "step": 1178
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.894893111638956e-05,
+      "loss": 4.3556,
+      "step": 1179
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.894299287410927e-05,
+      "loss": 4.1024,
+      "step": 1180
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.893705463182899e-05,
+      "loss": 4.1706,
+      "step": 1181
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.89311163895487e-05,
+      "loss": 4.3934,
+      "step": 1182
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.892517814726841e-05,
+      "loss": 4.2918,
+      "step": 1183
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.891923990498812e-05,
+      "loss": 4.1251,
+      "step": 1184
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.891330166270784e-05,
+      "loss": 3.9312,
+      "step": 1185
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.890736342042756e-05,
+      "loss": 4.1394,
+      "step": 1186
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.890142517814728e-05,
+      "loss": 3.9789,
+      "step": 1187
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.889548693586699e-05,
+      "loss": 4.3013,
+      "step": 1188
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.88895486935867e-05,
+      "loss": 4.0698,
+      "step": 1189
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.888361045130642e-05,
+      "loss": 4.2572,
+      "step": 1190
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.887767220902612e-05,
+      "loss": 4.6749,
+      "step": 1191
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.887173396674586e-05,
+      "loss": 6.3022,
+      "step": 1192
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.886579572446556e-05,
+      "loss": 4.3254,
+      "step": 1193
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.885985748218528e-05,
+      "loss": 4.5276,
+      "step": 1194
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.885391923990499e-05,
+      "loss": 4.3036,
+      "step": 1195
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.884798099762471e-05,
+      "loss": 4.4677,
+      "step": 1196
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.884204275534442e-05,
+      "loss": 4.2354,
+      "step": 1197
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.883610451306414e-05,
+      "loss": 4.323,
+      "step": 1198
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.883016627078386e-05,
+      "loss": 4.4029,
+      "step": 1199
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.882422802850356e-05,
+      "loss": 4.1119,
+      "step": 1200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.881828978622329e-05,
+      "loss": 4.2821,
+      "step": 1201
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.881235154394299e-05,
+      "loss": 4.4601,
+      "step": 1202
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.880641330166271e-05,
+      "loss": 4.1707,
+      "step": 1203
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.880047505938242e-05,
+      "loss": 4.2332,
+      "step": 1204
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.879453681710214e-05,
+      "loss": 4.1487,
+      "step": 1205
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.878859857482186e-05,
+      "loss": 4.4109,
+      "step": 1206
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.878266033254158e-05,
+      "loss": 4.5674,
+      "step": 1207
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.878266033254158e-05,
+      "loss": 4.495,
+      "step": 1208
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.877672209026129e-05,
+      "loss": 4.4116,
+      "step": 1209
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.877078384798101e-05,
+      "loss": 4.4481,
+      "step": 1210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.876484560570071e-05,
+      "loss": 4.1894,
+      "step": 1211
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.875890736342042e-05,
+      "loss": 4.1882,
+      "step": 1212
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.875296912114015e-05,
+      "loss": 4.9425,
+      "step": 1213
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.874703087885986e-05,
+      "loss": 4.2668,
+      "step": 1214
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.874109263657958e-05,
+      "loss": 4.3125,
+      "step": 1215
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.873515439429929e-05,
+      "loss": 4.2517,
+      "step": 1216
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.872921615201901e-05,
+      "loss": 4.1149,
+      "step": 1217
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.872327790973872e-05,
+      "loss": 4.2978,
+      "step": 1218
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.871733966745844e-05,
+      "loss": 4.2464,
+      "step": 1219
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.871140142517816e-05,
+      "loss": 4.3928,
+      "step": 1220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.870546318289786e-05,
+      "loss": 4.2514,
+      "step": 1221
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.869952494061758e-05,
+      "loss": 4.4974,
+      "step": 1222
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.869358669833729e-05,
+      "loss": 4.1234,
+      "step": 1223
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.868764845605701e-05,
+      "loss": 4.3919,
+      "step": 1224
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.868171021377672e-05,
+      "loss": 5.0076,
+      "step": 1225
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.867577197149644e-05,
+      "loss": 4.497,
+      "step": 1226
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.866983372921616e-05,
+      "loss": 4.237,
+      "step": 1227
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.866389548693588e-05,
+      "loss": 4.4512,
+      "step": 1228
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.865795724465559e-05,
+      "loss": 4.3414,
+      "step": 1229
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.86520190023753e-05,
+      "loss": 3.9671,
+      "step": 1230
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.864608076009501e-05,
+      "loss": 4.4519,
+      "step": 1231
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.864014251781473e-05,
+      "loss": 4.2704,
+      "step": 1232
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.863420427553445e-05,
+      "loss": 4.4546,
+      "step": 1233
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.862826603325416e-05,
+      "loss": 4.5089,
+      "step": 1234
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.862232779097388e-05,
+      "loss": 4.3015,
+      "step": 1235
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.861638954869359e-05,
+      "loss": 4.3311,
+      "step": 1236
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.861045130641331e-05,
+      "loss": 4.3353,
+      "step": 1237
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.860451306413301e-05,
+      "loss": 4.3515,
+      "step": 1238
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.859857482185274e-05,
+      "loss": 4.2812,
+      "step": 1239
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.859263657957246e-05,
+      "loss": 3.8219,
+      "step": 1240
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.858669833729218e-05,
+      "loss": 4.6426,
+      "step": 1241
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.858076009501188e-05,
+      "loss": 4.602,
+      "step": 1242
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.857482185273159e-05,
+      "loss": 4.612,
+      "step": 1243
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.856888361045131e-05,
+      "loss": 4.4442,
+      "step": 1244
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.856294536817102e-05,
+      "loss": 4.8317,
+      "step": 1245
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.855700712589074e-05,
+      "loss": 4.5549,
+      "step": 1246
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.855106888361046e-05,
+      "loss": 4.6514,
+      "step": 1247
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.854513064133018e-05,
+      "loss": 4.4699,
+      "step": 1248
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.853919239904988e-05,
+      "loss": 4.4293,
+      "step": 1249
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.85332541567696e-05,
+      "loss": 4.2406,
+      "step": 1250
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.852731591448931e-05,
+      "loss": 4.5166,
+      "step": 1251
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.852137767220903e-05,
+      "loss": 4.2058,
+      "step": 1252
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.851543942992875e-05,
+      "loss": 4.387,
+      "step": 1253
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.850950118764846e-05,
+      "loss": 4.3538,
+      "step": 1254
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.850356294536818e-05,
+      "loss": 4.8358,
+      "step": 1255
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.849762470308789e-05,
+      "loss": 3.9956,
+      "step": 1256
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.84916864608076e-05,
+      "loss": 4.3482,
+      "step": 1257
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.848574821852731e-05,
+      "loss": 4.1965,
+      "step": 1258
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.847980997624703e-05,
+      "loss": 4.2172,
+      "step": 1259
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.847387173396675e-05,
+      "loss": 4.3746,
+      "step": 1260
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.846793349168647e-05,
+      "loss": 4.4868,
+      "step": 1261
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.846199524940618e-05,
+      "loss": 4.2933,
+      "step": 1262
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.845605700712589e-05,
+      "loss": 4.0732,
+      "step": 1263
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.845011876484561e-05,
+      "loss": 3.996,
+      "step": 1264
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.844418052256532e-05,
+      "loss": 4.212,
+      "step": 1265
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.843824228028504e-05,
+      "loss": 4.2823,
+      "step": 1266
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.843230403800476e-05,
+      "loss": 4.07,
+      "step": 1267
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.842636579572448e-05,
+      "loss": 4.397,
+      "step": 1268
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.842042755344418e-05,
+      "loss": 4.3032,
+      "step": 1269
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.84144893111639e-05,
+      "loss": 4.2726,
+      "step": 1270
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.840855106888361e-05,
+      "loss": 4.1112,
+      "step": 1271
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.840261282660333e-05,
+      "loss": 4.2683,
+      "step": 1272
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.839667458432305e-05,
+      "loss": 4.3266,
+      "step": 1273
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.839073634204276e-05,
+      "loss": 4.1674,
+      "step": 1274
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.838479809976248e-05,
+      "loss": 4.4032,
+      "step": 1275
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.837885985748219e-05,
+      "loss": 4.1303,
+      "step": 1276
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.83729216152019e-05,
+      "loss": 4.8279,
+      "step": 1277
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.836698337292161e-05,
+      "loss": 4.2929,
+      "step": 1278
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.836104513064133e-05,
+      "loss": 4.3896,
+      "step": 1279
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.835510688836105e-05,
+      "loss": 3.9265,
+      "step": 1280
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.834916864608077e-05,
+      "loss": 4.2379,
+      "step": 1281
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.834323040380048e-05,
+      "loss": 4.1862,
+      "step": 1282
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.833729216152019e-05,
+      "loss": 4.3963,
+      "step": 1283
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.833135391923991e-05,
+      "loss": 4.0334,
+      "step": 1284
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.832541567695961e-05,
+      "loss": 4.3461,
+      "step": 1285
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.831947743467935e-05,
+      "loss": 4.6112,
+      "step": 1286
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.831353919239905e-05,
+      "loss": 4.2056,
+      "step": 1287
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.830760095011878e-05,
+      "loss": 4.5514,
+      "step": 1288
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.830166270783848e-05,
+      "loss": 4.472,
+      "step": 1289
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.82957244655582e-05,
+      "loss": 4.3207,
+      "step": 1290
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.828978622327791e-05,
+      "loss": 4.3308,
+      "step": 1291
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.828384798099763e-05,
+      "loss": 4.8702,
+      "step": 1292
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.827790973871735e-05,
+      "loss": 4.2398,
+      "step": 1293
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.827197149643706e-05,
+      "loss": 4.4093,
+      "step": 1294
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.826603325415678e-05,
+      "loss": 4.3676,
+      "step": 1295
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.826009501187648e-05,
+      "loss": 4.3275,
+      "step": 1296
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.82541567695962e-05,
+      "loss": 4.0797,
+      "step": 1297
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.824821852731591e-05,
+      "loss": 4.1516,
+      "step": 1298
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.824228028503563e-05,
+      "loss": 4.3498,
+      "step": 1299
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.823634204275535e-05,
+      "loss": 4.2887,
+      "step": 1300
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.823040380047507e-05,
+      "loss": 4.348,
+      "step": 1301
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.822446555819478e-05,
+      "loss": 4.4661,
+      "step": 1302
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.82185273159145e-05,
+      "loss": 4.2657,
+      "step": 1303
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.82125890736342e-05,
+      "loss": 4.3613,
+      "step": 1304
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.820665083135391e-05,
+      "loss": 4.1599,
+      "step": 1305
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.820071258907365e-05,
+      "loss": 4.1816,
+      "step": 1306
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.819477434679335e-05,
+      "loss": 4.1606,
+      "step": 1307
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.818883610451307e-05,
+      "loss": 4.2371,
+      "step": 1308
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.818289786223278e-05,
+      "loss": 4.2306,
+      "step": 1309
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.81769596199525e-05,
+      "loss": 4.2853,
+      "step": 1310
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.817102137767221e-05,
+      "loss": 4.3859,
+      "step": 1311
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.816508313539193e-05,
+      "loss": 4.7774,
+      "step": 1312
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.815914489311165e-05,
+      "loss": 4.0024,
+      "step": 1313
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.815320665083137e-05,
+      "loss": 4.1803,
+      "step": 1314
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.814726840855108e-05,
+      "loss": 4.3828,
+      "step": 1315
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.814133016627078e-05,
+      "loss": 4.1582,
+      "step": 1316
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.81353919239905e-05,
+      "loss": 4.6699,
+      "step": 1317
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.812945368171021e-05,
+      "loss": 4.3753,
+      "step": 1318
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.812351543942993e-05,
+      "loss": 4.235,
+      "step": 1319
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.811757719714965e-05,
+      "loss": 4.2084,
+      "step": 1320
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.811163895486937e-05,
+      "loss": 4.1285,
+      "step": 1321
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.810570071258908e-05,
+      "loss": 4.8052,
+      "step": 1322
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.80997624703088e-05,
+      "loss": 4.4917,
+      "step": 1323
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.80938242280285e-05,
+      "loss": 4.3461,
+      "step": 1324
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.808788598574821e-05,
+      "loss": 4.3979,
+      "step": 1325
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.808194774346795e-05,
+      "loss": 4.3739,
+      "step": 1326
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.807600950118765e-05,
+      "loss": 4.363,
+      "step": 1327
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.807007125890737e-05,
+      "loss": 4.4282,
+      "step": 1328
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.806413301662708e-05,
+      "loss": 4.4785,
+      "step": 1329
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.80581947743468e-05,
+      "loss": 4.1677,
+      "step": 1330
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.80522565320665e-05,
+      "loss": 4.2401,
+      "step": 1331
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.804631828978623e-05,
+      "loss": 4.3421,
+      "step": 1332
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.804038004750595e-05,
+      "loss": 4.2048,
+      "step": 1333
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.803444180522567e-05,
+      "loss": 4.3433,
+      "step": 1334
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.802850356294537e-05,
+      "loss": 4.7555,
+      "step": 1335
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.802256532066508e-05,
+      "loss": 4.2027,
+      "step": 1336
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.80166270783848e-05,
+      "loss": 4.3186,
+      "step": 1337
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.801068883610451e-05,
+      "loss": 4.6348,
+      "step": 1338
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.800475059382423e-05,
+      "loss": 4.2551,
+      "step": 1339
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.799881235154395e-05,
+      "loss": 4.4183,
+      "step": 1340
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.799287410926367e-05,
+      "loss": 4.2958,
+      "step": 1341
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.798693586698338e-05,
+      "loss": 3.9798,
+      "step": 1342
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.79809976247031e-05,
+      "loss": 4.8579,
+      "step": 1343
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.79750593824228e-05,
+      "loss": 4.1298,
+      "step": 1344
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.796912114014252e-05,
+      "loss": 4.3091,
+      "step": 1345
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.796318289786224e-05,
+      "loss": 4.4876,
+      "step": 1346
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.795724465558195e-05,
+      "loss": 4.2172,
+      "step": 1347
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.795130641330167e-05,
+      "loss": 4.4832,
+      "step": 1348
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.794536817102138e-05,
+      "loss": 4.2124,
+      "step": 1349
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.79394299287411e-05,
+      "loss": 4.1326,
+      "step": 1350
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.79334916864608e-05,
+      "loss": 4.4193,
+      "step": 1351
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.792755344418053e-05,
+      "loss": 4.4244,
+      "step": 1352
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.792161520190025e-05,
+      "loss": 4.3251,
+      "step": 1353
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.791567695961997e-05,
+      "loss": 4.3669,
+      "step": 1354
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.790973871733967e-05,
+      "loss": 4.3021,
+      "step": 1355
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.790380047505938e-05,
+      "loss": 4.5098,
+      "step": 1356
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.78978622327791e-05,
+      "loss": 4.2609,
+      "step": 1357
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.789192399049881e-05,
+      "loss": 4.4382,
+      "step": 1358
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.788598574821853e-05,
+      "loss": 4.1726,
+      "step": 1359
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.788004750593825e-05,
+      "loss": 4.0598,
+      "step": 1360
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.787410926365797e-05,
+      "loss": 4.1932,
+      "step": 1361
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.786817102137767e-05,
+      "loss": 4.3436,
+      "step": 1362
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.78622327790974e-05,
+      "loss": 4.3365,
+      "step": 1363
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.78562945368171e-05,
+      "loss": 4.1953,
+      "step": 1364
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.785035629453682e-05,
+      "loss": 4.5171,
+      "step": 1365
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.784441805225654e-05,
+      "loss": 4.2169,
+      "step": 1366
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.783847980997625e-05,
+      "loss": 4.2286,
+      "step": 1367
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.783254156769597e-05,
+      "loss": 4.3668,
+      "step": 1368
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.782660332541568e-05,
+      "loss": 4.0577,
+      "step": 1369
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.78206650831354e-05,
+      "loss": 4.1589,
+      "step": 1370
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.78147268408551e-05,
+      "loss": 4.2192,
+      "step": 1371
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.780878859857482e-05,
+      "loss": 4.3003,
+      "step": 1372
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.780285035629454e-05,
+      "loss": 4.3444,
+      "step": 1373
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.779691211401426e-05,
+      "loss": 4.1404,
+      "step": 1374
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.779097387173397e-05,
+      "loss": 4.3094,
+      "step": 1375
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.778503562945369e-05,
+      "loss": 4.2367,
+      "step": 1376
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.77790973871734e-05,
+      "loss": 4.087,
+      "step": 1377
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.77731591448931e-05,
+      "loss": 4.0528,
+      "step": 1378
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.776722090261283e-05,
+      "loss": 4.434,
+      "step": 1379
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.776128266033255e-05,
+      "loss": 4.15,
+      "step": 1380
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.775534441805227e-05,
+      "loss": 4.0942,
+      "step": 1381
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.774940617577197e-05,
+      "loss": 4.3856,
+      "step": 1382
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.77434679334917e-05,
+      "loss": 4.4474,
+      "step": 1383
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.77375296912114e-05,
+      "loss": 4.0875,
+      "step": 1384
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.773159144893112e-05,
+      "loss": 4.841,
+      "step": 1385
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.772565320665084e-05,
+      "loss": 4.2094,
+      "step": 1386
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.771971496437055e-05,
+      "loss": 4.3964,
+      "step": 1387
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.771377672209027e-05,
+      "loss": 4.7692,
+      "step": 1388
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.770783847980998e-05,
+      "loss": 3.9678,
+      "step": 1389
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.77019002375297e-05,
+      "loss": 4.0961,
+      "step": 1390
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.76959619952494e-05,
+      "loss": 4.4407,
+      "step": 1391
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.769002375296912e-05,
+      "loss": 4.5129,
+      "step": 1392
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.768408551068884e-05,
+      "loss": 4.2493,
+      "step": 1393
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.767814726840856e-05,
+      "loss": 4.2046,
+      "step": 1394
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.767220902612827e-05,
+      "loss": 4.0701,
+      "step": 1395
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.766627078384799e-05,
+      "loss": 4.1927,
+      "step": 1396
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.76603325415677e-05,
+      "loss": 4.1267,
+      "step": 1397
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.76543942992874e-05,
+      "loss": 4.0898,
+      "step": 1398
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.764845605700714e-05,
+      "loss": 4.266,
+      "step": 1399
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.764251781472685e-05,
+      "loss": 4.3667,
+      "step": 1400
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.763657957244657e-05,
+      "loss": 4.1423,
+      "step": 1401
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.763064133016627e-05,
+      "loss": 4.929,
+      "step": 1402
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.762470308788599e-05,
+      "loss": 3.9936,
+      "step": 1403
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.76187648456057e-05,
+      "loss": 4.7452,
+      "step": 1404
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.761282660332542e-05,
+      "loss": 4.0139,
+      "step": 1405
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.760688836104514e-05,
+      "loss": 4.6419,
+      "step": 1406
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.760095011876486e-05,
+      "loss": 4.2254,
+      "step": 1407
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.759501187648457e-05,
+      "loss": 4.1012,
+      "step": 1408
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.758907363420427e-05,
+      "loss": 4.1201,
+      "step": 1409
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.7583135391924e-05,
+      "loss": 4.1089,
+      "step": 1410
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.75771971496437e-05,
+      "loss": 4.3952,
+      "step": 1411
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.757125890736342e-05,
+      "loss": 4.4949,
+      "step": 1412
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.756532066508314e-05,
+      "loss": 4.3253,
+      "step": 1413
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.755938242280286e-05,
+      "loss": 4.0718,
+      "step": 1414
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.755344418052257e-05,
+      "loss": 4.2946,
+      "step": 1415
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.754750593824229e-05,
+      "loss": 4.0121,
+      "step": 1416
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.7541567695962e-05,
+      "loss": 4.0407,
+      "step": 1417
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.75356294536817e-05,
+      "loss": 3.9672,
+      "step": 1418
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.752969121140144e-05,
+      "loss": 4.4225,
+      "step": 1419
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.752375296912114e-05,
+      "loss": 4.4329,
+      "step": 1420
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.751781472684086e-05,
+      "loss": 4.4406,
+      "step": 1421
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.751187648456057e-05,
+      "loss": 4.3162,
+      "step": 1422
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.750593824228029e-05,
+      "loss": 4.2936,
+      "step": 1423
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.75e-05,
+      "loss": 4.158,
+      "step": 1424
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.749406175771972e-05,
+      "loss": 4.2452,
+      "step": 1425
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.748812351543944e-05,
+      "loss": 4.3367,
+      "step": 1426
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.748218527315916e-05,
+      "loss": 4.1184,
+      "step": 1427
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.747624703087887e-05,
+      "loss": 4.1725,
+      "step": 1428
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.747030878859857e-05,
+      "loss": 4.3674,
+      "step": 1429
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.74643705463183e-05,
+      "loss": 4.4894,
+      "step": 1430
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.7458432304038e-05,
+      "loss": 4.4073,
+      "step": 1431
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.745249406175772e-05,
+      "loss": 4.4417,
+      "step": 1432
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.744655581947744e-05,
+      "loss": 4.4357,
+      "step": 1433
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.744061757719716e-05,
+      "loss": 4.3079,
+      "step": 1434
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.743467933491687e-05,
+      "loss": 4.2782,
+      "step": 1435
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.742874109263659e-05,
+      "loss": 4.9306,
+      "step": 1436
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.74228028503563e-05,
+      "loss": 4.4391,
+      "step": 1437
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.741686460807602e-05,
+      "loss": 4.1006,
+      "step": 1438
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.741092636579574e-05,
+      "loss": 4.3989,
+      "step": 1439
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.740498812351544e-05,
+      "loss": 5.4388,
+      "step": 1440
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.739904988123516e-05,
+      "loss": 4.2219,
+      "step": 1441
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.739311163895487e-05,
+      "loss": 4.228,
+      "step": 1442
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.738717339667459e-05,
+      "loss": 4.0727,
+      "step": 1443
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.73812351543943e-05,
+      "loss": 4.5409,
+      "step": 1444
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.737529691211402e-05,
+      "loss": 4.2025,
+      "step": 1445
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.736935866983374e-05,
+      "loss": 4.0784,
+      "step": 1446
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.736342042755346e-05,
+      "loss": 4.0798,
+      "step": 1447
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.735748218527316e-05,
+      "loss": 4.3746,
+      "step": 1448
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.735154394299287e-05,
+      "loss": 4.0739,
+      "step": 1449
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.734560570071259e-05,
+      "loss": 4.2473,
+      "step": 1450
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.73396674584323e-05,
+      "loss": 4.5388,
+      "step": 1451
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.733372921615202e-05,
+      "loss": 4.294,
+      "step": 1452
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.732779097387174e-05,
+      "loss": 4.445,
+      "step": 1453
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.732185273159146e-05,
+      "loss": 4.1096,
+      "step": 1454
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.731591448931117e-05,
+      "loss": 4.5229,
+      "step": 1455
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.730997624703089e-05,
+      "loss": 4.1706,
+      "step": 1456
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.73040380047506e-05,
+      "loss": 4.1121,
+      "step": 1457
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.729809976247031e-05,
+      "loss": 3.9869,
+      "step": 1458
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.729216152019003e-05,
+      "loss": 4.1061,
+      "step": 1459
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.728622327790974e-05,
+      "loss": 4.0387,
+      "step": 1460
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.728028503562946e-05,
+      "loss": 3.9587,
+      "step": 1461
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.727434679334917e-05,
+      "loss": 4.3285,
+      "step": 1462
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.726840855106889e-05,
+      "loss": 4.3117,
+      "step": 1463
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.72624703087886e-05,
+      "loss": 4.8798,
+      "step": 1464
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.725653206650832e-05,
+      "loss": 4.462,
+      "step": 1465
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.725059382422804e-05,
+      "loss": 4.288,
+      "step": 1466
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.724465558194776e-05,
+      "loss": 4.2433,
+      "step": 1467
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.723871733966746e-05,
+      "loss": 4.2105,
+      "step": 1468
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.723277909738718e-05,
+      "loss": 4.2794,
+      "step": 1469
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.722684085510689e-05,
+      "loss": 3.9473,
+      "step": 1470
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.72209026128266e-05,
+      "loss": 4.2169,
+      "step": 1471
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.721496437054632e-05,
+      "loss": 4.9202,
+      "step": 1472
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.720902612826604e-05,
+      "loss": 4.1318,
+      "step": 1473
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.720308788598576e-05,
+      "loss": 4.0178,
+      "step": 1474
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.719714964370547e-05,
+      "loss": 4.2188,
+      "step": 1475
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.719121140142519e-05,
+      "loss": 4.2682,
+      "step": 1476
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.718527315914489e-05,
+      "loss": 4.4745,
+      "step": 1477
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.717933491686461e-05,
+      "loss": 4.211,
+      "step": 1478
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.717339667458433e-05,
+      "loss": 4.2261,
+      "step": 1479
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.716745843230404e-05,
+      "loss": 4.2734,
+      "step": 1480
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.716152019002376e-05,
+      "loss": 4.3338,
+      "step": 1481
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.715558194774347e-05,
+      "loss": 4.2977,
+      "step": 1482
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.714964370546319e-05,
+      "loss": 4.1156,
+      "step": 1483
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.71437054631829e-05,
+      "loss": 4.3965,
+      "step": 1484
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.713776722090261e-05,
+      "loss": 4.9895,
+      "step": 1485
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.713182897862234e-05,
+      "loss": 4.5147,
+      "step": 1486
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.712589073634206e-05,
+      "loss": 4.4322,
+      "step": 1487
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.711995249406176e-05,
+      "loss": 4.3484,
+      "step": 1488
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.711401425178148e-05,
+      "loss": 4.5708,
+      "step": 1489
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.710807600950119e-05,
+      "loss": 4.5842,
+      "step": 1490
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.71021377672209e-05,
+      "loss": 4.4674,
+      "step": 1491
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.709619952494062e-05,
+      "loss": 4.2663,
+      "step": 1492
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.709026128266034e-05,
+      "loss": 4.1823,
+      "step": 1493
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.708432304038006e-05,
+      "loss": 4.2159,
+      "step": 1494
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.707838479809976e-05,
+      "loss": 4.3264,
+      "step": 1495
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.707244655581948e-05,
+      "loss": 4.1584,
+      "step": 1496
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.706650831353919e-05,
+      "loss": 4.3297,
+      "step": 1497
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.706057007125891e-05,
+      "loss": 4.2283,
+      "step": 1498
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.705463182897863e-05,
+      "loss": 4.265,
+      "step": 1499
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.704869358669835e-05,
+      "loss": 4.1667,
+      "step": 1500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.704275534441806e-05,
+      "loss": 4.4821,
+      "step": 1501
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.703681710213777e-05,
+      "loss": 4.4389,
+      "step": 1502
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.703087885985749e-05,
+      "loss": 4.168,
+      "step": 1503
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.702494061757719e-05,
+      "loss": 4.2614,
+      "step": 1504
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.701900237529691e-05,
+      "loss": 4.406,
+      "step": 1505
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.701306413301663e-05,
+      "loss": 4.0407,
+      "step": 1506
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.700712589073635e-05,
+      "loss": 4.207,
+      "step": 1507
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.700118764845606e-05,
+      "loss": 4.1764,
+      "step": 1508
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.699524940617578e-05,
+      "loss": 4.2739,
+      "step": 1509
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.698931116389549e-05,
+      "loss": 4.0963,
+      "step": 1510
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.698337292161521e-05,
+      "loss": 4.2545,
+      "step": 1511
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.697743467933492e-05,
+      "loss": 4.1082,
+      "step": 1512
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.697149643705464e-05,
+      "loss": 4.157,
+      "step": 1513
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.696555819477436e-05,
+      "loss": 4.6307,
+      "step": 1514
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.695961995249406e-05,
+      "loss": 4.3001,
+      "step": 1515
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.695368171021378e-05,
+      "loss": 4.1973,
+      "step": 1516
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.694774346793349e-05,
+      "loss": 4.1625,
+      "step": 1517
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.694180522565321e-05,
+      "loss": 4.2576,
+      "step": 1518
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.693586698337293e-05,
+      "loss": 4.1369,
+      "step": 1519
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.692992874109265e-05,
+      "loss": 4.1519,
+      "step": 1520
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.692399049881236e-05,
+      "loss": 4.1734,
+      "step": 1521
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.691805225653206e-05,
+      "loss": 4.495,
+      "step": 1522
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.691211401425179e-05,
+      "loss": 4.0354,
+      "step": 1523
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.690617577197149e-05,
+      "loss": 4.3285,
+      "step": 1524
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.690023752969121e-05,
+      "loss": 4.2471,
+      "step": 1525
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.689429928741093e-05,
+      "loss": 4.2783,
+      "step": 1526
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.688836104513065e-05,
+      "loss": 4.4574,
+      "step": 1527
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.688242280285036e-05,
+      "loss": 4.2896,
+      "step": 1528
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.687648456057008e-05,
+      "loss": 4.0481,
+      "step": 1529
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.687054631828979e-05,
+      "loss": 4.5021,
+      "step": 1530
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.686460807600951e-05,
+      "loss": 4.2584,
+      "step": 1531
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.685866983372923e-05,
+      "loss": 4.3685,
+      "step": 1532
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.685273159144893e-05,
+      "loss": 4.2734,
+      "step": 1533
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.684679334916865e-05,
+      "loss": 4.3808,
+      "step": 1534
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.684085510688836e-05,
+      "loss": 4.2464,
+      "step": 1535
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.683491686460808e-05,
+      "loss": 4.6061,
+      "step": 1536
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.682897862232779e-05,
+      "loss": 4.3647,
+      "step": 1537
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.682304038004751e-05,
+      "loss": 5.0083,
+      "step": 1538
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.681710213776723e-05,
+      "loss": 4.1827,
+      "step": 1539
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.681116389548695e-05,
+      "loss": 4.2275,
+      "step": 1540
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.680522565320666e-05,
+      "loss": 4.4101,
+      "step": 1541
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.679928741092638e-05,
+      "loss": 4.3097,
+      "step": 1542
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.679334916864608e-05,
+      "loss": 4.5104,
+      "step": 1543
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.678741092636579e-05,
+      "loss": 4.2735,
+      "step": 1544
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.678147268408551e-05,
+      "loss": 4.439,
+      "step": 1545
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.677553444180523e-05,
+      "loss": 4.1948,
+      "step": 1546
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.676959619952495e-05,
+      "loss": 4.2537,
+      "step": 1547
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.676365795724466e-05,
+      "loss": 4.1198,
+      "step": 1548
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.675771971496438e-05,
+      "loss": 4.433,
+      "step": 1549
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.675178147268409e-05,
+      "loss": 4.2581,
+      "step": 1550
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.67458432304038e-05,
+      "loss": 4.3466,
+      "step": 1551
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.673990498812353e-05,
+      "loss": 4.7106,
+      "step": 1552
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.673396674584323e-05,
+      "loss": 4.1474,
+      "step": 1553
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.672802850356295e-05,
+      "loss": 4.0709,
+      "step": 1554
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.672209026128266e-05,
+      "loss": 4.2768,
+      "step": 1555
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.671615201900238e-05,
+      "loss": 4.0824,
+      "step": 1556
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.671021377672209e-05,
+      "loss": 4.3274,
+      "step": 1557
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.670427553444181e-05,
+      "loss": 4.2951,
+      "step": 1558
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.669833729216153e-05,
+      "loss": 4.0788,
+      "step": 1559
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.669239904988125e-05,
+      "loss": 4.4068,
+      "step": 1560
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.668646080760096e-05,
+      "loss": 4.4109,
+      "step": 1561
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.668052256532068e-05,
+      "loss": 4.5331,
+      "step": 1562
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.667458432304038e-05,
+      "loss": 4.5664,
+      "step": 1563
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.666864608076009e-05,
+      "loss": 4.3291,
+      "step": 1564
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.666270783847981e-05,
+      "loss": 3.9441,
+      "step": 1565
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.665676959619953e-05,
+      "loss": 4.4099,
+      "step": 1566
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.665083135391925e-05,
+      "loss": 4.4323,
+      "step": 1567
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.664489311163896e-05,
+      "loss": 4.2168,
+      "step": 1568
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.663895486935868e-05,
+      "loss": 4.3002,
+      "step": 1569
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.663301662707838e-05,
+      "loss": 4.3666,
+      "step": 1570
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.66270783847981e-05,
+      "loss": 4.2715,
+      "step": 1571
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.662114014251783e-05,
+      "loss": 4.2721,
+      "step": 1572
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.661520190023755e-05,
+      "loss": 4.4518,
+      "step": 1573
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.660926365795725e-05,
+      "loss": 4.0633,
+      "step": 1574
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.660332541567696e-05,
+      "loss": 4.1192,
+      "step": 1575
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.659738717339668e-05,
+      "loss": 4.3142,
+      "step": 1576
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.659144893111639e-05,
+      "loss": 4.1945,
+      "step": 1577
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.65855106888361e-05,
+      "loss": 4.2115,
+      "step": 1578
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.657957244655583e-05,
+      "loss": 4.0866,
+      "step": 1579
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.657363420427555e-05,
+      "loss": 4.1643,
+      "step": 1580
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.656769596199525e-05,
+      "loss": 4.1644,
+      "step": 1581
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.656175771971497e-05,
+      "loss": 4.211,
+      "step": 1582
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.655581947743468e-05,
+      "loss": 4.127,
+      "step": 1583
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.654988123515439e-05,
+      "loss": 4.2726,
+      "step": 1584
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.654394299287411e-05,
+      "loss": 4.349,
+      "step": 1585
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.653800475059383e-05,
+      "loss": 4.191,
+      "step": 1586
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.653206650831355e-05,
+      "loss": 4.244,
+      "step": 1587
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.652612826603326e-05,
+      "loss": 4.3755,
+      "step": 1588
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.652019002375298e-05,
+      "loss": 4.9399,
+      "step": 1589
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.651425178147268e-05,
+      "loss": 4.4854,
+      "step": 1590
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.65083135391924e-05,
+      "loss": 4.4845,
+      "step": 1591
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.650237529691212e-05,
+      "loss": 4.0511,
+      "step": 1592
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.649643705463184e-05,
+      "loss": 4.7299,
+      "step": 1593
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.649049881235155e-05,
+      "loss": 4.2234,
+      "step": 1594
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.648456057007126e-05,
+      "loss": 4.2185,
+      "step": 1595
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.647862232779098e-05,
+      "loss": 4.4352,
+      "step": 1596
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.647268408551068e-05,
+      "loss": 4.1786,
+      "step": 1597
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.64667458432304e-05,
+      "loss": 4.1917,
+      "step": 1598
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.646080760095013e-05,
+      "loss": 4.2939,
+      "step": 1599
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.645486935866985e-05,
+      "loss": 4.5956,
+      "step": 1600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.644893111638955e-05,
+      "loss": 4.4033,
+      "step": 1601
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.644299287410927e-05,
+      "loss": 4.4716,
+      "step": 1602
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.643705463182898e-05,
+      "loss": 4.3027,
+      "step": 1603
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.64311163895487e-05,
+      "loss": 4.1509,
+      "step": 1604
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.642517814726841e-05,
+      "loss": 4.4022,
+      "step": 1605
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.641923990498813e-05,
+      "loss": 4.2769,
+      "step": 1606
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.641330166270785e-05,
+      "loss": 4.4685,
+      "step": 1607
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.640736342042755e-05,
+      "loss": 4.2132,
+      "step": 1608
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.640142517814727e-05,
+      "loss": 4.1678,
+      "step": 1609
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.639548693586698e-05,
+      "loss": 4.1829,
+      "step": 1610
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.63895486935867e-05,
+      "loss": 4.438,
+      "step": 1611
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.638361045130642e-05,
+      "loss": 4.2825,
+      "step": 1612
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.637767220902614e-05,
+      "loss": 4.2355,
+      "step": 1613
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.637173396674585e-05,
+      "loss": 4.316,
+      "step": 1614
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.636579572446556e-05,
+      "loss": 4.3366,
+      "step": 1615
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.635985748218528e-05,
+      "loss": 4.1831,
+      "step": 1616
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.635391923990498e-05,
+      "loss": 4.2545,
+      "step": 1617
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.63479809976247e-05,
+      "loss": 4.5595,
+      "step": 1618
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.634204275534442e-05,
+      "loss": 4.306,
+      "step": 1619
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.633610451306414e-05,
+      "loss": 4.27,
+      "step": 1620
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.633016627078385e-05,
+      "loss": 4.2785,
+      "step": 1621
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.632422802850357e-05,
+      "loss": 4.0893,
+      "step": 1622
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.631828978622328e-05,
+      "loss": 4.6126,
+      "step": 1623
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.6312351543943e-05,
+      "loss": 4.108,
+      "step": 1624
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.63064133016627e-05,
+      "loss": 4.4504,
+      "step": 1625
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.630047505938243e-05,
+      "loss": 4.0397,
+      "step": 1626
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.629453681710215e-05,
+      "loss": 4.0649,
+      "step": 1627
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.628859857482185e-05,
+      "loss": 4.2987,
+      "step": 1628
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.628266033254157e-05,
+      "loss": 4.1208,
+      "step": 1629
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.627672209026128e-05,
+      "loss": 4.1263,
+      "step": 1630
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.6270783847981e-05,
+      "loss": 3.9581,
+      "step": 1631
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.626484560570072e-05,
+      "loss": 4.3145,
+      "step": 1632
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.625890736342044e-05,
+      "loss": 4.4159,
+      "step": 1633
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.625296912114015e-05,
+      "loss": 4.2869,
+      "step": 1634
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.624703087885987e-05,
+      "loss": 4.583,
+      "step": 1635
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.624109263657958e-05,
+      "loss": 4.1255,
+      "step": 1636
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.623515439429928e-05,
+      "loss": 4.2918,
+      "step": 1637
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.6229216152019e-05,
+      "loss": 4.1264,
+      "step": 1638
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.622327790973872e-05,
+      "loss": 4.3896,
+      "step": 1639
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.621733966745844e-05,
+      "loss": 3.9866,
+      "step": 1640
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.621140142517815e-05,
+      "loss": 5.2777,
+      "step": 1641
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.620546318289787e-05,
+      "loss": 3.8689,
+      "step": 1642
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.619952494061758e-05,
+      "loss": 4.2341,
+      "step": 1643
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.61935866983373e-05,
+      "loss": 4.4449,
+      "step": 1644
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.6187648456057e-05,
+      "loss": 4.2204,
+      "step": 1645
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.618171021377672e-05,
+      "loss": 4.3777,
+      "step": 1646
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.617577197149645e-05,
+      "loss": 4.1052,
+      "step": 1647
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.616983372921615e-05,
+      "loss": 4.199,
+      "step": 1648
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.616389548693587e-05,
+      "loss": 4.3598,
+      "step": 1649
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.615795724465558e-05,
+      "loss": 4.5093,
+      "step": 1650
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.61520190023753e-05,
+      "loss": 3.9398,
+      "step": 1651
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.614608076009502e-05,
+      "loss": 4.4636,
+      "step": 1652
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.614014251781474e-05,
+      "loss": 3.9443,
+      "step": 1653
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.613420427553445e-05,
+      "loss": 4.2371,
+      "step": 1654
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.612826603325417e-05,
+      "loss": 4.262,
+      "step": 1655
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.612232779097387e-05,
+      "loss": 4.2601,
+      "step": 1656
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.611638954869358e-05,
+      "loss": 4.3374,
+      "step": 1657
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.61104513064133e-05,
+      "loss": 4.2747,
+      "step": 1658
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.610451306413302e-05,
+      "loss": 4.1254,
+      "step": 1659
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.609857482185274e-05,
+      "loss": 4.1879,
+      "step": 1660
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.609263657957245e-05,
+      "loss": 4.2419,
+      "step": 1661
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.608669833729217e-05,
+      "loss": 4.2899,
+      "step": 1662
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.608076009501188e-05,
+      "loss": 4.2525,
+      "step": 1663
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.60748218527316e-05,
+      "loss": 4.4076,
+      "step": 1664
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.606888361045132e-05,
+      "loss": 4.1646,
+      "step": 1665
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.606294536817104e-05,
+      "loss": 4.2739,
+      "step": 1666
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.605700712589074e-05,
+      "loss": 4.1789,
+      "step": 1667
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.605106888361045e-05,
+      "loss": 4.3275,
+      "step": 1668
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.604513064133017e-05,
+      "loss": 4.2096,
+      "step": 1669
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.603919239904988e-05,
+      "loss": 4.2592,
+      "step": 1670
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.60332541567696e-05,
+      "loss": 4.0019,
+      "step": 1671
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.602731591448932e-05,
+      "loss": 4.0745,
+      "step": 1672
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.602137767220904e-05,
+      "loss": 4.7949,
+      "step": 1673
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.601543942992875e-05,
+      "loss": 4.3798,
+      "step": 1674
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.600950118764847e-05,
+      "loss": 4.0885,
+      "step": 1675
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.600356294536817e-05,
+      "loss": 4.3752,
+      "step": 1676
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.59976247030879e-05,
+      "loss": 4.1144,
+      "step": 1677
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.59916864608076e-05,
+      "loss": 4.0393,
+      "step": 1678
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.598574821852732e-05,
+      "loss": 4.6357,
+      "step": 1679
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.597980997624704e-05,
+      "loss": 4.0906,
+      "step": 1680
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.597387173396675e-05,
+      "loss": 4.3847,
+      "step": 1681
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.596793349168647e-05,
+      "loss": 4.1169,
+      "step": 1682
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.596199524940617e-05,
+      "loss": 4.4335,
+      "step": 1683
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.59560570071259e-05,
+      "loss": 4.0529,
+      "step": 1684
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.595011876484562e-05,
+      "loss": 4.2258,
+      "step": 1685
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.594418052256534e-05,
+      "loss": 4.2188,
+      "step": 1686
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.593824228028504e-05,
+      "loss": 4.5329,
+      "step": 1687
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.593230403800475e-05,
+      "loss": 4.1853,
+      "step": 1688
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.592636579572447e-05,
+      "loss": 5.6473,
+      "step": 1689
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.592042755344418e-05,
+      "loss": 4.6438,
+      "step": 1690
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.59144893111639e-05,
+      "loss": 4.4361,
+      "step": 1691
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.590855106888362e-05,
+      "loss": 3.9353,
+      "step": 1692
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.590261282660334e-05,
+      "loss": 4.1754,
+      "step": 1693
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.589667458432304e-05,
+      "loss": 4.7416,
+      "step": 1694
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.589073634204276e-05,
+      "loss": 4.3098,
+      "step": 1695
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.588479809976247e-05,
+      "loss": 4.2658,
+      "step": 1696
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.587885985748219e-05,
+      "loss": 4.1401,
+      "step": 1697
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.58729216152019e-05,
+      "loss": 4.28,
+      "step": 1698
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.586698337292162e-05,
+      "loss": 4.1508,
+      "step": 1699
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.586104513064134e-05,
+      "loss": 4.4967,
+      "step": 1700
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.585510688836105e-05,
+      "loss": 4.2288,
+      "step": 1701
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.584916864608077e-05,
+      "loss": 4.2901,
+      "step": 1702
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.584323040380047e-05,
+      "loss": 4.01,
+      "step": 1703
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.58372921615202e-05,
+      "loss": 4.378,
+      "step": 1704
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.583135391923991e-05,
+      "loss": 4.4804,
+      "step": 1705
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.582541567695963e-05,
+      "loss": 4.028,
+      "step": 1706
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.581947743467934e-05,
+      "loss": 3.9743,
+      "step": 1707
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.581353919239906e-05,
+      "loss": 4.3011,
+      "step": 1708
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.580760095011877e-05,
+      "loss": 4.1696,
+      "step": 1709
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.580166270783848e-05,
+      "loss": 3.9831,
+      "step": 1710
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.57957244655582e-05,
+      "loss": 4.0186,
+      "step": 1711
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.578978622327792e-05,
+      "loss": 4.1186,
+      "step": 1712
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.578384798099764e-05,
+      "loss": 4.0852,
+      "step": 1713
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.577790973871734e-05,
+      "loss": 3.9722,
+      "step": 1714
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.577197149643706e-05,
+      "loss": 4.1281,
+      "step": 1715
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.576603325415677e-05,
+      "loss": 4.4702,
+      "step": 1716
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.576009501187649e-05,
+      "loss": 4.4904,
+      "step": 1717
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.57541567695962e-05,
+      "loss": 4.1465,
+      "step": 1718
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.574821852731592e-05,
+      "loss": 4.3882,
+      "step": 1719
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.574228028503564e-05,
+      "loss": 4.0788,
+      "step": 1720
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.573634204275535e-05,
+      "loss": 4.2593,
+      "step": 1721
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.573040380047507e-05,
+      "loss": 4.1902,
+      "step": 1722
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.572446555819477e-05,
+      "loss": 4.3995,
+      "step": 1723
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.571852731591449e-05,
+      "loss": 4.0596,
+      "step": 1724
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.571258907363421e-05,
+      "loss": 4.0745,
+      "step": 1725
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.570665083135393e-05,
+      "loss": 4.1768,
+      "step": 1726
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.570071258907364e-05,
+      "loss": 3.9723,
+      "step": 1727
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.569477434679336e-05,
+      "loss": 4.3541,
+      "step": 1728
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.568883610451307e-05,
+      "loss": 4.1805,
+      "step": 1729
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.568289786223277e-05,
+      "loss": 4.1069,
+      "step": 1730
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.56769596199525e-05,
+      "loss": 4.0984,
+      "step": 1731
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.567102137767221e-05,
+      "loss": 4.4418,
+      "step": 1732
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.566508313539194e-05,
+      "loss": 4.2336,
+      "step": 1733
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.565914489311164e-05,
+      "loss": 4.0743,
+      "step": 1734
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.565320665083136e-05,
+      "loss": 4.0848,
+      "step": 1735
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.564726840855107e-05,
+      "loss": 4.5083,
+      "step": 1736
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.564133016627079e-05,
+      "loss": 4.1161,
+      "step": 1737
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.56353919239905e-05,
+      "loss": 4.2335,
+      "step": 1738
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.562945368171023e-05,
+      "loss": 4.248,
+      "step": 1739
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.562351543942994e-05,
+      "loss": 4.1143,
+      "step": 1740
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.561757719714964e-05,
+      "loss": 3.8488,
+      "step": 1741
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.561163895486936e-05,
+      "loss": 4.491,
+      "step": 1742
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.560570071258907e-05,
+      "loss": 4.1543,
+      "step": 1743
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.559976247030879e-05,
+      "loss": 4.0953,
+      "step": 1744
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.559382422802851e-05,
+      "loss": 4.5162,
+      "step": 1745
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.558788598574823e-05,
+      "loss": 4.1187,
+      "step": 1746
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.558194774346794e-05,
+      "loss": 4.3129,
+      "step": 1747
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.557600950118766e-05,
+      "loss": 4.0262,
+      "step": 1748
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.557007125890737e-05,
+      "loss": 4.703,
+      "step": 1749
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.556413301662707e-05,
+      "loss": 4.5655,
+      "step": 1750
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.555819477434679e-05,
+      "loss": 4.231,
+      "step": 1751
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.555225653206651e-05,
+      "loss": 4.2313,
+      "step": 1752
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.554631828978623e-05,
+      "loss": 4.312,
+      "step": 1753
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.554038004750594e-05,
+      "loss": 4.1246,
+      "step": 1754
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.553444180522566e-05,
+      "loss": 4.1037,
+      "step": 1755
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.552850356294537e-05,
+      "loss": 4.2215,
+      "step": 1756
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.552256532066509e-05,
+      "loss": 4.7056,
+      "step": 1757
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.55166270783848e-05,
+      "loss": 4.4011,
+      "step": 1758
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.551068883610453e-05,
+      "loss": 3.9713,
+      "step": 1759
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.550475059382424e-05,
+      "loss": 4.0985,
+      "step": 1760
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.549881235154394e-05,
+      "loss": 4.2046,
+      "step": 1761
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.549287410926366e-05,
+      "loss": 4.551,
+      "step": 1762
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.548693586698337e-05,
+      "loss": 4.4012,
+      "step": 1763
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.548099762470309e-05,
+      "loss": 4.1437,
+      "step": 1764
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.547505938242281e-05,
+      "loss": 3.9546,
+      "step": 1765
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.546912114014253e-05,
+      "loss": 4.1364,
+      "step": 1766
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.546318289786224e-05,
+      "loss": 4.1632,
+      "step": 1767
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.545724465558196e-05,
+      "loss": 4.36,
+      "step": 1768
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.545130641330166e-05,
+      "loss": 3.9983,
+      "step": 1769
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.544536817102139e-05,
+      "loss": 4.0201,
+      "step": 1770
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.543942992874109e-05,
+      "loss": 4.2814,
+      "step": 1771
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.543349168646081e-05,
+      "loss": 4.1499,
+      "step": 1772
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.542755344418053e-05,
+      "loss": 3.9681,
+      "step": 1773
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.542161520190024e-05,
+      "loss": 4.0189,
+      "step": 1774
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.541567695961996e-05,
+      "loss": 4.0786,
+      "step": 1775
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.540973871733967e-05,
+      "loss": 4.3777,
+      "step": 1776
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.540380047505939e-05,
+      "loss": 4.0744,
+      "step": 1777
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.539786223277911e-05,
+      "loss": 4.5677,
+      "step": 1778
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.539192399049883e-05,
+      "loss": 4.3892,
+      "step": 1779
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.538598574821853e-05,
+      "loss": 4.2255,
+      "step": 1780
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.538004750593824e-05,
+      "loss": 5.0142,
+      "step": 1781
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.537410926365796e-05,
+      "loss": 4.1532,
+      "step": 1782
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.536817102137767e-05,
+      "loss": 4.9852,
+      "step": 1783
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.536223277909739e-05,
+      "loss": 3.7744,
+      "step": 1784
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.535629453681711e-05,
+      "loss": 3.853,
+      "step": 1785
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.535035629453683e-05,
+      "loss": 3.7358,
+      "step": 1786
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.534441805225654e-05,
+      "loss": 3.9738,
+      "step": 1787
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.533847980997626e-05,
+      "loss": 3.8856,
+      "step": 1788
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.533254156769596e-05,
+      "loss": 4.0318,
+      "step": 1789
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.532660332541568e-05,
+      "loss": 3.9099,
+      "step": 1790
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.532066508313539e-05,
+      "loss": 3.9777,
+      "step": 1791
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.531472684085511e-05,
+      "loss": 4.2006,
+      "step": 1792
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.530878859857483e-05,
+      "loss": 3.8265,
+      "step": 1793
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.530285035629454e-05,
+      "loss": 3.8738,
+      "step": 1794
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.529691211401426e-05,
+      "loss": 4.3039,
+      "step": 1795
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.529097387173397e-05,
+      "loss": 3.8549,
+      "step": 1796
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.528503562945369e-05,
+      "loss": 3.902,
+      "step": 1797
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.52790973871734e-05,
+      "loss": 3.6783,
+      "step": 1798
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.527315914489313e-05,
+      "loss": 4.1442,
+      "step": 1799
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.526722090261283e-05,
+      "loss": 3.9854,
+      "step": 1800
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.526128266033255e-05,
+      "loss": 3.8976,
+      "step": 1801
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.525534441805226e-05,
+      "loss": 3.9677,
+      "step": 1802
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.524940617577197e-05,
+      "loss": 3.9179,
+      "step": 1803
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.524346793349169e-05,
+      "loss": 3.7239,
+      "step": 1804
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.523752969121141e-05,
+      "loss": 3.6542,
+      "step": 1805
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.523159144893113e-05,
+      "loss": 3.6801,
+      "step": 1806
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.522565320665083e-05,
+      "loss": 3.9706,
+      "step": 1807
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.521971496437056e-05,
+      "loss": 4.0895,
+      "step": 1808
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.521377672209026e-05,
+      "loss": 3.8656,
+      "step": 1809
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.520783847980998e-05,
+      "loss": 3.756,
+      "step": 1810
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.520190023752969e-05,
+      "loss": 3.8435,
+      "step": 1811
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.519596199524941e-05,
+      "loss": 3.7798,
+      "step": 1812
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.519002375296913e-05,
+      "loss": 3.7547,
+      "step": 1813
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.518408551068884e-05,
+      "loss": 3.8161,
+      "step": 1814
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.517814726840856e-05,
+      "loss": 3.9586,
+      "step": 1815
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.517220902612826e-05,
+      "loss": 3.8294,
+      "step": 1816
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.516627078384798e-05,
+      "loss": 3.6153,
+      "step": 1817
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.51603325415677e-05,
+      "loss": 3.7703,
+      "step": 1818
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.515439429928743e-05,
+      "loss": 3.8353,
+      "step": 1819
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.514845605700713e-05,
+      "loss": 3.483,
+      "step": 1820
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.514251781472685e-05,
+      "loss": 3.8574,
+      "step": 1821
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.513657957244656e-05,
+      "loss": 3.8388,
+      "step": 1822
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.513064133016627e-05,
+      "loss": 3.7562,
+      "step": 1823
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.512470308788599e-05,
+      "loss": 3.7407,
+      "step": 1824
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.51187648456057e-05,
+      "loss": 3.7423,
+      "step": 1825
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.511282660332543e-05,
+      "loss": 3.955,
+      "step": 1826
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.510688836104513e-05,
+      "loss": 3.7274,
+      "step": 1827
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.510095011876485e-05,
+      "loss": 3.6471,
+      "step": 1828
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.509501187648456e-05,
+      "loss": 3.5441,
+      "step": 1829
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.508907363420428e-05,
+      "loss": 3.9466,
+      "step": 1830
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.508313539192399e-05,
+      "loss": 4.1684,
+      "step": 1831
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.507719714964372e-05,
+      "loss": 4.8085,
+      "step": 1832
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.507125890736343e-05,
+      "loss": 3.8518,
+      "step": 1833
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.506532066508314e-05,
+      "loss": 3.5879,
+      "step": 1834
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.505938242280286e-05,
+      "loss": 4.0974,
+      "step": 1835
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.505344418052256e-05,
+      "loss": 4.1797,
+      "step": 1836
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.504750593824228e-05,
+      "loss": 3.8017,
+      "step": 1837
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.5041567695962e-05,
+      "loss": 3.8318,
+      "step": 1838
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.503562945368172e-05,
+      "loss": 4.0839,
+      "step": 1839
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.502969121140143e-05,
+      "loss": 3.6026,
+      "step": 1840
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.502375296912115e-05,
+      "loss": 3.896,
+      "step": 1841
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.501781472684086e-05,
+      "loss": 3.8405,
+      "step": 1842
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.501187648456056e-05,
+      "loss": 3.7653,
+      "step": 1843
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.500593824228028e-05,
+      "loss": 3.6986,
+      "step": 1844
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.5e-05,
+      "loss": 3.7552,
+      "step": 1845
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.499406175771973e-05,
+      "loss": 3.8847,
+      "step": 1846
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.498812351543943e-05,
+      "loss": 3.9502,
+      "step": 1847
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.498218527315915e-05,
+      "loss": 4.0071,
+      "step": 1848
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.497624703087886e-05,
+      "loss": 4.0183,
+      "step": 1849
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.497030878859858e-05,
+      "loss": 3.8775,
+      "step": 1850
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.496437054631829e-05,
+      "loss": 4.1459,
+      "step": 1851
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.495843230403802e-05,
+      "loss": 4.0685,
+      "step": 1852
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.495249406175773e-05,
+      "loss": 4.0627,
+      "step": 1853
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.494655581947743e-05,
+      "loss": 4.1835,
+      "step": 1854
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.494061757719715e-05,
+      "loss": 4.0042,
+      "step": 1855
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.493467933491686e-05,
+      "loss": 3.7133,
+      "step": 1856
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.492874109263658e-05,
+      "loss": 3.7688,
+      "step": 1857
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.49228028503563e-05,
+      "loss": 3.7688,
+      "step": 1858
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.491686460807602e-05,
+      "loss": 4.0088,
+      "step": 1859
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.491092636579573e-05,
+      "loss": 3.7115,
+      "step": 1860
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.490498812351545e-05,
+      "loss": 3.9357,
+      "step": 1861
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.489904988123516e-05,
+      "loss": 3.5856,
+      "step": 1862
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.489311163895488e-05,
+      "loss": 3.6749,
+      "step": 1863
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.488717339667458e-05,
+      "loss": 3.7522,
+      "step": 1864
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.48812351543943e-05,
+      "loss": 3.9344,
+      "step": 1865
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.487529691211402e-05,
+      "loss": 3.8,
+      "step": 1866
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.486935866983373e-05,
+      "loss": 3.9871,
+      "step": 1867
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.486342042755345e-05,
+      "loss": 3.836,
+      "step": 1868
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.485748218527316e-05,
+      "loss": 3.8661,
+      "step": 1869
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.485154394299288e-05,
+      "loss": 3.7306,
+      "step": 1870
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.484560570071259e-05,
+      "loss": 3.983,
+      "step": 1871
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.483966745843232e-05,
+      "loss": 3.7257,
+      "step": 1872
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.483372921615203e-05,
+      "loss": 3.8124,
+      "step": 1873
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.482779097387175e-05,
+      "loss": 3.8737,
+      "step": 1874
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.482185273159145e-05,
+      "loss": 3.9062,
+      "step": 1875
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.481591448931116e-05,
+      "loss": 3.8507,
+      "step": 1876
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.480997624703088e-05,
+      "loss": 3.8679,
+      "step": 1877
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.48040380047506e-05,
+      "loss": 3.8116,
+      "step": 1878
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.479809976247032e-05,
+      "loss": 3.8318,
+      "step": 1879
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.479216152019003e-05,
+      "loss": 4.0547,
+      "step": 1880
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.478622327790975e-05,
+      "loss": 3.6798,
+      "step": 1881
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.478028503562946e-05,
+      "loss": 4.4849,
+      "step": 1882
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.477434679334918e-05,
+      "loss": 3.9434,
+      "step": 1883
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.476840855106888e-05,
+      "loss": 3.9218,
+      "step": 1884
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.47624703087886e-05,
+      "loss": 4.1803,
+      "step": 1885
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.475653206650832e-05,
+      "loss": 4.0016,
+      "step": 1886
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.475059382422803e-05,
+      "loss": 4.145,
+      "step": 1887
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.474465558194775e-05,
+      "loss": 3.8218,
+      "step": 1888
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.473871733966746e-05,
+      "loss": 3.8106,
+      "step": 1889
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.473277909738718e-05,
+      "loss": 3.7448,
+      "step": 1890
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.472684085510688e-05,
+      "loss": 4.0239,
+      "step": 1891
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.472090261282662e-05,
+      "loss": 3.8824,
+      "step": 1892
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.471496437054632e-05,
+      "loss": 4.3555,
+      "step": 1893
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.470902612826605e-05,
+      "loss": 3.9832,
+      "step": 1894
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.470308788598575e-05,
+      "loss": 3.771,
+      "step": 1895
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.469714964370546e-05,
+      "loss": 3.765,
+      "step": 1896
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.469121140142518e-05,
+      "loss": 3.7343,
+      "step": 1897
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.46852731591449e-05,
+      "loss": 4.0804,
+      "step": 1898
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.467933491686462e-05,
+      "loss": 3.8938,
+      "step": 1899
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.467339667458433e-05,
+      "loss": 3.7207,
+      "step": 1900
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.466745843230405e-05,
+      "loss": 3.8443,
+      "step": 1901
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.466152019002375e-05,
+      "loss": 3.8877,
+      "step": 1902
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.465558194774347e-05,
+      "loss": 3.7304,
+      "step": 1903
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.464964370546318e-05,
+      "loss": 3.6802,
+      "step": 1904
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.46437054631829e-05,
+      "loss": 4.1135,
+      "step": 1905
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.463776722090262e-05,
+      "loss": 4.2347,
+      "step": 1906
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.463182897862233e-05,
+      "loss": 3.8311,
+      "step": 1907
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.462589073634205e-05,
+      "loss": 4.0919,
+      "step": 1908
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.461995249406176e-05,
+      "loss": 3.7767,
+      "step": 1909
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.461401425178148e-05,
+      "loss": 3.7452,
+      "step": 1910
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.46080760095012e-05,
+      "loss": 3.6634,
+      "step": 1911
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.460213776722092e-05,
+      "loss": 4.1031,
+      "step": 1912
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.459619952494062e-05,
+      "loss": 4.1899,
+      "step": 1913
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.459026128266034e-05,
+      "loss": 4.0701,
+      "step": 1914
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.458432304038005e-05,
+      "loss": 3.8868,
+      "step": 1915
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.457838479809976e-05,
+      "loss": 3.6908,
+      "step": 1916
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.457244655581948e-05,
+      "loss": 3.7068,
+      "step": 1917
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.45665083135392e-05,
+      "loss": 3.9562,
+      "step": 1918
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.456057007125892e-05,
+      "loss": 3.695,
+      "step": 1919
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.455463182897863e-05,
+      "loss": 3.8667,
+      "step": 1920
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.454869358669835e-05,
+      "loss": 3.6668,
+      "step": 1921
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.454275534441805e-05,
+      "loss": 3.7616,
+      "step": 1922
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.453681710213777e-05,
+      "loss": 3.8644,
+      "step": 1923
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.453087885985748e-05,
+      "loss": 3.6665,
+      "step": 1924
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.45249406175772e-05,
+      "loss": 3.8519,
+      "step": 1925
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.451900237529692e-05,
+      "loss": 3.6134,
+      "step": 1926
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.451306413301663e-05,
+      "loss": 3.6783,
+      "step": 1927
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.450712589073635e-05,
+      "loss": 3.4762,
+      "step": 1928
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.450118764845605e-05,
+      "loss": 4.1754,
+      "step": 1929
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.449524940617577e-05,
+      "loss": 3.4711,
+      "step": 1930
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.44893111638955e-05,
+      "loss": 3.7814,
+      "step": 1931
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.448337292161522e-05,
+      "loss": 3.8611,
+      "step": 1932
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.447743467933492e-05,
+      "loss": 3.939,
+      "step": 1933
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.447149643705464e-05,
+      "loss": 3.7128,
+      "step": 1934
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.446555819477435e-05,
+      "loss": 4.0181,
+      "step": 1935
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.445961995249407e-05,
+      "loss": 4.0318,
+      "step": 1936
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.445368171021378e-05,
+      "loss": 3.7151,
+      "step": 1937
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.44477434679335e-05,
+      "loss": 3.964,
+      "step": 1938
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.444180522565322e-05,
+      "loss": 4.0393,
+      "step": 1939
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.443586698337292e-05,
+      "loss": 4.0118,
+      "step": 1940
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.442992874109264e-05,
+      "loss": 3.8236,
+      "step": 1941
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.442399049881235e-05,
+      "loss": 3.8811,
+      "step": 1942
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.441805225653207e-05,
+      "loss": 3.8578,
+      "step": 1943
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.441211401425178e-05,
+      "loss": 3.7324,
+      "step": 1944
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.440617577197151e-05,
+      "loss": 4.0404,
+      "step": 1945
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.440023752969122e-05,
+      "loss": 3.9863,
+      "step": 1946
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.439429928741093e-05,
+      "loss": 3.7555,
+      "step": 1947
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.438836104513065e-05,
+      "loss": 3.8121,
+      "step": 1948
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.438242280285035e-05,
+      "loss": 4.241,
+      "step": 1949
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.437648456057007e-05,
+      "loss": 3.7697,
+      "step": 1950
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.43705463182898e-05,
+      "loss": 3.7029,
+      "step": 1951
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.436460807600951e-05,
+      "loss": 3.5954,
+      "step": 1952
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.435866983372922e-05,
+      "loss": 3.7441,
+      "step": 1953
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.435273159144894e-05,
+      "loss": 3.8349,
+      "step": 1954
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.434679334916865e-05,
+      "loss": 3.8642,
+      "step": 1955
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.434085510688837e-05,
+      "loss": 3.7598,
+      "step": 1956
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.433491686460808e-05,
+      "loss": 3.9189,
+      "step": 1957
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.43289786223278e-05,
+      "loss": 3.6961,
+      "step": 1958
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.432304038004752e-05,
+      "loss": 3.679,
+      "step": 1959
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.431710213776722e-05,
+      "loss": 4.0866,
+      "step": 1960
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.431116389548694e-05,
+      "loss": 3.7061,
+      "step": 1961
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.430522565320665e-05,
+      "loss": 3.8556,
+      "step": 1962
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.429928741092637e-05,
+      "loss": 3.9506,
+      "step": 1963
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.429334916864608e-05,
+      "loss": 3.8813,
+      "step": 1964
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.428741092636581e-05,
+      "loss": 3.7952,
+      "step": 1965
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.428147268408552e-05,
+      "loss": 3.6712,
+      "step": 1966
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.427553444180524e-05,
+      "loss": 3.6452,
+      "step": 1967
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.426959619952495e-05,
+      "loss": 3.6997,
+      "step": 1968
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.426365795724465e-05,
+      "loss": 4.0393,
+      "step": 1969
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.425771971496437e-05,
+      "loss": 3.7712,
+      "step": 1970
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.425178147268409e-05,
+      "loss": 3.8342,
+      "step": 1971
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.424584323040381e-05,
+      "loss": 3.9219,
+      "step": 1972
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.423990498812352e-05,
+      "loss": 3.5528,
+      "step": 1973
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.423396674584324e-05,
+      "loss": 4.0666,
+      "step": 1974
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.422802850356295e-05,
+      "loss": 3.9581,
+      "step": 1975
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.422209026128267e-05,
+      "loss": 3.8004,
+      "step": 1976
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.421615201900237e-05,
+      "loss": 3.9038,
+      "step": 1977
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.42102137767221e-05,
+      "loss": 3.7618,
+      "step": 1978
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.420427553444181e-05,
+      "loss": 3.6414,
+      "step": 1979
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.419833729216152e-05,
+      "loss": 4.1028,
+      "step": 1980
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.419239904988124e-05,
+      "loss": 3.5554,
+      "step": 1981
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.418646080760095e-05,
+      "loss": 3.8719,
+      "step": 1982
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.418052256532067e-05,
+      "loss": 3.8998,
+      "step": 1983
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.417458432304038e-05,
+      "loss": 3.6846,
+      "step": 1984
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.416864608076011e-05,
+      "loss": 4.0649,
+      "step": 1985
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.416270783847982e-05,
+      "loss": 3.911,
+      "step": 1986
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.415676959619954e-05,
+      "loss": 4.1922,
+      "step": 1987
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.415083135391924e-05,
+      "loss": 4.0347,
+      "step": 1988
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.414489311163895e-05,
+      "loss": 3.9544,
+      "step": 1989
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.413895486935867e-05,
+      "loss": 3.8124,
+      "step": 1990
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.413301662707839e-05,
+      "loss": 4.0515,
+      "step": 1991
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.412707838479811e-05,
+      "loss": 3.9833,
+      "step": 1992
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.412114014251782e-05,
+      "loss": 3.6887,
+      "step": 1993
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.411520190023754e-05,
+      "loss": 3.8622,
+      "step": 1994
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.410926365795725e-05,
+      "loss": 3.8257,
+      "step": 1995
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.410332541567697e-05,
+      "loss": 3.921,
+      "step": 1996
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.409738717339667e-05,
+      "loss": 3.8157,
+      "step": 1997
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.409144893111639e-05,
+      "loss": 3.7706,
+      "step": 1998
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.408551068883611e-05,
+      "loss": 3.9073,
+      "step": 1999
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.407957244655582e-05,
+      "loss": 3.8361,
+      "step": 2000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.407363420427554e-05,
+      "loss": 3.9597,
+      "step": 2001
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.406769596199525e-05,
+      "loss": 3.6773,
+      "step": 2002
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.406175771971497e-05,
+      "loss": 3.8423,
+      "step": 2003
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.405581947743467e-05,
+      "loss": 3.811,
+      "step": 2004
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.404988123515441e-05,
+      "loss": 3.5753,
+      "step": 2005
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.404394299287412e-05,
+      "loss": 3.849,
+      "step": 2006
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.403800475059384e-05,
+      "loss": 3.8106,
+      "step": 2007
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.403206650831354e-05,
+      "loss": 4.0251,
+      "step": 2008
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.402612826603325e-05,
+      "loss": 4.1087,
+      "step": 2009
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.402019002375297e-05,
+      "loss": 4.2084,
+      "step": 2010
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.401425178147269e-05,
+      "loss": 4.1282,
+      "step": 2011
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.400831353919241e-05,
+      "loss": 3.7762,
+      "step": 2012
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.400237529691212e-05,
+      "loss": 3.8588,
+      "step": 2013
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.399643705463184e-05,
+      "loss": 3.9858,
+      "step": 2014
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.399049881235154e-05,
+      "loss": 4.2191,
+      "step": 2015
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.398456057007126e-05,
+      "loss": 3.6513,
+      "step": 2016
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.397862232779097e-05,
+      "loss": 3.8899,
+      "step": 2017
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.397268408551069e-05,
+      "loss": 3.6812,
+      "step": 2018
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.396674584323041e-05,
+      "loss": 3.9128,
+      "step": 2019
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.396080760095012e-05,
+      "loss": 3.9739,
+      "step": 2020
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.395486935866984e-05,
+      "loss": 4.0342,
+      "step": 2021
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.394893111638955e-05,
+      "loss": 3.9102,
+      "step": 2022
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.394299287410927e-05,
+      "loss": 3.9781,
+      "step": 2023
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.393705463182899e-05,
+      "loss": 4.043,
+      "step": 2024
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.393111638954871e-05,
+      "loss": 3.9896,
+      "step": 2025
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.392517814726841e-05,
+      "loss": 3.7685,
+      "step": 2026
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.391923990498813e-05,
+      "loss": 4.5375,
+      "step": 2027
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.391330166270784e-05,
+      "loss": 3.7761,
+      "step": 2028
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.390736342042756e-05,
+      "loss": 3.8048,
+      "step": 2029
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.390142517814727e-05,
+      "loss": 3.9209,
+      "step": 2030
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.389548693586699e-05,
+      "loss": 3.8157,
+      "step": 2031
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.388954869358671e-05,
+      "loss": 4.1878,
+      "step": 2032
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.388361045130642e-05,
+      "loss": 3.7775,
+      "step": 2033
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.387767220902614e-05,
+      "loss": 3.9315,
+      "step": 2034
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.387173396674584e-05,
+      "loss": 3.9996,
+      "step": 2035
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.386579572446556e-05,
+      "loss": 3.9253,
+      "step": 2036
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.385985748218527e-05,
+      "loss": 4.0344,
+      "step": 2037
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.385391923990499e-05,
+      "loss": 3.6907,
+      "step": 2038
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.384798099762471e-05,
+      "loss": 3.6758,
+      "step": 2039
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.384204275534442e-05,
+      "loss": 3.9639,
+      "step": 2040
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.383610451306414e-05,
+      "loss": 3.7063,
+      "step": 2041
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.383016627078384e-05,
+      "loss": 4.1121,
+      "step": 2042
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.382422802850357e-05,
+      "loss": 4.0519,
+      "step": 2043
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.381828978622329e-05,
+      "loss": 4.0443,
+      "step": 2044
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.3812351543943e-05,
+      "loss": 3.9396,
+      "step": 2045
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.380641330166271e-05,
+      "loss": 3.7892,
+      "step": 2046
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.380047505938243e-05,
+      "loss": 3.9769,
+      "step": 2047
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.379453681710214e-05,
+      "loss": 3.6581,
+      "step": 2048
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.378859857482186e-05,
+      "loss": 4.0338,
+      "step": 2049
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.378266033254157e-05,
+      "loss": 3.6637,
+      "step": 2050
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.377672209026129e-05,
+      "loss": 3.8152,
+      "step": 2051
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.377078384798101e-05,
+      "loss": 3.8157,
+      "step": 2052
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.376484560570071e-05,
+      "loss": 3.7723,
+      "step": 2053
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.375890736342043e-05,
+      "loss": 3.8129,
+      "step": 2054
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.375296912114014e-05,
+      "loss": 3.8637,
+      "step": 2055
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.374703087885986e-05,
+      "loss": 3.8721,
+      "step": 2056
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.374109263657957e-05,
+      "loss": 3.7036,
+      "step": 2057
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.373515439429929e-05,
+      "loss": 3.5971,
+      "step": 2058
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.372921615201901e-05,
+      "loss": 4.2934,
+      "step": 2059
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.372327790973873e-05,
+      "loss": 3.9779,
+      "step": 2060
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.371733966745844e-05,
+      "loss": 3.9323,
+      "step": 2061
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.371140142517814e-05,
+      "loss": 3.7754,
+      "step": 2062
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.370546318289786e-05,
+      "loss": 3.972,
+      "step": 2063
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.369952494061758e-05,
+      "loss": 4.0639,
+      "step": 2064
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.36935866983373e-05,
+      "loss": 3.8884,
+      "step": 2065
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.368764845605701e-05,
+      "loss": 3.5926,
+      "step": 2066
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.368171021377673e-05,
+      "loss": 3.9393,
+      "step": 2067
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.367577197149644e-05,
+      "loss": 3.7009,
+      "step": 2068
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.366983372921616e-05,
+      "loss": 3.7181,
+      "step": 2069
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.366389548693587e-05,
+      "loss": 3.4975,
+      "step": 2070
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.365795724465559e-05,
+      "loss": 3.6686,
+      "step": 2071
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.36520190023753e-05,
+      "loss": 3.7205,
+      "step": 2072
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.364608076009501e-05,
+      "loss": 3.6827,
+      "step": 2073
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.364014251781473e-05,
+      "loss": 3.6423,
+      "step": 2074
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.363420427553444e-05,
+      "loss": 3.9018,
+      "step": 2075
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.362826603325416e-05,
+      "loss": 3.7833,
+      "step": 2076
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.362232779097387e-05,
+      "loss": 3.8347,
+      "step": 2077
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.36163895486936e-05,
+      "loss": 3.7915,
+      "step": 2078
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.361045130641331e-05,
+      "loss": 3.4989,
+      "step": 2079
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.360451306413303e-05,
+      "loss": 4.3252,
+      "step": 2080
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.359857482185274e-05,
+      "loss": 3.5966,
+      "step": 2081
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.359263657957244e-05,
+      "loss": 3.6507,
+      "step": 2082
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.358669833729216e-05,
+      "loss": 4.1271,
+      "step": 2083
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.358076009501188e-05,
+      "loss": 4.0866,
+      "step": 2084
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.35748218527316e-05,
+      "loss": 4.0113,
+      "step": 2085
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.356888361045131e-05,
+      "loss": 3.9565,
+      "step": 2086
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.356294536817103e-05,
+      "loss": 4.0659,
+      "step": 2087
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.355700712589074e-05,
+      "loss": 3.8289,
+      "step": 2088
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.355106888361046e-05,
+      "loss": 3.5548,
+      "step": 2089
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.354513064133016e-05,
+      "loss": 3.6344,
+      "step": 2090
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.353919239904988e-05,
+      "loss": 3.8106,
+      "step": 2091
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.35332541567696e-05,
+      "loss": 3.6908,
+      "step": 2092
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.352731591448931e-05,
+      "loss": 4.1085,
+      "step": 2093
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.352137767220903e-05,
+      "loss": 4.1693,
+      "step": 2094
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.351543942992874e-05,
+      "loss": 3.9732,
+      "step": 2095
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.350950118764846e-05,
+      "loss": 3.8248,
+      "step": 2096
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.350356294536817e-05,
+      "loss": 4.0246,
+      "step": 2097
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.34976247030879e-05,
+      "loss": 3.9588,
+      "step": 2098
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.349168646080761e-05,
+      "loss": 3.7082,
+      "step": 2099
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.348574821852733e-05,
+      "loss": 3.9041,
+      "step": 2100
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.347980997624703e-05,
+      "loss": 3.9336,
+      "step": 2101
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.347387173396675e-05,
+      "loss": 3.8134,
+      "step": 2102
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.346793349168646e-05,
+      "loss": 3.8962,
+      "step": 2103
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.346199524940618e-05,
+      "loss": 3.7342,
+      "step": 2104
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.34560570071259e-05,
+      "loss": 3.7635,
+      "step": 2105
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.345011876484561e-05,
+      "loss": 3.8504,
+      "step": 2106
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.344418052256533e-05,
+      "loss": 3.6203,
+      "step": 2107
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.343824228028504e-05,
+      "loss": 3.9006,
+      "step": 2108
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.343230403800476e-05,
+      "loss": 3.8813,
+      "step": 2109
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.342636579572446e-05,
+      "loss": 3.9489,
+      "step": 2110
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.342042755344418e-05,
+      "loss": 3.8003,
+      "step": 2111
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.34144893111639e-05,
+      "loss": 3.9229,
+      "step": 2112
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.340855106888361e-05,
+      "loss": 3.7907,
+      "step": 2113
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.340261282660333e-05,
+      "loss": 4.0251,
+      "step": 2114
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.339667458432304e-05,
+      "loss": 3.9196,
+      "step": 2115
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.339073634204276e-05,
+      "loss": 3.7876,
+      "step": 2116
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.338479809976247e-05,
+      "loss": 3.5601,
+      "step": 2117
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.33788598574822e-05,
+      "loss": 3.6961,
+      "step": 2118
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.33729216152019e-05,
+      "loss": 3.9165,
+      "step": 2119
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.336698337292163e-05,
+      "loss": 4.116,
+      "step": 2120
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.336104513064133e-05,
+      "loss": 3.7858,
+      "step": 2121
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.335510688836105e-05,
+      "loss": 3.8643,
+      "step": 2122
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.334916864608076e-05,
+      "loss": 3.7828,
+      "step": 2123
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.334323040380048e-05,
+      "loss": 3.6895,
+      "step": 2124
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.33372921615202e-05,
+      "loss": 3.8249,
+      "step": 2125
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.333135391923991e-05,
+      "loss": 3.7706,
+      "step": 2126
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.332541567695963e-05,
+      "loss": 3.8187,
+      "step": 2127
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.331947743467933e-05,
+      "loss": 3.6167,
+      "step": 2128
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.331353919239906e-05,
+      "loss": 3.728,
+      "step": 2129
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.330760095011876e-05,
+      "loss": 3.3851,
+      "step": 2130
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.330166270783848e-05,
+      "loss": 4.0012,
+      "step": 2131
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.32957244655582e-05,
+      "loss": 3.7296,
+      "step": 2132
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.328978622327792e-05,
+      "loss": 3.7157,
+      "step": 2133
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.328384798099763e-05,
+      "loss": 3.9688,
+      "step": 2134
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.327790973871734e-05,
+      "loss": 3.9217,
+      "step": 2135
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.327197149643706e-05,
+      "loss": 3.9241,
+      "step": 2136
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.326603325415676e-05,
+      "loss": 3.8184,
+      "step": 2137
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.32600950118765e-05,
+      "loss": 3.7695,
+      "step": 2138
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.32541567695962e-05,
+      "loss": 3.7934,
+      "step": 2139
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.324821852731592e-05,
+      "loss": 3.8578,
+      "step": 2140
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.324228028503563e-05,
+      "loss": 4.1954,
+      "step": 2141
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.323634204275535e-05,
+      "loss": 4.1272,
+      "step": 2142
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.323040380047506e-05,
+      "loss": 3.7801,
+      "step": 2143
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.322446555819478e-05,
+      "loss": 3.7871,
+      "step": 2144
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.32185273159145e-05,
+      "loss": 3.8409,
+      "step": 2145
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.32125890736342e-05,
+      "loss": 4.1446,
+      "step": 2146
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.320665083135393e-05,
+      "loss": 4.0786,
+      "step": 2147
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.320071258907363e-05,
+      "loss": 3.9027,
+      "step": 2148
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.319477434679335e-05,
+      "loss": 3.8474,
+      "step": 2149
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.318883610451306e-05,
+      "loss": 3.8781,
+      "step": 2150
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.318289786223278e-05,
+      "loss": 3.8247,
+      "step": 2151
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.31769596199525e-05,
+      "loss": 4.0052,
+      "step": 2152
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.317102137767222e-05,
+      "loss": 3.7889,
+      "step": 2153
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.316508313539193e-05,
+      "loss": 3.7669,
+      "step": 2154
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.315914489311164e-05,
+      "loss": 4.1916,
+      "step": 2155
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.315320665083136e-05,
+      "loss": 3.7981,
+      "step": 2156
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.314726840855108e-05,
+      "loss": 3.8619,
+      "step": 2157
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.31413301662708e-05,
+      "loss": 3.7273,
+      "step": 2158
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.31353919239905e-05,
+      "loss": 4.0068,
+      "step": 2159
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.312945368171022e-05,
+      "loss": 3.8477,
+      "step": 2160
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.312351543942993e-05,
+      "loss": 3.9205,
+      "step": 2161
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.311757719714965e-05,
+      "loss": 3.5598,
+      "step": 2162
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.311163895486936e-05,
+      "loss": 3.8943,
+      "step": 2163
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.310570071258908e-05,
+      "loss": 3.8085,
+      "step": 2164
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.30997624703088e-05,
+      "loss": 3.854,
+      "step": 2165
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.30938242280285e-05,
+      "loss": 3.7438,
+      "step": 2166
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.308788598574823e-05,
+      "loss": 3.774,
+      "step": 2167
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.308194774346793e-05,
+      "loss": 3.7764,
+      "step": 2168
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.307600950118765e-05,
+      "loss": 4.1391,
+      "step": 2169
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.307007125890736e-05,
+      "loss": 3.9479,
+      "step": 2170
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.306413301662708e-05,
+      "loss": 3.7187,
+      "step": 2171
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.30581947743468e-05,
+      "loss": 3.82,
+      "step": 2172
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.305225653206652e-05,
+      "loss": 3.8284,
+      "step": 2173
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.304631828978623e-05,
+      "loss": 3.7026,
+      "step": 2174
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.304038004750593e-05,
+      "loss": 3.9347,
+      "step": 2175
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.303444180522565e-05,
+      "loss": 3.8637,
+      "step": 2176
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.302850356294537e-05,
+      "loss": 3.9824,
+      "step": 2177
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.30225653206651e-05,
+      "loss": 4.1117,
+      "step": 2178
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.30166270783848e-05,
+      "loss": 4.2317,
+      "step": 2179
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.301068883610452e-05,
+      "loss": 3.6549,
+      "step": 2180
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.300475059382423e-05,
+      "loss": 4.1646,
+      "step": 2181
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.299881235154395e-05,
+      "loss": 3.8887,
+      "step": 2182
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.299287410926366e-05,
+      "loss": 3.7116,
+      "step": 2183
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.298693586698338e-05,
+      "loss": 3.5156,
+      "step": 2184
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.29809976247031e-05,
+      "loss": 3.8922,
+      "step": 2185
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.29750593824228e-05,
+      "loss": 3.7892,
+      "step": 2186
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.296912114014252e-05,
+      "loss": 4.02,
+      "step": 2187
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.296318289786223e-05,
+      "loss": 3.8409,
+      "step": 2188
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.295724465558195e-05,
+      "loss": 3.72,
+      "step": 2189
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.295130641330166e-05,
+      "loss": 3.6814,
+      "step": 2190
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.294536817102139e-05,
+      "loss": 3.972,
+      "step": 2191
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.29394299287411e-05,
+      "loss": 3.9988,
+      "step": 2192
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.293349168646082e-05,
+      "loss": 4.1289,
+      "step": 2193
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.292755344418053e-05,
+      "loss": 4.2095,
+      "step": 2194
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.292161520190025e-05,
+      "loss": 4.053,
+      "step": 2195
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.291567695961995e-05,
+      "loss": 3.9777,
+      "step": 2196
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.290973871733967e-05,
+      "loss": 3.7623,
+      "step": 2197
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.29038004750594e-05,
+      "loss": 3.6487,
+      "step": 2198
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.28978622327791e-05,
+      "loss": 3.8182,
+      "step": 2199
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.289192399049882e-05,
+      "loss": 4.2794,
+      "step": 2200
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.288598574821853e-05,
+      "loss": 3.9003,
+      "step": 2201
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.288004750593825e-05,
+      "loss": 3.8206,
+      "step": 2202
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.287410926365795e-05,
+      "loss": 3.6953,
+      "step": 2203
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.286817102137768e-05,
+      "loss": 3.7684,
+      "step": 2204
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.28622327790974e-05,
+      "loss": 4.226,
+      "step": 2205
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.28562945368171e-05,
+      "loss": 3.9642,
+      "step": 2206
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.285035629453682e-05,
+      "loss": 3.6621,
+      "step": 2207
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.284441805225653e-05,
+      "loss": 3.8457,
+      "step": 2208
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.283847980997625e-05,
+      "loss": 3.7908,
+      "step": 2209
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.283254156769596e-05,
+      "loss": 4.022,
+      "step": 2210
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.282660332541569e-05,
+      "loss": 3.923,
+      "step": 2211
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.28206650831354e-05,
+      "loss": 3.7003,
+      "step": 2212
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.281472684085512e-05,
+      "loss": 3.7864,
+      "step": 2213
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.280878859857482e-05,
+      "loss": 3.9053,
+      "step": 2214
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.280285035629455e-05,
+      "loss": 4.0372,
+      "step": 2215
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.279691211401425e-05,
+      "loss": 3.7273,
+      "step": 2216
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.279097387173397e-05,
+      "loss": 3.7483,
+      "step": 2217
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.278503562945369e-05,
+      "loss": 3.8378,
+      "step": 2218
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.27790973871734e-05,
+      "loss": 3.6516,
+      "step": 2219
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.277315914489312e-05,
+      "loss": 4.1146,
+      "step": 2220
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.276722090261283e-05,
+      "loss": 3.7575,
+      "step": 2221
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.276128266033255e-05,
+      "loss": 3.6765,
+      "step": 2222
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.275534441805225e-05,
+      "loss": 4.1195,
+      "step": 2223
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.274940617577197e-05,
+      "loss": 3.668,
+      "step": 2224
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.27434679334917e-05,
+      "loss": 3.6837,
+      "step": 2225
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.273752969121141e-05,
+      "loss": 4.1511,
+      "step": 2226
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.273159144893112e-05,
+      "loss": 3.7772,
+      "step": 2227
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.272565320665083e-05,
+      "loss": 3.7475,
+      "step": 2228
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.271971496437055e-05,
+      "loss": 3.6411,
+      "step": 2229
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.271377672209026e-05,
+      "loss": 3.7971,
+      "step": 2230
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.270783847980999e-05,
+      "loss": 3.7916,
+      "step": 2231
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.27019002375297e-05,
+      "loss": 3.9424,
+      "step": 2232
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.269596199524942e-05,
+      "loss": 4.1777,
+      "step": 2233
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.269002375296912e-05,
+      "loss": 3.8871,
+      "step": 2234
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.268408551068884e-05,
+      "loss": 4.0519,
+      "step": 2235
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.267814726840855e-05,
+      "loss": 3.9878,
+      "step": 2236
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.267220902612827e-05,
+      "loss": 4.2697,
+      "step": 2237
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.266627078384799e-05,
+      "loss": 4.169,
+      "step": 2238
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.26603325415677e-05,
+      "loss": 3.7008,
+      "step": 2239
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.265439429928742e-05,
+      "loss": 3.7581,
+      "step": 2240
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.264845605700713e-05,
+      "loss": 4.0872,
+      "step": 2241
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.264251781472685e-05,
+      "loss": 3.857,
+      "step": 2242
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.263657957244655e-05,
+      "loss": 3.8972,
+      "step": 2243
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.263064133016627e-05,
+      "loss": 4.0778,
+      "step": 2244
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.262470308788599e-05,
+      "loss": 4.1687,
+      "step": 2245
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.261876484560571e-05,
+      "loss": 3.8644,
+      "step": 2246
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.261282660332542e-05,
+      "loss": 3.9273,
+      "step": 2247
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.260688836104513e-05,
+      "loss": 3.8316,
+      "step": 2248
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.260095011876485e-05,
+      "loss": 3.6608,
+      "step": 2249
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.259501187648455e-05,
+      "loss": 3.6638,
+      "step": 2250
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.258907363420429e-05,
+      "loss": 3.7746,
+      "step": 2251
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.2583135391924e-05,
+      "loss": 3.8458,
+      "step": 2252
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.257719714964372e-05,
+      "loss": 3.9136,
+      "step": 2253
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.257125890736342e-05,
+      "loss": 4.0477,
+      "step": 2254
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.256532066508314e-05,
+      "loss": 3.7786,
+      "step": 2255
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.255938242280285e-05,
+      "loss": 4.1063,
+      "step": 2256
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.255344418052257e-05,
+      "loss": 3.8734,
+      "step": 2257
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.254750593824229e-05,
+      "loss": 3.7594,
+      "step": 2258
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.2541567695962e-05,
+      "loss": 3.8608,
+      "step": 2259
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.253562945368172e-05,
+      "loss": 3.8261,
+      "step": 2260
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.252969121140142e-05,
+      "loss": 3.8043,
+      "step": 2261
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.252375296912114e-05,
+      "loss": 3.9649,
+      "step": 2262
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.251781472684085e-05,
+      "loss": 3.6677,
+      "step": 2263
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.251187648456057e-05,
+      "loss": 3.947,
+      "step": 2264
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.250593824228029e-05,
+      "loss": 3.8181,
+      "step": 2265
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.250000000000001e-05,
+      "loss": 3.7578,
+      "step": 2266
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.249406175771972e-05,
+      "loss": 4.0013,
+      "step": 2267
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.248812351543944e-05,
+      "loss": 3.8331,
+      "step": 2268
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.248218527315915e-05,
+      "loss": 3.9782,
+      "step": 2269
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.247624703087885e-05,
+      "loss": 3.8446,
+      "step": 2270
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.247030878859859e-05,
+      "loss": 4.032,
+      "step": 2271
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.24643705463183e-05,
+      "loss": 3.549,
+      "step": 2272
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.245843230403801e-05,
+      "loss": 3.9218,
+      "step": 2273
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.245249406175772e-05,
+      "loss": 3.7437,
+      "step": 2274
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.244655581947744e-05,
+      "loss": 3.7121,
+      "step": 2275
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.244061757719715e-05,
+      "loss": 3.6255,
+      "step": 2276
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.243467933491687e-05,
+      "loss": 3.8391,
+      "step": 2277
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.242874109263659e-05,
+      "loss": 3.7643,
+      "step": 2278
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.24228028503563e-05,
+      "loss": 4.232,
+      "step": 2279
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.241686460807602e-05,
+      "loss": 4.1758,
+      "step": 2280
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.241092636579572e-05,
+      "loss": 3.7279,
+      "step": 2281
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.240498812351544e-05,
+      "loss": 3.6979,
+      "step": 2282
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.239904988123515e-05,
+      "loss": 3.5919,
+      "step": 2283
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.239311163895487e-05,
+      "loss": 3.5123,
+      "step": 2284
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.238717339667459e-05,
+      "loss": 4.4183,
+      "step": 2285
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.238123515439431e-05,
+      "loss": 3.7386,
+      "step": 2286
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.237529691211402e-05,
+      "loss": 4.0425,
+      "step": 2287
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.236935866983374e-05,
+      "loss": 4.0513,
+      "step": 2288
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.236342042755344e-05,
+      "loss": 3.6861,
+      "step": 2289
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.235748218527317e-05,
+      "loss": 3.9303,
+      "step": 2290
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.235154394299289e-05,
+      "loss": 3.7783,
+      "step": 2291
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.234560570071259e-05,
+      "loss": 3.927,
+      "step": 2292
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.233966745843231e-05,
+      "loss": 4.0426,
+      "step": 2293
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.233372921615202e-05,
+      "loss": 3.9999,
+      "step": 2294
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.232779097387174e-05,
+      "loss": 3.9948,
+      "step": 2295
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.232185273159145e-05,
+      "loss": 3.7549,
+      "step": 2296
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.231591448931117e-05,
+      "loss": 4.3104,
+      "step": 2297
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.230997624703089e-05,
+      "loss": 3.9306,
+      "step": 2298
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.230403800475061e-05,
+      "loss": 3.7529,
+      "step": 2299
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.229809976247031e-05,
+      "loss": 3.8095,
+      "step": 2300
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.229216152019002e-05,
+      "loss": 3.9263,
+      "step": 2301
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.228622327790974e-05,
+      "loss": 3.9386,
+      "step": 2302
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.228028503562945e-05,
+      "loss": 3.9901,
+      "step": 2303
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.227434679334917e-05,
+      "loss": 3.9723,
+      "step": 2304
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.226840855106889e-05,
+      "loss": 3.9073,
+      "step": 2305
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.226247030878861e-05,
+      "loss": 3.6759,
+      "step": 2306
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.225653206650832e-05,
+      "loss": 3.7763,
+      "step": 2307
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.225059382422804e-05,
+      "loss": 3.9855,
+      "step": 2308
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.224465558194774e-05,
+      "loss": 3.8044,
+      "step": 2309
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.223871733966746e-05,
+      "loss": 3.7894,
+      "step": 2310
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.223277909738718e-05,
+      "loss": 3.5844,
+      "step": 2311
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.222684085510689e-05,
+      "loss": 4.1678,
+      "step": 2312
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.222090261282661e-05,
+      "loss": 3.6515,
+      "step": 2313
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.221496437054632e-05,
+      "loss": 3.7428,
+      "step": 2314
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.220902612826604e-05,
+      "loss": 4.1606,
+      "step": 2315
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.220308788598575e-05,
+      "loss": 3.5746,
+      "step": 2316
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.219714964370547e-05,
+      "loss": 3.7739,
+      "step": 2317
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.219121140142519e-05,
+      "loss": 3.7043,
+      "step": 2318
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.21852731591449e-05,
+      "loss": 4.0782,
+      "step": 2319
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.217933491686461e-05,
+      "loss": 3.7061,
+      "step": 2320
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.217339667458432e-05,
+      "loss": 3.9926,
+      "step": 2321
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.216745843230404e-05,
+      "loss": 3.8475,
+      "step": 2322
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.216152019002375e-05,
+      "loss": 3.793,
+      "step": 2323
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.215558194774348e-05,
+      "loss": 3.8377,
+      "step": 2324
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.214964370546319e-05,
+      "loss": 3.6739,
+      "step": 2325
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.214370546318291e-05,
+      "loss": 3.7777,
+      "step": 2326
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.213776722090262e-05,
+      "loss": 3.8204,
+      "step": 2327
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.213182897862234e-05,
+      "loss": 3.6186,
+      "step": 2328
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.212589073634204e-05,
+      "loss": 3.891,
+      "step": 2329
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.211995249406176e-05,
+      "loss": 3.7935,
+      "step": 2330
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.211401425178148e-05,
+      "loss": 3.6366,
+      "step": 2331
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.210807600950119e-05,
+      "loss": 3.601,
+      "step": 2332
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.210213776722091e-05,
+      "loss": 3.9553,
+      "step": 2333
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.209619952494062e-05,
+      "loss": 3.6276,
+      "step": 2334
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.209026128266034e-05,
+      "loss": 4.0258,
+      "step": 2335
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.208432304038004e-05,
+      "loss": 3.9752,
+      "step": 2336
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.207838479809976e-05,
+      "loss": 3.9162,
+      "step": 2337
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.207244655581948e-05,
+      "loss": 3.9886,
+      "step": 2338
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.20665083135392e-05,
+      "loss": 3.6902,
+      "step": 2339
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.206057007125891e-05,
+      "loss": 3.8212,
+      "step": 2340
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.205463182897862e-05,
+      "loss": 3.8886,
+      "step": 2341
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.204869358669834e-05,
+      "loss": 3.8906,
+      "step": 2342
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.204275534441805e-05,
+      "loss": 3.8474,
+      "step": 2343
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.203681710213778e-05,
+      "loss": 3.9055,
+      "step": 2344
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.203087885985749e-05,
+      "loss": 3.8376,
+      "step": 2345
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.202494061757721e-05,
+      "loss": 3.8913,
+      "step": 2346
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.201900237529691e-05,
+      "loss": 3.9372,
+      "step": 2347
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.201306413301663e-05,
+      "loss": 3.8845,
+      "step": 2348
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.200712589073634e-05,
+      "loss": 3.9151,
+      "step": 2349
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.200118764845606e-05,
+      "loss": 3.9417,
+      "step": 2350
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.199524940617578e-05,
+      "loss": 3.8267,
+      "step": 2351
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.198931116389549e-05,
+      "loss": 3.7264,
+      "step": 2352
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.198337292161521e-05,
+      "loss": 3.9971,
+      "step": 2353
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.197743467933492e-05,
+      "loss": 3.7816,
+      "step": 2354
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.197149643705464e-05,
+      "loss": 3.6185,
+      "step": 2355
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.196555819477434e-05,
+      "loss": 3.9907,
+      "step": 2356
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.195961995249406e-05,
+      "loss": 3.802,
+      "step": 2357
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.195368171021378e-05,
+      "loss": 3.7336,
+      "step": 2358
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.19477434679335e-05,
+      "loss": 3.9396,
+      "step": 2359
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.194180522565321e-05,
+      "loss": 3.7117,
+      "step": 2360
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.193586698337293e-05,
+      "loss": 3.8743,
+      "step": 2361
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.192992874109264e-05,
+      "loss": 3.8864,
+      "step": 2362
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.192399049881234e-05,
+      "loss": 3.9532,
+      "step": 2363
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.191805225653208e-05,
+      "loss": 3.8274,
+      "step": 2364
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.191211401425179e-05,
+      "loss": 3.7667,
+      "step": 2365
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.19061757719715e-05,
+      "loss": 3.8852,
+      "step": 2366
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.190023752969121e-05,
+      "loss": 3.7237,
+      "step": 2367
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.189429928741093e-05,
+      "loss": 3.8833,
+      "step": 2368
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.188836104513064e-05,
+      "loss": 3.7736,
+      "step": 2369
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.188242280285036e-05,
+      "loss": 4.1424,
+      "step": 2370
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.187648456057008e-05,
+      "loss": 3.7474,
+      "step": 2371
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.187054631828979e-05,
+      "loss": 3.811,
+      "step": 2372
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.186460807600951e-05,
+      "loss": 3.9466,
+      "step": 2373
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.185866983372921e-05,
+      "loss": 3.6972,
+      "step": 2374
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.185273159144893e-05,
+      "loss": 3.7348,
+      "step": 2375
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.184679334916864e-05,
+      "loss": 3.766,
+      "step": 2376
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.184085510688836e-05,
+      "loss": 3.7628,
+      "step": 2377
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.183491686460808e-05,
+      "loss": 4.1485,
+      "step": 2378
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.18289786223278e-05,
+      "loss": 3.7915,
+      "step": 2379
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.182304038004751e-05,
+      "loss": 3.5942,
+      "step": 2380
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.181710213776723e-05,
+      "loss": 3.7905,
+      "step": 2381
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.181116389548694e-05,
+      "loss": 3.5964,
+      "step": 2382
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.180522565320664e-05,
+      "loss": 4.1464,
+      "step": 2383
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.179928741092638e-05,
+      "loss": 3.4826,
+      "step": 2384
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.179334916864608e-05,
+      "loss": 4.0832,
+      "step": 2385
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.17874109263658e-05,
+      "loss": 4.0264,
+      "step": 2386
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.178147268408551e-05,
+      "loss": 3.9645,
+      "step": 2387
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.177553444180523e-05,
+      "loss": 3.8456,
+      "step": 2388
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.176959619952494e-05,
+      "loss": 3.8594,
+      "step": 2389
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.176365795724466e-05,
+      "loss": 4.008,
+      "step": 2390
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.175771971496438e-05,
+      "loss": 4.0024,
+      "step": 2391
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.17517814726841e-05,
+      "loss": 3.913,
+      "step": 2392
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.17458432304038e-05,
+      "loss": 4.0062,
+      "step": 2393
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.173990498812351e-05,
+      "loss": 3.8194,
+      "step": 2394
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.173396674584323e-05,
+      "loss": 3.7402,
+      "step": 2395
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.172802850356294e-05,
+      "loss": 3.8941,
+      "step": 2396
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.172209026128266e-05,
+      "loss": 3.6733,
+      "step": 2397
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.171615201900238e-05,
+      "loss": 3.9813,
+      "step": 2398
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.17102137767221e-05,
+      "loss": 3.8457,
+      "step": 2399
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.170427553444181e-05,
+      "loss": 3.7724,
+      "step": 2400
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.169833729216153e-05,
+      "loss": 3.8552,
+      "step": 2401
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.169239904988124e-05,
+      "loss": 3.7204,
+      "step": 2402
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.168646080760096e-05,
+      "loss": 4.068,
+      "step": 2403
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.168052256532068e-05,
+      "loss": 3.6425,
+      "step": 2404
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.167458432304038e-05,
+      "loss": 3.7741,
+      "step": 2405
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.16686460807601e-05,
+      "loss": 3.8787,
+      "step": 2406
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.166270783847981e-05,
+      "loss": 3.8472,
+      "step": 2407
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.165676959619953e-05,
+      "loss": 3.8279,
+      "step": 2408
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.165083135391924e-05,
+      "loss": 3.7915,
+      "step": 2409
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.164489311163896e-05,
+      "loss": 3.6768,
+      "step": 2410
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.163895486935868e-05,
+      "loss": 4.1882,
+      "step": 2411
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.16330166270784e-05,
+      "loss": 3.6928,
+      "step": 2412
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.16270783847981e-05,
+      "loss": 3.6647,
+      "step": 2413
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.162114014251781e-05,
+      "loss": 4.0149,
+      "step": 2414
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.161520190023753e-05,
+      "loss": 4.2171,
+      "step": 2415
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.160926365795724e-05,
+      "loss": 3.9095,
+      "step": 2416
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.160332541567696e-05,
+      "loss": 3.7637,
+      "step": 2417
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.159738717339668e-05,
+      "loss": 4.1305,
+      "step": 2418
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.15914489311164e-05,
+      "loss": 3.5557,
+      "step": 2419
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.158551068883611e-05,
+      "loss": 4.0735,
+      "step": 2420
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.157957244655583e-05,
+      "loss": 3.8926,
+      "step": 2421
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.157363420427553e-05,
+      "loss": 3.7705,
+      "step": 2422
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.156769596199525e-05,
+      "loss": 4.0214,
+      "step": 2423
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.156175771971497e-05,
+      "loss": 3.7592,
+      "step": 2424
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.155581947743468e-05,
+      "loss": 3.9398,
+      "step": 2425
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.15498812351544e-05,
+      "loss": 3.775,
+      "step": 2426
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.154394299287411e-05,
+      "loss": 3.8757,
+      "step": 2427
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.153800475059383e-05,
+      "loss": 3.8285,
+      "step": 2428
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.153206650831354e-05,
+      "loss": 4.1164,
+      "step": 2429
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.152612826603326e-05,
+      "loss": 3.6547,
+      "step": 2430
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.152019002375298e-05,
+      "loss": 3.9094,
+      "step": 2431
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.15142517814727e-05,
+      "loss": 3.8391,
+      "step": 2432
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.15083135391924e-05,
+      "loss": 3.574,
+      "step": 2433
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.150237529691211e-05,
+      "loss": 3.4054,
+      "step": 2434
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.149643705463183e-05,
+      "loss": 3.9443,
+      "step": 2435
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.149049881235154e-05,
+      "loss": 3.7127,
+      "step": 2436
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.148456057007126e-05,
+      "loss": 3.9031,
+      "step": 2437
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.147862232779098e-05,
+      "loss": 4.1991,
+      "step": 2438
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.14726840855107e-05,
+      "loss": 3.7434,
+      "step": 2439
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.14667458432304e-05,
+      "loss": 3.9538,
+      "step": 2440
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.146080760095013e-05,
+      "loss": 4.1466,
+      "step": 2441
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.145486935866983e-05,
+      "loss": 3.7233,
+      "step": 2442
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.144893111638955e-05,
+      "loss": 3.8744,
+      "step": 2443
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.144299287410927e-05,
+      "loss": 3.9121,
+      "step": 2444
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.143705463182898e-05,
+      "loss": 3.9266,
+      "step": 2445
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.14311163895487e-05,
+      "loss": 3.8944,
+      "step": 2446
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.142517814726841e-05,
+      "loss": 3.8807,
+      "step": 2447
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.141923990498813e-05,
+      "loss": 3.9241,
+      "step": 2448
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.141330166270783e-05,
+      "loss": 3.8425,
+      "step": 2449
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.140736342042756e-05,
+      "loss": 3.9082,
+      "step": 2450
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.140142517814728e-05,
+      "loss": 3.7674,
+      "step": 2451
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.1395486935867e-05,
+      "loss": 4.1111,
+      "step": 2452
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.13895486935867e-05,
+      "loss": 3.8155,
+      "step": 2453
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.138361045130642e-05,
+      "loss": 4.0009,
+      "step": 2454
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.137767220902613e-05,
+      "loss": 3.7724,
+      "step": 2455
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.137173396674584e-05,
+      "loss": 3.8072,
+      "step": 2456
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.136579572446557e-05,
+      "loss": 4.2404,
+      "step": 2457
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.135985748218528e-05,
+      "loss": 3.8499,
+      "step": 2458
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.1353919239905e-05,
+      "loss": 3.684,
+      "step": 2459
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.13479809976247e-05,
+      "loss": 3.7955,
+      "step": 2460
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.134204275534442e-05,
+      "loss": 4.03,
+      "step": 2461
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.133610451306413e-05,
+      "loss": 3.8098,
+      "step": 2462
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.133016627078385e-05,
+      "loss": 3.7233,
+      "step": 2463
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.132422802850357e-05,
+      "loss": 3.9442,
+      "step": 2464
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.131828978622329e-05,
+      "loss": 3.9118,
+      "step": 2465
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.1312351543943e-05,
+      "loss": 3.8458,
+      "step": 2466
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.13064133016627e-05,
+      "loss": 3.6388,
+      "step": 2467
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.130047505938243e-05,
+      "loss": 4.064,
+      "step": 2468
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.129453681710213e-05,
+      "loss": 3.7515,
+      "step": 2469
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.128859857482185e-05,
+      "loss": 3.6558,
+      "step": 2470
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.128266033254157e-05,
+      "loss": 3.9568,
+      "step": 2471
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.12767220902613e-05,
+      "loss": 3.7134,
+      "step": 2472
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.1270783847981e-05,
+      "loss": 4.0063,
+      "step": 2473
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.126484560570072e-05,
+      "loss": 3.709,
+      "step": 2474
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.125890736342043e-05,
+      "loss": 3.725,
+      "step": 2475
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.125296912114014e-05,
+      "loss": 3.9869,
+      "step": 2476
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.124703087885987e-05,
+      "loss": 3.75,
+      "step": 2477
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.124109263657958e-05,
+      "loss": 3.9015,
+      "step": 2478
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.12351543942993e-05,
+      "loss": 3.7263,
+      "step": 2479
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.1229216152019e-05,
+      "loss": 4.0147,
+      "step": 2480
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.122327790973872e-05,
+      "loss": 3.4561,
+      "step": 2481
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.121733966745843e-05,
+      "loss": 3.8066,
+      "step": 2482
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.121140142517815e-05,
+      "loss": 3.6465,
+      "step": 2483
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.120546318289787e-05,
+      "loss": 3.6034,
+      "step": 2484
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.119952494061759e-05,
+      "loss": 3.7692,
+      "step": 2485
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.11935866983373e-05,
+      "loss": 4.8459,
+      "step": 2486
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.1187648456057e-05,
+      "loss": 3.9671,
+      "step": 2487
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.118171021377673e-05,
+      "loss": 4.0852,
+      "step": 2488
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.117577197149643e-05,
+      "loss": 4.027,
+      "step": 2489
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.116983372921615e-05,
+      "loss": 3.8659,
+      "step": 2490
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.116389548693587e-05,
+      "loss": 3.7706,
+      "step": 2491
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.115795724465559e-05,
+      "loss": 3.9211,
+      "step": 2492
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.11520190023753e-05,
+      "loss": 3.7648,
+      "step": 2493
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.114608076009502e-05,
+      "loss": 3.9413,
+      "step": 2494
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.114014251781473e-05,
+      "loss": 3.672,
+      "step": 2495
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.113420427553445e-05,
+      "loss": 3.8431,
+      "step": 2496
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.112826603325417e-05,
+      "loss": 4.0186,
+      "step": 2497
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.112232779097387e-05,
+      "loss": 3.7599,
+      "step": 2498
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.11163895486936e-05,
+      "loss": 3.6585,
+      "step": 2499
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.11104513064133e-05,
+      "loss": 3.5506,
+      "step": 2500
+    },
+    {
+      "epoch": 2.8,
+      "eval_loss": 4.492813587188721,
+      "eval_runtime": 899.5193,
+      "eval_samples_per_second": 2.937,
+      "eval_steps_per_second": 0.368,
+      "eval_wer": 1.8771717572391908,
+      "step": 2500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.110451306413302e-05,
+      "loss": 4.4715,
+      "step": 2501
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.109857482185273e-05,
+      "loss": 3.7647,
+      "step": 2502
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.109263657957245e-05,
+      "loss": 3.6216,
+      "step": 2503
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.108669833729217e-05,
+      "loss": 4.0743,
+      "step": 2504
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.108076009501189e-05,
+      "loss": 3.7252,
+      "step": 2505
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.10748218527316e-05,
+      "loss": 4.018,
+      "step": 2506
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.10688836104513e-05,
+      "loss": 3.7672,
+      "step": 2507
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.106294536817102e-05,
+      "loss": 4.0627,
+      "step": 2508
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.105700712589073e-05,
+      "loss": 3.8853,
+      "step": 2509
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.105106888361045e-05,
+      "loss": 4.0378,
+      "step": 2510
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.104513064133017e-05,
+      "loss": 3.9224,
+      "step": 2511
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.103919239904989e-05,
+      "loss": 3.6413,
+      "step": 2512
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.10332541567696e-05,
+      "loss": 3.9063,
+      "step": 2513
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.102731591448932e-05,
+      "loss": 3.828,
+      "step": 2514
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.102137767220903e-05,
+      "loss": 3.9463,
+      "step": 2515
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.101543942992875e-05,
+      "loss": 4.2303,
+      "step": 2516
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.100950118764847e-05,
+      "loss": 3.9163,
+      "step": 2517
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.100356294536817e-05,
+      "loss": 3.7456,
+      "step": 2518
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.09976247030879e-05,
+      "loss": 3.7836,
+      "step": 2519
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.09916864608076e-05,
+      "loss": 3.943,
+      "step": 2520
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.098574821852732e-05,
+      "loss": 4.1882,
+      "step": 2521
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.097980997624703e-05,
+      "loss": 3.9216,
+      "step": 2522
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.097387173396675e-05,
+      "loss": 4.079,
+      "step": 2523
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.096793349168647e-05,
+      "loss": 3.928,
+      "step": 2524
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.096199524940619e-05,
+      "loss": 3.6287,
+      "step": 2525
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.09560570071259e-05,
+      "loss": 3.9699,
+      "step": 2526
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.095011876484562e-05,
+      "loss": 3.9758,
+      "step": 2527
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.094418052256532e-05,
+      "loss": 3.8028,
+      "step": 2528
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.093824228028503e-05,
+      "loss": 3.9568,
+      "step": 2529
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.093230403800475e-05,
+      "loss": 3.7256,
+      "step": 2530
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.092636579572447e-05,
+      "loss": 3.6867,
+      "step": 2531
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.092042755344419e-05,
+      "loss": 3.8429,
+      "step": 2532
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.09144893111639e-05,
+      "loss": 3.3626,
+      "step": 2533
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.090855106888362e-05,
+      "loss": 3.6205,
+      "step": 2534
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.090261282660332e-05,
+      "loss": 3.8452,
+      "step": 2535
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.089667458432304e-05,
+      "loss": 4.1429,
+      "step": 2536
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.089073634204277e-05,
+      "loss": 3.8285,
+      "step": 2537
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.088479809976247e-05,
+      "loss": 3.8738,
+      "step": 2538
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.087885985748219e-05,
+      "loss": 3.997,
+      "step": 2539
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.08729216152019e-05,
+      "loss": 3.6897,
+      "step": 2540
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.086698337292162e-05,
+      "loss": 3.9334,
+      "step": 2541
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.086104513064133e-05,
+      "loss": 3.7478,
+      "step": 2542
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.085510688836105e-05,
+      "loss": 3.8858,
+      "step": 2543
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.084916864608077e-05,
+      "loss": 3.9292,
+      "step": 2544
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.084323040380049e-05,
+      "loss": 4.0312,
+      "step": 2545
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.08372921615202e-05,
+      "loss": 3.8808,
+      "step": 2546
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.083135391923991e-05,
+      "loss": 3.8943,
+      "step": 2547
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.082541567695962e-05,
+      "loss": 3.6872,
+      "step": 2548
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.081947743467933e-05,
+      "loss": 3.8353,
+      "step": 2549
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.081353919239905e-05,
+      "loss": 3.8598,
+      "step": 2550
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.080760095011877e-05,
+      "loss": 3.7231,
+      "step": 2551
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.080166270783849e-05,
+      "loss": 3.8049,
+      "step": 2552
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.07957244655582e-05,
+      "loss": 3.6938,
+      "step": 2553
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.078978622327792e-05,
+      "loss": 3.816,
+      "step": 2554
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.078384798099762e-05,
+      "loss": 3.6118,
+      "step": 2555
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.077790973871734e-05,
+      "loss": 3.7146,
+      "step": 2556
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.077197149643706e-05,
+      "loss": 4.0785,
+      "step": 2557
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.076603325415678e-05,
+      "loss": 3.7163,
+      "step": 2558
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.076009501187649e-05,
+      "loss": 3.8356,
+      "step": 2559
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.07541567695962e-05,
+      "loss": 3.7709,
+      "step": 2560
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.074821852731592e-05,
+      "loss": 3.6377,
+      "step": 2561
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.074228028503563e-05,
+      "loss": 3.6401,
+      "step": 2562
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.073634204275535e-05,
+      "loss": 4.0615,
+      "step": 2563
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.073040380047507e-05,
+      "loss": 3.742,
+      "step": 2564
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.072446555819479e-05,
+      "loss": 3.7632,
+      "step": 2565
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.071852731591449e-05,
+      "loss": 3.749,
+      "step": 2566
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.071258907363421e-05,
+      "loss": 3.5402,
+      "step": 2567
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.070665083135392e-05,
+      "loss": 3.8125,
+      "step": 2568
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.070071258907363e-05,
+      "loss": 3.8695,
+      "step": 2569
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.069477434679336e-05,
+      "loss": 3.8474,
+      "step": 2570
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.068883610451307e-05,
+      "loss": 3.5647,
+      "step": 2571
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.068289786223279e-05,
+      "loss": 3.6698,
+      "step": 2572
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.06769596199525e-05,
+      "loss": 3.7421,
+      "step": 2573
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.067102137767222e-05,
+      "loss": 3.8698,
+      "step": 2574
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.066508313539192e-05,
+      "loss": 3.6925,
+      "step": 2575
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.065914489311164e-05,
+      "loss": 3.6451,
+      "step": 2576
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.065320665083136e-05,
+      "loss": 4.1154,
+      "step": 2577
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.064726840855108e-05,
+      "loss": 3.8532,
+      "step": 2578
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.064133016627079e-05,
+      "loss": 3.4555,
+      "step": 2579
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.06353919239905e-05,
+      "loss": 4.0216,
+      "step": 2580
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.062945368171022e-05,
+      "loss": 3.9705,
+      "step": 2581
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.062351543942992e-05,
+      "loss": 3.8851,
+      "step": 2582
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.061757719714964e-05,
+      "loss": 3.4314,
+      "step": 2583
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.061163895486936e-05,
+      "loss": 3.1727,
+      "step": 2584
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.060570071258908e-05,
+      "loss": 3.9149,
+      "step": 2585
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.059976247030879e-05,
+      "loss": 3.7971,
+      "step": 2586
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.059382422802851e-05,
+      "loss": 3.7516,
+      "step": 2587
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.058788598574822e-05,
+      "loss": 3.8903,
+      "step": 2588
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.058194774346794e-05,
+      "loss": 4.0466,
+      "step": 2589
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.057600950118766e-05,
+      "loss": 3.7303,
+      "step": 2590
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.057007125890737e-05,
+      "loss": 4.0373,
+      "step": 2591
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.056413301662709e-05,
+      "loss": 3.945,
+      "step": 2592
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.05581947743468e-05,
+      "loss": 3.986,
+      "step": 2593
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.055225653206651e-05,
+      "loss": 4.1228,
+      "step": 2594
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.054631828978622e-05,
+      "loss": 3.9339,
+      "step": 2595
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.054038004750594e-05,
+      "loss": 3.8114,
+      "step": 2596
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.053444180522566e-05,
+      "loss": 3.9459,
+      "step": 2597
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.052850356294538e-05,
+      "loss": 3.8068,
+      "step": 2598
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.052256532066509e-05,
+      "loss": 3.6713,
+      "step": 2599
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.05166270783848e-05,
+      "loss": 3.6288,
+      "step": 2600
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.051068883610452e-05,
+      "loss": 3.7698,
+      "step": 2601
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.050475059382422e-05,
+      "loss": 3.8335,
+      "step": 2602
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.049881235154394e-05,
+      "loss": 3.8141,
+      "step": 2603
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.049287410926366e-05,
+      "loss": 3.787,
+      "step": 2604
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.048693586698338e-05,
+      "loss": 3.7091,
+      "step": 2605
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.048099762470309e-05,
+      "loss": 3.7576,
+      "step": 2606
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.047505938242281e-05,
+      "loss": 3.8146,
+      "step": 2607
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.046912114014252e-05,
+      "loss": 3.665,
+      "step": 2608
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.046318289786224e-05,
+      "loss": 3.9334,
+      "step": 2609
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.045724465558196e-05,
+      "loss": 3.9812,
+      "step": 2610
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.045130641330167e-05,
+      "loss": 3.7028,
+      "step": 2611
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.044536817102139e-05,
+      "loss": 3.5496,
+      "step": 2612
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.043942992874109e-05,
+      "loss": 3.8854,
+      "step": 2613
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.043349168646081e-05,
+      "loss": 3.536,
+      "step": 2614
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.042755344418052e-05,
+      "loss": 3.8106,
+      "step": 2615
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.042161520190024e-05,
+      "loss": 4.0601,
+      "step": 2616
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.041567695961996e-05,
+      "loss": 3.6072,
+      "step": 2617
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.040973871733968e-05,
+      "loss": 3.3227,
+      "step": 2618
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.040380047505939e-05,
+      "loss": 3.8911,
+      "step": 2619
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.039786223277911e-05,
+      "loss": 3.7211,
+      "step": 2620
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.039192399049881e-05,
+      "loss": 3.7726,
+      "step": 2621
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.038598574821852e-05,
+      "loss": 3.6226,
+      "step": 2622
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.038004750593824e-05,
+      "loss": 3.9119,
+      "step": 2623
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.037410926365796e-05,
+      "loss": 4.1018,
+      "step": 2624
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.036817102137768e-05,
+      "loss": 3.624,
+      "step": 2625
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.036223277909739e-05,
+      "loss": 3.6477,
+      "step": 2626
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.035629453681711e-05,
+      "loss": 3.5309,
+      "step": 2627
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.035035629453682e-05,
+      "loss": 3.3045,
+      "step": 2628
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.034441805225654e-05,
+      "loss": 3.7028,
+      "step": 2629
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.033847980997626e-05,
+      "loss": 3.8234,
+      "step": 2630
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.033254156769598e-05,
+      "loss": 3.453,
+      "step": 2631
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.032660332541568e-05,
+      "loss": 3.4376,
+      "step": 2632
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.032066508313539e-05,
+      "loss": 3.4629,
+      "step": 2633
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.031472684085511e-05,
+      "loss": 3.4059,
+      "step": 2634
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.030878859857482e-05,
+      "loss": 3.9744,
+      "step": 2635
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.030285035629454e-05,
+      "loss": 3.7448,
+      "step": 2636
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.029691211401426e-05,
+      "loss": 3.6864,
+      "step": 2637
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.029097387173398e-05,
+      "loss": 3.6703,
+      "step": 2638
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.028503562945369e-05,
+      "loss": 4.1994,
+      "step": 2639
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.02790973871734e-05,
+      "loss": 3.7079,
+      "step": 2640
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.027315914489311e-05,
+      "loss": 3.7417,
+      "step": 2641
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.026722090261282e-05,
+      "loss": 4.1451,
+      "step": 2642
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.026128266033254e-05,
+      "loss": 3.9297,
+      "step": 2643
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.025534441805226e-05,
+      "loss": 3.6047,
+      "step": 2644
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.024940617577198e-05,
+      "loss": 3.7906,
+      "step": 2645
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.024346793349169e-05,
+      "loss": 3.7283,
+      "step": 2646
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.023752969121141e-05,
+      "loss": 3.8446,
+      "step": 2647
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.023159144893112e-05,
+      "loss": 3.8317,
+      "step": 2648
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.022565320665084e-05,
+      "loss": 3.8977,
+      "step": 2649
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.021971496437056e-05,
+      "loss": 3.6951,
+      "step": 2650
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.021377672209028e-05,
+      "loss": 3.8038,
+      "step": 2651
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.020783847980998e-05,
+      "loss": 3.6661,
+      "step": 2652
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.020190023752969e-05,
+      "loss": 3.7529,
+      "step": 2653
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.019596199524941e-05,
+      "loss": 3.6221,
+      "step": 2654
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.019002375296912e-05,
+      "loss": 3.6721,
+      "step": 2655
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.018408551068884e-05,
+      "loss": 3.9018,
+      "step": 2656
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.017814726840856e-05,
+      "loss": 3.9209,
+      "step": 2657
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.017220902612828e-05,
+      "loss": 3.844,
+      "step": 2658
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.016627078384798e-05,
+      "loss": 3.5487,
+      "step": 2659
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.01603325415677e-05,
+      "loss": 4.0318,
+      "step": 2660
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.015439429928741e-05,
+      "loss": 3.8398,
+      "step": 2661
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.014845605700713e-05,
+      "loss": 3.9736,
+      "step": 2662
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.014251781472684e-05,
+      "loss": 3.7691,
+      "step": 2663
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.013657957244656e-05,
+      "loss": 3.9498,
+      "step": 2664
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.013064133016628e-05,
+      "loss": 3.477,
+      "step": 2665
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.012470308788599e-05,
+      "loss": 3.7467,
+      "step": 2666
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.011876484560571e-05,
+      "loss": 4.1127,
+      "step": 2667
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.011282660332541e-05,
+      "loss": 3.7861,
+      "step": 2668
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.010688836104513e-05,
+      "loss": 3.6245,
+      "step": 2669
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.010095011876485e-05,
+      "loss": 3.9032,
+      "step": 2670
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.009501187648457e-05,
+      "loss": 3.5986,
+      "step": 2671
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.008907363420428e-05,
+      "loss": 3.6922,
+      "step": 2672
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.008313539192399e-05,
+      "loss": 4.1977,
+      "step": 2673
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.007719714964371e-05,
+      "loss": 3.1606,
+      "step": 2674
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.007125890736342e-05,
+      "loss": 3.3739,
+      "step": 2675
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.006532066508314e-05,
+      "loss": 3.0675,
+      "step": 2676
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.005938242280286e-05,
+      "loss": 3.4861,
+      "step": 2677
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.005344418052258e-05,
+      "loss": 3.435,
+      "step": 2678
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.004750593824228e-05,
+      "loss": 3.7157,
+      "step": 2679
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.0041567695962e-05,
+      "loss": 3.571,
+      "step": 2680
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 9.003562945368171e-05,
+      "loss": 3.5093,
+      "step": 2681
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 9.002969121140143e-05,
+      "loss": 3.4079,
+      "step": 2682
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 9.002375296912114e-05,
+      "loss": 3.4635,
+      "step": 2683
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 9.001781472684086e-05,
+      "loss": 3.4903,
+      "step": 2684
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 9.001187648456058e-05,
+      "loss": 3.6592,
+      "step": 2685
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 9.000593824228029e-05,
+      "loss": 3.2572,
+      "step": 2686
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 9e-05,
+      "loss": 3.5961,
+      "step": 2687
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 8.999406175771971e-05,
+      "loss": 3.3546,
+      "step": 2688
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 8.998812351543943e-05,
+      "loss": 3.3584,
+      "step": 2689
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.998218527315915e-05,
+      "loss": 3.4062,
+      "step": 2690
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.997624703087887e-05,
+      "loss": 3.4339,
+      "step": 2691
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.997030878859858e-05,
+      "loss": 3.6216,
+      "step": 2692
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.99643705463183e-05,
+      "loss": 3.4937,
+      "step": 2693
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.995843230403801e-05,
+      "loss": 3.4999,
+      "step": 2694
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.995249406175771e-05,
+      "loss": 3.4452,
+      "step": 2695
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.994655581947743e-05,
+      "loss": 3.3459,
+      "step": 2696
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.994061757719716e-05,
+      "loss": 3.403,
+      "step": 2697
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.993467933491688e-05,
+      "loss": 3.4013,
+      "step": 2698
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.992874109263658e-05,
+      "loss": 3.5615,
+      "step": 2699
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.99228028503563e-05,
+      "loss": 3.3959,
+      "step": 2700
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.991686460807601e-05,
+      "loss": 3.5614,
+      "step": 2701
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.991092636579573e-05,
+      "loss": 3.4597,
+      "step": 2702
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.990498812351545e-05,
+      "loss": 3.466,
+      "step": 2703
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.989904988123516e-05,
+      "loss": 3.2678,
+      "step": 2704
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.989311163895488e-05,
+      "loss": 3.7314,
+      "step": 2705
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.988717339667458e-05,
+      "loss": 3.3357,
+      "step": 2706
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.98812351543943e-05,
+      "loss": 3.3506,
+      "step": 2707
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 8.987529691211401e-05,
+      "loss": 3.1728,
+      "step": 2708
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 8.986935866983373e-05,
+      "loss": 3.2948,
+      "step": 2709
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 8.986342042755345e-05,
+      "loss": 3.1334,
+      "step": 2710
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 8.985748218527317e-05,
+      "loss": 3.5172,
+      "step": 2711
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 8.985154394299288e-05,
+      "loss": 3.2604,
+      "step": 2712
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 8.98456057007126e-05,
+      "loss": 3.3369,
+      "step": 2713
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 8.98396674584323e-05,
+      "loss": 3.4115,
+      "step": 2714
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 8.983372921615201e-05,
+      "loss": 3.1878,
+      "step": 2715
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 8.982779097387173e-05,
+      "loss": 3.2363,
+      "step": 2716
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.982185273159145e-05,
+      "loss": 3.1557,
+      "step": 2717
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.981591448931117e-05,
+      "loss": 3.2771,
+      "step": 2718
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.980997624703088e-05,
+      "loss": 3.3616,
+      "step": 2719
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.98040380047506e-05,
+      "loss": 3.4235,
+      "step": 2720
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.979809976247031e-05,
+      "loss": 3.0862,
+      "step": 2721
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.979216152019003e-05,
+      "loss": 3.1692,
+      "step": 2722
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.978622327790975e-05,
+      "loss": 3.2702,
+      "step": 2723
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.978028503562947e-05,
+      "loss": 3.4664,
+      "step": 2724
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.977434679334918e-05,
+      "loss": 3.0959,
+      "step": 2725
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 8.976840855106888e-05,
+      "loss": 2.6833,
+      "step": 2726
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 8.97624703087886e-05,
+      "loss": 3.5797,
+      "step": 2727
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 8.975653206650831e-05,
+      "loss": 3.4362,
+      "step": 2728
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 8.975059382422803e-05,
+      "loss": 3.3635,
+      "step": 2729
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 8.974465558194775e-05,
+      "loss": 3.252,
+      "step": 2730
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 8.973871733966747e-05,
+      "loss": 3.3736,
+      "step": 2731
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 8.973277909738718e-05,
+      "loss": 3.5185,
+      "step": 2732
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 8.97268408551069e-05,
+      "loss": 3.6019,
+      "step": 2733
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.97209026128266e-05,
+      "loss": 3.2806,
+      "step": 2734
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.971496437054631e-05,
+      "loss": 3.3971,
+      "step": 2735
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.970902612826603e-05,
+      "loss": 3.5337,
+      "step": 2736
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.970308788598575e-05,
+      "loss": 3.2895,
+      "step": 2737
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.969714964370547e-05,
+      "loss": 3.4222,
+      "step": 2738
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.969121140142518e-05,
+      "loss": 3.3529,
+      "step": 2739
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.96852731591449e-05,
+      "loss": 3.3541,
+      "step": 2740
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.967933491686461e-05,
+      "loss": 3.5352,
+      "step": 2741
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 8.967339667458433e-05,
+      "loss": 3.586,
+      "step": 2742
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.966745843230405e-05,
+      "loss": 3.4343,
+      "step": 2743
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.966152019002377e-05,
+      "loss": 3.334,
+      "step": 2744
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.965558194774347e-05,
+      "loss": 3.1966,
+      "step": 2745
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.964964370546318e-05,
+      "loss": 3.404,
+      "step": 2746
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.96437054631829e-05,
+      "loss": 3.2478,
+      "step": 2747
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.963776722090261e-05,
+      "loss": 3.3262,
+      "step": 2748
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.963182897862233e-05,
+      "loss": 3.3951,
+      "step": 2749
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.962589073634205e-05,
+      "loss": 3.3637,
+      "step": 2750
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.961995249406177e-05,
+      "loss": 3.5007,
+      "step": 2751
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.961401425178148e-05,
+      "loss": 3.4374,
+      "step": 2752
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.96080760095012e-05,
+      "loss": 3.4865,
+      "step": 2753
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.96021377672209e-05,
+      "loss": 3.2785,
+      "step": 2754
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.959619952494062e-05,
+      "loss": 3.5638,
+      "step": 2755
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.959026128266033e-05,
+      "loss": 3.2457,
+      "step": 2756
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.958432304038005e-05,
+      "loss": 3.4282,
+      "step": 2757
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.957838479809977e-05,
+      "loss": 3.3267,
+      "step": 2758
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.957244655581948e-05,
+      "loss": 3.0742,
+      "step": 2759
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.95665083135392e-05,
+      "loss": 3.2875,
+      "step": 2760
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.95605700712589e-05,
+      "loss": 3.2943,
+      "step": 2761
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.955463182897863e-05,
+      "loss": 3.2181,
+      "step": 2762
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.954869358669835e-05,
+      "loss": 3.314,
+      "step": 2763
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.954275534441807e-05,
+      "loss": 3.3099,
+      "step": 2764
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.953681710213777e-05,
+      "loss": 3.3149,
+      "step": 2765
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.953087885985748e-05,
+      "loss": 3.2497,
+      "step": 2766
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.95249406175772e-05,
+      "loss": 3.4307,
+      "step": 2767
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.951900237529691e-05,
+      "loss": 3.376,
+      "step": 2768
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 8.951306413301663e-05,
+      "loss": 3.2684,
+      "step": 2769
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.950712589073635e-05,
+      "loss": 3.259,
+      "step": 2770
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.950118764845607e-05,
+      "loss": 2.9972,
+      "step": 2771
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.949524940617578e-05,
+      "loss": 3.0711,
+      "step": 2772
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.94893111638955e-05,
+      "loss": 3.1685,
+      "step": 2773
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.94833729216152e-05,
+      "loss": 2.8482,
+      "step": 2774
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.947743467933492e-05,
+      "loss": 3.161,
+      "step": 2775
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.947149643705463e-05,
+      "loss": 2.8279,
+      "step": 2776
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.946555819477435e-05,
+      "loss": 3.626,
+      "step": 2777
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.945961995249407e-05,
+      "loss": 3.4511,
+      "step": 2778
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.945368171021378e-05,
+      "loss": 3.5893,
+      "step": 2779
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.94477434679335e-05,
+      "loss": 3.375,
+      "step": 2780
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.94418052256532e-05,
+      "loss": 3.317,
+      "step": 2781
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.943586698337292e-05,
+      "loss": 3.3964,
+      "step": 2782
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.942992874109264e-05,
+      "loss": 3.5314,
+      "step": 2783
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.942399049881237e-05,
+      "loss": 3.2697,
+      "step": 2784
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.941805225653207e-05,
+      "loss": 3.4639,
+      "step": 2785
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.941211401425179e-05,
+      "loss": 3.4671,
+      "step": 2786
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.94061757719715e-05,
+      "loss": 3.3799,
+      "step": 2787
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.94002375296912e-05,
+      "loss": 3.5013,
+      "step": 2788
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.939429928741093e-05,
+      "loss": 3.4277,
+      "step": 2789
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.938836104513065e-05,
+      "loss": 3.5202,
+      "step": 2790
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.938242280285037e-05,
+      "loss": 3.6615,
+      "step": 2791
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.937648456057007e-05,
+      "loss": 3.3518,
+      "step": 2792
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.93705463182898e-05,
+      "loss": 3.5344,
+      "step": 2793
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.93646080760095e-05,
+      "loss": 3.7516,
+      "step": 2794
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.935866983372922e-05,
+      "loss": 3.4119,
+      "step": 2795
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 8.935273159144893e-05,
+      "loss": 3.5008,
+      "step": 2796
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.934679334916865e-05,
+      "loss": 3.6436,
+      "step": 2797
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.934085510688837e-05,
+      "loss": 3.5914,
+      "step": 2798
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.933491686460808e-05,
+      "loss": 3.3579,
+      "step": 2799
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.93289786223278e-05,
+      "loss": 3.5624,
+      "step": 2800
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.93230403800475e-05,
+      "loss": 3.2951,
+      "step": 2801
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.931710213776722e-05,
+      "loss": 3.409,
+      "step": 2802
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.931116389548694e-05,
+      "loss": 3.3868,
+      "step": 2803
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.930522565320666e-05,
+      "loss": 3.4278,
+      "step": 2804
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.929928741092637e-05,
+      "loss": 3.192,
+      "step": 2805
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.929334916864609e-05,
+      "loss": 3.6972,
+      "step": 2806
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.92874109263658e-05,
+      "loss": 3.2009,
+      "step": 2807
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.92814726840855e-05,
+      "loss": 3.1873,
+      "step": 2808
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.927553444180523e-05,
+      "loss": 3.3182,
+      "step": 2809
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.926959619952495e-05,
+      "loss": 3.4492,
+      "step": 2810
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.926365795724467e-05,
+      "loss": 3.3258,
+      "step": 2811
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.925771971496437e-05,
+      "loss": 3.5376,
+      "step": 2812
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.925178147268409e-05,
+      "loss": 3.2133,
+      "step": 2813
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 8.92458432304038e-05,
+      "loss": 3.3527,
+      "step": 2814
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.923990498812352e-05,
+      "loss": 3.3132,
+      "step": 2815
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.923396674584324e-05,
+      "loss": 3.3486,
+      "step": 2816
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.922802850356296e-05,
+      "loss": 3.2458,
+      "step": 2817
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.922209026128267e-05,
+      "loss": 3.1736,
+      "step": 2818
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.921615201900237e-05,
+      "loss": 3.1214,
+      "step": 2819
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.92102137767221e-05,
+      "loss": 3.1319,
+      "step": 2820
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.92042755344418e-05,
+      "loss": 3.2131,
+      "step": 2821
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.919833729216152e-05,
+      "loss": 3.1744,
+      "step": 2822
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.919239904988124e-05,
+      "loss": 3.4892,
+      "step": 2823
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.918646080760096e-05,
+      "loss": 2.9446,
+      "step": 2824
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.918052256532067e-05,
+      "loss": 2.8372,
+      "step": 2825
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.917458432304039e-05,
+      "loss": 2.5446,
+      "step": 2826
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.91686460807601e-05,
+      "loss": 3.7032,
+      "step": 2827
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.916270783847982e-05,
+      "loss": 3.3049,
+      "step": 2828
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.915676959619952e-05,
+      "loss": 3.6621,
+      "step": 2829
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.915083135391924e-05,
+      "loss": 3.436,
+      "step": 2830
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.914489311163896e-05,
+      "loss": 3.6833,
+      "step": 2831
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.913895486935867e-05,
+      "loss": 3.617,
+      "step": 2832
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.913301662707839e-05,
+      "loss": 3.3622,
+      "step": 2833
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.91270783847981e-05,
+      "loss": 3.5339,
+      "step": 2834
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.912114014251782e-05,
+      "loss": 3.3206,
+      "step": 2835
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.911520190023754e-05,
+      "loss": 3.4256,
+      "step": 2836
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.910926365795726e-05,
+      "loss": 3.545,
+      "step": 2837
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.910332541567697e-05,
+      "loss": 3.3884,
+      "step": 2838
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.909738717339667e-05,
+      "loss": 3.5845,
+      "step": 2839
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.90914489311164e-05,
+      "loss": 3.538,
+      "step": 2840
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.90855106888361e-05,
+      "loss": 3.3098,
+      "step": 2841
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.907957244655582e-05,
+      "loss": 3.3815,
+      "step": 2842
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.907363420427554e-05,
+      "loss": 3.5518,
+      "step": 2843
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.906769596199526e-05,
+      "loss": 3.4724,
+      "step": 2844
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.906175771971497e-05,
+      "loss": 3.4915,
+      "step": 2845
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.905581947743469e-05,
+      "loss": 3.4375,
+      "step": 2846
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.90498812351544e-05,
+      "loss": 3.5709,
+      "step": 2847
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.904394299287412e-05,
+      "loss": 3.3609,
+      "step": 2848
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.903800475059382e-05,
+      "loss": 3.3205,
+      "step": 2849
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.903206650831354e-05,
+      "loss": 3.2707,
+      "step": 2850
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.902612826603326e-05,
+      "loss": 3.3343,
+      "step": 2851
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.902019002375297e-05,
+      "loss": 3.6056,
+      "step": 2852
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.901425178147269e-05,
+      "loss": 3.2558,
+      "step": 2853
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.90083135391924e-05,
+      "loss": 3.7381,
+      "step": 2854
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.900237529691212e-05,
+      "loss": 3.2933,
+      "step": 2855
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.899643705463184e-05,
+      "loss": 3.4688,
+      "step": 2856
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.899049881235156e-05,
+      "loss": 3.5557,
+      "step": 2857
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.898456057007127e-05,
+      "loss": 3.2739,
+      "step": 2858
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.897862232779099e-05,
+      "loss": 3.3442,
+      "step": 2859
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.897268408551069e-05,
+      "loss": 3.3846,
+      "step": 2860
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.89667458432304e-05,
+      "loss": 3.2657,
+      "step": 2861
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.896080760095012e-05,
+      "loss": 3.5134,
+      "step": 2862
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.895486935866984e-05,
+      "loss": 3.2608,
+      "step": 2863
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.894893111638956e-05,
+      "loss": 3.2734,
+      "step": 2864
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.894299287410927e-05,
+      "loss": 3.2155,
+      "step": 2865
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.893705463182899e-05,
+      "loss": 3.2757,
+      "step": 2866
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.89311163895487e-05,
+      "loss": 3.4553,
+      "step": 2867
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.892517814726841e-05,
+      "loss": 3.4498,
+      "step": 2868
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.891923990498812e-05,
+      "loss": 3.1517,
+      "step": 2869
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.891330166270784e-05,
+      "loss": 3.0596,
+      "step": 2870
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.890736342042756e-05,
+      "loss": 3.1817,
+      "step": 2871
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.890142517814727e-05,
+      "loss": 2.871,
+      "step": 2872
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.889548693586699e-05,
+      "loss": 3.0313,
+      "step": 2873
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.88895486935867e-05,
+      "loss": 3.0244,
+      "step": 2874
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.888361045130642e-05,
+      "loss": 2.8008,
+      "step": 2875
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.887767220902614e-05,
+      "loss": 2.4118,
+      "step": 2876
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.887173396674586e-05,
+      "loss": 3.4359,
+      "step": 2877
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.886579572446556e-05,
+      "loss": 3.4465,
+      "step": 2878
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.885985748218528e-05,
+      "loss": 3.5715,
+      "step": 2879
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.885391923990499e-05,
+      "loss": 3.5405,
+      "step": 2880
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.88479809976247e-05,
+      "loss": 3.498,
+      "step": 2881
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.884204275534442e-05,
+      "loss": 3.3388,
+      "step": 2882
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.883610451306414e-05,
+      "loss": 3.3536,
+      "step": 2883
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.883016627078386e-05,
+      "loss": 3.5638,
+      "step": 2884
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.882422802850357e-05,
+      "loss": 3.6139,
+      "step": 2885
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.881828978622329e-05,
+      "loss": 3.6631,
+      "step": 2886
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.881235154394299e-05,
+      "loss": 3.2613,
+      "step": 2887
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.880641330166271e-05,
+      "loss": 3.6119,
+      "step": 2888
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.880047505938242e-05,
+      "loss": 3.5328,
+      "step": 2889
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.879453681710215e-05,
+      "loss": 3.3936,
+      "step": 2890
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.878859857482186e-05,
+      "loss": 3.537,
+      "step": 2891
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.878266033254157e-05,
+      "loss": 3.54,
+      "step": 2892
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.877672209026129e-05,
+      "loss": 3.418,
+      "step": 2893
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.8770783847981e-05,
+      "loss": 3.378,
+      "step": 2894
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.876484560570072e-05,
+      "loss": 3.5357,
+      "step": 2895
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.875890736342044e-05,
+      "loss": 3.2631,
+      "step": 2896
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.875296912114016e-05,
+      "loss": 3.4193,
+      "step": 2897
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.874703087885986e-05,
+      "loss": 3.3771,
+      "step": 2898
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.874109263657958e-05,
+      "loss": 3.3464,
+      "step": 2899
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.873515439429929e-05,
+      "loss": 3.7011,
+      "step": 2900
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.8729216152019e-05,
+      "loss": 3.2202,
+      "step": 2901
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.872327790973872e-05,
+      "loss": 3.4403,
+      "step": 2902
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.871733966745844e-05,
+      "loss": 3.3805,
+      "step": 2903
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.871140142517816e-05,
+      "loss": 3.3342,
+      "step": 2904
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.870546318289786e-05,
+      "loss": 3.5223,
+      "step": 2905
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.869952494061758e-05,
+      "loss": 3.3181,
+      "step": 2906
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.869358669833729e-05,
+      "loss": 3.3527,
+      "step": 2907
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.868764845605701e-05,
+      "loss": 3.4028,
+      "step": 2908
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.868171021377672e-05,
+      "loss": 3.3329,
+      "step": 2909
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.867577197149645e-05,
+      "loss": 3.4525,
+      "step": 2910
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.866983372921616e-05,
+      "loss": 3.4378,
+      "step": 2911
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.866389548693587e-05,
+      "loss": 3.271,
+      "step": 2912
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.865795724465559e-05,
+      "loss": 3.3421,
+      "step": 2913
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.86520190023753e-05,
+      "loss": 3.4318,
+      "step": 2914
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.864608076009501e-05,
+      "loss": 3.137,
+      "step": 2915
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.864014251781473e-05,
+      "loss": 3.0916,
+      "step": 2916
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.863420427553445e-05,
+      "loss": 3.2168,
+      "step": 2917
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.862826603325416e-05,
+      "loss": 3.1569,
+      "step": 2918
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.862232779097388e-05,
+      "loss": 3.3764,
+      "step": 2919
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.861638954869359e-05,
+      "loss": 3.1122,
+      "step": 2920
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.861045130641331e-05,
+      "loss": 2.9334,
+      "step": 2921
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.860451306413302e-05,
+      "loss": 3.0518,
+      "step": 2922
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.859857482185274e-05,
+      "loss": 3.4676,
+      "step": 2923
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.859263657957246e-05,
+      "loss": 2.9875,
+      "step": 2924
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.858669833729216e-05,
+      "loss": 2.9674,
+      "step": 2925
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.858076009501188e-05,
+      "loss": 2.9525,
+      "step": 2926
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.857482185273159e-05,
+      "loss": 3.6674,
+      "step": 2927
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.856888361045131e-05,
+      "loss": 3.6227,
+      "step": 2928
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.856294536817102e-05,
+      "loss": 3.3195,
+      "step": 2929
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.855700712589075e-05,
+      "loss": 3.4026,
+      "step": 2930
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.855106888361046e-05,
+      "loss": 3.4216,
+      "step": 2931
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.854513064133016e-05,
+      "loss": 3.1633,
+      "step": 2932
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.853919239904989e-05,
+      "loss": 3.565,
+      "step": 2933
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.853325415676959e-05,
+      "loss": 3.5741,
+      "step": 2934
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.852731591448931e-05,
+      "loss": 3.5552,
+      "step": 2935
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.852137767220903e-05,
+      "loss": 3.4277,
+      "step": 2936
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.851543942992875e-05,
+      "loss": 3.3635,
+      "step": 2937
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.850950118764846e-05,
+      "loss": 3.3268,
+      "step": 2938
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.850356294536818e-05,
+      "loss": 3.3734,
+      "step": 2939
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.849762470308789e-05,
+      "loss": 3.2342,
+      "step": 2940
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.849168646080761e-05,
+      "loss": 3.4811,
+      "step": 2941
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.848574821852731e-05,
+      "loss": 3.455,
+      "step": 2942
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.847980997624703e-05,
+      "loss": 3.3735,
+      "step": 2943
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.847387173396676e-05,
+      "loss": 3.4277,
+      "step": 2944
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.846793349168646e-05,
+      "loss": 3.4529,
+      "step": 2945
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.846199524940618e-05,
+      "loss": 3.4719,
+      "step": 2946
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.845605700712589e-05,
+      "loss": 3.2146,
+      "step": 2947
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.845011876484561e-05,
+      "loss": 3.345,
+      "step": 2948
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.844418052256533e-05,
+      "loss": 3.3112,
+      "step": 2949
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.843824228028505e-05,
+      "loss": 3.4842,
+      "step": 2950
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.843230403800476e-05,
+      "loss": 3.388,
+      "step": 2951
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.842636579572448e-05,
+      "loss": 3.2822,
+      "step": 2952
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.842042755344418e-05,
+      "loss": 3.2476,
+      "step": 2953
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.841448931116389e-05,
+      "loss": 3.314,
+      "step": 2954
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.840855106888361e-05,
+      "loss": 3.3372,
+      "step": 2955
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.840261282660333e-05,
+      "loss": 3.5643,
+      "step": 2956
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.839667458432305e-05,
+      "loss": 3.3066,
+      "step": 2957
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.839073634204276e-05,
+      "loss": 3.619,
+      "step": 2958
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.838479809976248e-05,
+      "loss": 3.5506,
+      "step": 2959
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.837885985748219e-05,
+      "loss": 3.4786,
+      "step": 2960
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.83729216152019e-05,
+      "loss": 3.3008,
+      "step": 2961
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.836698337292161e-05,
+      "loss": 3.5551,
+      "step": 2962
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.836104513064133e-05,
+      "loss": 3.2878,
+      "step": 2963
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.835510688836105e-05,
+      "loss": 3.1459,
+      "step": 2964
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.834916864608076e-05,
+      "loss": 3.5309,
+      "step": 2965
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.834323040380048e-05,
+      "loss": 3.3384,
+      "step": 2966
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.833729216152019e-05,
+      "loss": 3.198,
+      "step": 2967
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.833135391923991e-05,
+      "loss": 3.1384,
+      "step": 2968
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.832541567695963e-05,
+      "loss": 3.4589,
+      "step": 2969
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.831947743467935e-05,
+      "loss": 3.1247,
+      "step": 2970
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.831353919239906e-05,
+      "loss": 3.4123,
+      "step": 2971
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.830760095011878e-05,
+      "loss": 3.0003,
+      "step": 2972
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.830166270783848e-05,
+      "loss": 2.9313,
+      "step": 2973
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.829572446555819e-05,
+      "loss": 2.7487,
+      "step": 2974
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.828978622327791e-05,
+      "loss": 2.8043,
+      "step": 2975
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.828384798099763e-05,
+      "loss": 2.4944,
+      "step": 2976
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.827790973871735e-05,
+      "loss": 3.6382,
+      "step": 2977
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.827197149643706e-05,
+      "loss": 3.3741,
+      "step": 2978
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.826603325415678e-05,
+      "loss": 3.4551,
+      "step": 2979
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.826009501187648e-05,
+      "loss": 3.3396,
+      "step": 2980
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.82541567695962e-05,
+      "loss": 3.4879,
+      "step": 2981
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.824821852731591e-05,
+      "loss": 3.7206,
+      "step": 2982
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.824228028503565e-05,
+      "loss": 3.4491,
+      "step": 2983
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.823634204275535e-05,
+      "loss": 3.721,
+      "step": 2984
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.823040380047506e-05,
+      "loss": 3.5473,
+      "step": 2985
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.822446555819478e-05,
+      "loss": 3.3798,
+      "step": 2986
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.821852731591449e-05,
+      "loss": 3.614,
+      "step": 2987
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.821258907363421e-05,
+      "loss": 3.3824,
+      "step": 2988
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.820665083135393e-05,
+      "loss": 3.3387,
+      "step": 2989
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.820071258907365e-05,
+      "loss": 3.2905,
+      "step": 2990
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.819477434679335e-05,
+      "loss": 3.2131,
+      "step": 2991
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.818883610451307e-05,
+      "loss": 3.534,
+      "step": 2992
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.818289786223278e-05,
+      "loss": 3.5054,
+      "step": 2993
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.817695961995249e-05,
+      "loss": 3.3325,
+      "step": 2994
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.817102137767221e-05,
+      "loss": 3.2832,
+      "step": 2995
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.816508313539193e-05,
+      "loss": 3.5243,
+      "step": 2996
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.815914489311165e-05,
+      "loss": 3.7036,
+      "step": 2997
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.815320665083136e-05,
+      "loss": 3.2205,
+      "step": 2998
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.814726840855108e-05,
+      "loss": 3.2458,
+      "step": 2999
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.814133016627078e-05,
+      "loss": 3.4592,
+      "step": 3000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.81353919239905e-05,
+      "loss": 3.3514,
+      "step": 3001
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.812945368171021e-05,
+      "loss": 3.2481,
+      "step": 3002
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.812351543942994e-05,
+      "loss": 3.275,
+      "step": 3003
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.811757719714965e-05,
+      "loss": 3.5407,
+      "step": 3004
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.811163895486936e-05,
+      "loss": 3.4012,
+      "step": 3005
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.810570071258908e-05,
+      "loss": 3.4484,
+      "step": 3006
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.809976247030879e-05,
+      "loss": 3.2955,
+      "step": 3007
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.80938242280285e-05,
+      "loss": 3.302,
+      "step": 3008
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.808788598574823e-05,
+      "loss": 3.3939,
+      "step": 3009
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.808194774346795e-05,
+      "loss": 3.3877,
+      "step": 3010
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.807600950118765e-05,
+      "loss": 3.2544,
+      "step": 3011
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.807007125890737e-05,
+      "loss": 3.0787,
+      "step": 3012
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.806413301662708e-05,
+      "loss": 3.4095,
+      "step": 3013
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.80581947743468e-05,
+      "loss": 3.2795,
+      "step": 3014
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.805225653206651e-05,
+      "loss": 3.2157,
+      "step": 3015
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.804631828978623e-05,
+      "loss": 3.3815,
+      "step": 3016
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.804038004750595e-05,
+      "loss": 3.2082,
+      "step": 3017
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.803444180522565e-05,
+      "loss": 2.9827,
+      "step": 3018
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.802850356294538e-05,
+      "loss": 3.17,
+      "step": 3019
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.802256532066508e-05,
+      "loss": 3.2886,
+      "step": 3020
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.80166270783848e-05,
+      "loss": 3.2431,
+      "step": 3021
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.801068883610451e-05,
+      "loss": 3.1543,
+      "step": 3022
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.800475059382424e-05,
+      "loss": 2.847,
+      "step": 3023
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.799881235154395e-05,
+      "loss": 2.7693,
+      "step": 3024
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.799287410926367e-05,
+      "loss": 2.6653,
+      "step": 3025
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.798693586698338e-05,
+      "loss": 2.4078,
+      "step": 3026
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.798099762470308e-05,
+      "loss": 3.5742,
+      "step": 3027
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.79750593824228e-05,
+      "loss": 3.4357,
+      "step": 3028
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.796912114014252e-05,
+      "loss": 3.5131,
+      "step": 3029
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.796318289786224e-05,
+      "loss": 3.7293,
+      "step": 3030
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.795724465558195e-05,
+      "loss": 3.2632,
+      "step": 3031
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.795130641330167e-05,
+      "loss": 3.3207,
+      "step": 3032
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.794536817102138e-05,
+      "loss": 3.528,
+      "step": 3033
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.79394299287411e-05,
+      "loss": 3.5384,
+      "step": 3034
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.79334916864608e-05,
+      "loss": 3.5509,
+      "step": 3035
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.792755344418053e-05,
+      "loss": 3.562,
+      "step": 3036
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.792161520190025e-05,
+      "loss": 3.3873,
+      "step": 3037
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.791567695961995e-05,
+      "loss": 3.4239,
+      "step": 3038
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.790973871733967e-05,
+      "loss": 4.0785,
+      "step": 3039
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.790380047505938e-05,
+      "loss": 3.4094,
+      "step": 3040
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.78978622327791e-05,
+      "loss": 3.2109,
+      "step": 3041
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.789192399049881e-05,
+      "loss": 3.1111,
+      "step": 3042
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.788598574821854e-05,
+      "loss": 3.4706,
+      "step": 3043
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.788004750593825e-05,
+      "loss": 3.4006,
+      "step": 3044
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.787410926365797e-05,
+      "loss": 3.2569,
+      "step": 3045
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.786817102137768e-05,
+      "loss": 3.2626,
+      "step": 3046
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.786223277909738e-05,
+      "loss": 3.1787,
+      "step": 3047
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.78562945368171e-05,
+      "loss": 3.2722,
+      "step": 3048
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.785035629453682e-05,
+      "loss": 3.3375,
+      "step": 3049
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.784441805225654e-05,
+      "loss": 3.3235,
+      "step": 3050
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.783847980997625e-05,
+      "loss": 3.5626,
+      "step": 3051
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.783254156769597e-05,
+      "loss": 3.6256,
+      "step": 3052
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.782660332541568e-05,
+      "loss": 3.3731,
+      "step": 3053
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.78206650831354e-05,
+      "loss": 3.3549,
+      "step": 3054
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.78147268408551e-05,
+      "loss": 3.1776,
+      "step": 3055
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.780878859857483e-05,
+      "loss": 3.0827,
+      "step": 3056
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.780285035629455e-05,
+      "loss": 3.4089,
+      "step": 3057
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.779691211401425e-05,
+      "loss": 3.2646,
+      "step": 3058
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.779097387173397e-05,
+      "loss": 3.3645,
+      "step": 3059
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.778503562945368e-05,
+      "loss": 3.1076,
+      "step": 3060
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.77790973871734e-05,
+      "loss": 3.5049,
+      "step": 3061
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.777315914489312e-05,
+      "loss": 3.3075,
+      "step": 3062
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.776722090261284e-05,
+      "loss": 3.0171,
+      "step": 3063
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.776128266033255e-05,
+      "loss": 3.2177,
+      "step": 3064
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.775534441805227e-05,
+      "loss": 3.2061,
+      "step": 3065
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.774940617577197e-05,
+      "loss": 3.1873,
+      "step": 3066
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.774346793349168e-05,
+      "loss": 3.3154,
+      "step": 3067
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.77375296912114e-05,
+      "loss": 3.1624,
+      "step": 3068
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.773159144893112e-05,
+      "loss": 3.4347,
+      "step": 3069
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.772565320665084e-05,
+      "loss": 3.0036,
+      "step": 3070
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.771971496437055e-05,
+      "loss": 3.1241,
+      "step": 3071
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.771377672209027e-05,
+      "loss": 2.964,
+      "step": 3072
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.770783847980998e-05,
+      "loss": 2.8891,
+      "step": 3073
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.77019002375297e-05,
+      "loss": 2.8799,
+      "step": 3074
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.76959619952494e-05,
+      "loss": 2.7104,
+      "step": 3075
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.769002375296912e-05,
+      "loss": 2.664,
+      "step": 3076
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.768408551068884e-05,
+      "loss": 3.6619,
+      "step": 3077
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.767814726840855e-05,
+      "loss": 3.47,
+      "step": 3078
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.767220902612827e-05,
+      "loss": 3.5485,
+      "step": 3079
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.766627078384798e-05,
+      "loss": 3.5353,
+      "step": 3080
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.76603325415677e-05,
+      "loss": 3.4585,
+      "step": 3081
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.765439429928742e-05,
+      "loss": 3.448,
+      "step": 3082
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.764845605700714e-05,
+      "loss": 3.3614,
+      "step": 3083
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.764251781472685e-05,
+      "loss": 3.2344,
+      "step": 3084
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.763657957244657e-05,
+      "loss": 3.5906,
+      "step": 3085
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.763064133016627e-05,
+      "loss": 3.4056,
+      "step": 3086
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.7624703087886e-05,
+      "loss": 3.5246,
+      "step": 3087
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.76187648456057e-05,
+      "loss": 3.3835,
+      "step": 3088
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.761282660332542e-05,
+      "loss": 3.5365,
+      "step": 3089
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.760688836104514e-05,
+      "loss": 3.2063,
+      "step": 3090
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.760095011876485e-05,
+      "loss": 3.2575,
+      "step": 3091
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.759501187648457e-05,
+      "loss": 3.3076,
+      "step": 3092
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.758907363420428e-05,
+      "loss": 3.3974,
+      "step": 3093
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.7583135391924e-05,
+      "loss": 3.395,
+      "step": 3094
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.75771971496437e-05,
+      "loss": 3.1588,
+      "step": 3095
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.757125890736342e-05,
+      "loss": 3.3646,
+      "step": 3096
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.756532066508314e-05,
+      "loss": 3.3062,
+      "step": 3097
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.755938242280285e-05,
+      "loss": 3.4029,
+      "step": 3098
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.755344418052257e-05,
+      "loss": 3.6587,
+      "step": 3099
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.754750593824228e-05,
+      "loss": 3.2721,
+      "step": 3100
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.7541567695962e-05,
+      "loss": 3.2521,
+      "step": 3101
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.753562945368172e-05,
+      "loss": 3.355,
+      "step": 3102
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.752969121140144e-05,
+      "loss": 3.3551,
+      "step": 3103
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.752375296912114e-05,
+      "loss": 3.2588,
+      "step": 3104
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.751781472684087e-05,
+      "loss": 3.2972,
+      "step": 3105
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.751187648456057e-05,
+      "loss": 3.4391,
+      "step": 3106
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.750593824228029e-05,
+      "loss": 3.3979,
+      "step": 3107
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.75e-05,
+      "loss": 3.533,
+      "step": 3108
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.749406175771972e-05,
+      "loss": 3.3266,
+      "step": 3109
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.748812351543944e-05,
+      "loss": 3.1687,
+      "step": 3110
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.748218527315915e-05,
+      "loss": 3.3135,
+      "step": 3111
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.747624703087887e-05,
+      "loss": 3.2718,
+      "step": 3112
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.747030878859857e-05,
+      "loss": 3.1231,
+      "step": 3113
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.74643705463183e-05,
+      "loss": 3.3003,
+      "step": 3114
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.7458432304038e-05,
+      "loss": 3.1662,
+      "step": 3115
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.745249406175773e-05,
+      "loss": 3.2141,
+      "step": 3116
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.744655581947744e-05,
+      "loss": 3.2226,
+      "step": 3117
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.744061757719716e-05,
+      "loss": 3.2537,
+      "step": 3118
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.743467933491687e-05,
+      "loss": 2.9254,
+      "step": 3119
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.742874109263658e-05,
+      "loss": 3.0881,
+      "step": 3120
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.74228028503563e-05,
+      "loss": 3.0874,
+      "step": 3121
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.741686460807602e-05,
+      "loss": 2.6832,
+      "step": 3122
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.741092636579574e-05,
+      "loss": 3.3416,
+      "step": 3123
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.740498812351544e-05,
+      "loss": 3.0658,
+      "step": 3124
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.739904988123516e-05,
+      "loss": 3.1696,
+      "step": 3125
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.739311163895487e-05,
+      "loss": 2.6013,
+      "step": 3126
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.738717339667459e-05,
+      "loss": 3.4781,
+      "step": 3127
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.73812351543943e-05,
+      "loss": 3.4203,
+      "step": 3128
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.737529691211402e-05,
+      "loss": 3.4989,
+      "step": 3129
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.736935866983374e-05,
+      "loss": 3.3763,
+      "step": 3130
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.736342042755345e-05,
+      "loss": 3.039,
+      "step": 3131
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.735748218527317e-05,
+      "loss": 3.3728,
+      "step": 3132
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.735154394299287e-05,
+      "loss": 3.453,
+      "step": 3133
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.734560570071259e-05,
+      "loss": 3.3069,
+      "step": 3134
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.73396674584323e-05,
+      "loss": 3.4316,
+      "step": 3135
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.733372921615203e-05,
+      "loss": 3.2467,
+      "step": 3136
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.732779097387174e-05,
+      "loss": 3.4902,
+      "step": 3137
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.732185273159146e-05,
+      "loss": 3.368,
+      "step": 3138
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.731591448931117e-05,
+      "loss": 3.0989,
+      "step": 3139
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.730997624703087e-05,
+      "loss": 3.2141,
+      "step": 3140
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.73040380047506e-05,
+      "loss": 3.2483,
+      "step": 3141
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.729809976247032e-05,
+      "loss": 3.9616,
+      "step": 3142
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.729216152019004e-05,
+      "loss": 3.2622,
+      "step": 3143
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.728622327790974e-05,
+      "loss": 3.3039,
+      "step": 3144
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.728028503562946e-05,
+      "loss": 3.208,
+      "step": 3145
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.727434679334917e-05,
+      "loss": 3.3183,
+      "step": 3146
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.726840855106889e-05,
+      "loss": 3.2283,
+      "step": 3147
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.72624703087886e-05,
+      "loss": 3.3412,
+      "step": 3148
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.725653206650832e-05,
+      "loss": 3.3142,
+      "step": 3149
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.725059382422804e-05,
+      "loss": 3.2207,
+      "step": 3150
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.724465558194774e-05,
+      "loss": 3.1692,
+      "step": 3151
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.723871733966746e-05,
+      "loss": 3.362,
+      "step": 3152
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.723277909738717e-05,
+      "loss": 3.1418,
+      "step": 3153
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.722684085510689e-05,
+      "loss": 3.3644,
+      "step": 3154
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.72209026128266e-05,
+      "loss": 3.1839,
+      "step": 3155
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.721496437054633e-05,
+      "loss": 3.4864,
+      "step": 3156
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.720902612826604e-05,
+      "loss": 3.0884,
+      "step": 3157
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.720308788598576e-05,
+      "loss": 3.4051,
+      "step": 3158
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.719714964370547e-05,
+      "loss": 3.0451,
+      "step": 3159
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.719121140142517e-05,
+      "loss": 3.1285,
+      "step": 3160
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.71852731591449e-05,
+      "loss": 3.2783,
+      "step": 3161
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.717933491686461e-05,
+      "loss": 3.1037,
+      "step": 3162
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.717339667458433e-05,
+      "loss": 3.1381,
+      "step": 3163
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.716745843230404e-05,
+      "loss": 3.0003,
+      "step": 3164
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.716152019002376e-05,
+      "loss": 3.1185,
+      "step": 3165
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.715558194774347e-05,
+      "loss": 3.1444,
+      "step": 3166
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.714964370546319e-05,
+      "loss": 3.0827,
+      "step": 3167
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.71437054631829e-05,
+      "loss": 3.0045,
+      "step": 3168
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.713776722090262e-05,
+      "loss": 2.9071,
+      "step": 3169
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.713182897862234e-05,
+      "loss": 2.9275,
+      "step": 3170
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.712589073634204e-05,
+      "loss": 2.7789,
+      "step": 3171
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.711995249406176e-05,
+      "loss": 3.3202,
+      "step": 3172
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.711401425178147e-05,
+      "loss": 2.9494,
+      "step": 3173
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.710807600950119e-05,
+      "loss": 3.0134,
+      "step": 3174
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.71021377672209e-05,
+      "loss": 2.3604,
+      "step": 3175
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.709619952494063e-05,
+      "loss": 2.4877,
+      "step": 3176
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.709026128266034e-05,
+      "loss": 3.3682,
+      "step": 3177
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.708432304038006e-05,
+      "loss": 3.5522,
+      "step": 3178
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.707838479809976e-05,
+      "loss": 3.4212,
+      "step": 3179
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.707244655581949e-05,
+      "loss": 3.612,
+      "step": 3180
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.706650831353919e-05,
+      "loss": 3.3466,
+      "step": 3181
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.706057007125891e-05,
+      "loss": 3.4352,
+      "step": 3182
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.705463182897863e-05,
+      "loss": 3.4445,
+      "step": 3183
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.704869358669834e-05,
+      "loss": 3.1813,
+      "step": 3184
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.704275534441806e-05,
+      "loss": 3.3551,
+      "step": 3185
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.703681710213777e-05,
+      "loss": 3.3772,
+      "step": 3186
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.703087885985749e-05,
+      "loss": 3.415,
+      "step": 3187
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.70249406175772e-05,
+      "loss": 3.2978,
+      "step": 3188
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.701900237529691e-05,
+      "loss": 3.5735,
+      "step": 3189
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.701306413301663e-05,
+      "loss": 3.2271,
+      "step": 3190
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.700712589073634e-05,
+      "loss": 3.6392,
+      "step": 3191
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.700118764845606e-05,
+      "loss": 3.1737,
+      "step": 3192
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.699524940617577e-05,
+      "loss": 3.3434,
+      "step": 3193
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.698931116389549e-05,
+      "loss": 3.3835,
+      "step": 3194
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.698337292161521e-05,
+      "loss": 3.1908,
+      "step": 3195
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.697743467933493e-05,
+      "loss": 3.2306,
+      "step": 3196
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.697149643705464e-05,
+      "loss": 3.2468,
+      "step": 3197
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.696555819477436e-05,
+      "loss": 3.3924,
+      "step": 3198
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.695961995249406e-05,
+      "loss": 3.2026,
+      "step": 3199
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.695368171021378e-05,
+      "loss": 3.3013,
+      "step": 3200
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.694774346793349e-05,
+      "loss": 3.2222,
+      "step": 3201
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.694180522565321e-05,
+      "loss": 3.1943,
+      "step": 3202
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.693586698337293e-05,
+      "loss": 3.2299,
+      "step": 3203
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.692992874109264e-05,
+      "loss": 2.9893,
+      "step": 3204
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.692399049881236e-05,
+      "loss": 3.0891,
+      "step": 3205
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.691805225653207e-05,
+      "loss": 3.4638,
+      "step": 3206
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.691211401425179e-05,
+      "loss": 3.2846,
+      "step": 3207
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.690617577197149e-05,
+      "loss": 3.2165,
+      "step": 3208
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.690023752969121e-05,
+      "loss": 3.0378,
+      "step": 3209
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.689429928741093e-05,
+      "loss": 3.0153,
+      "step": 3210
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.688836104513065e-05,
+      "loss": 2.941,
+      "step": 3211
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.688242280285036e-05,
+      "loss": 3.3507,
+      "step": 3212
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.687648456057007e-05,
+      "loss": 3.0724,
+      "step": 3213
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.687054631828979e-05,
+      "loss": 2.8572,
+      "step": 3214
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.686460807600951e-05,
+      "loss": 2.8489,
+      "step": 3215
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.685866983372923e-05,
+      "loss": 3.2128,
+      "step": 3216
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.685273159144894e-05,
+      "loss": 3.0592,
+      "step": 3217
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.684679334916866e-05,
+      "loss": 3.1836,
+      "step": 3218
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.684085510688836e-05,
+      "loss": 2.922,
+      "step": 3219
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.683491686460808e-05,
+      "loss": 2.7354,
+      "step": 3220
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.682897862232779e-05,
+      "loss": 2.8818,
+      "step": 3221
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.682304038004751e-05,
+      "loss": 2.7813,
+      "step": 3222
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.681710213776723e-05,
+      "loss": 2.7394,
+      "step": 3223
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.681116389548694e-05,
+      "loss": 2.5725,
+      "step": 3224
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.680522565320666e-05,
+      "loss": 2.957,
+      "step": 3225
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.679928741092636e-05,
+      "loss": 2.1902,
+      "step": 3226
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.679334916864608e-05,
+      "loss": 3.5278,
+      "step": 3227
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.678741092636579e-05,
+      "loss": 3.6728,
+      "step": 3228
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.678147268408553e-05,
+      "loss": 3.5286,
+      "step": 3229
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.677553444180523e-05,
+      "loss": 3.2726,
+      "step": 3230
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.676959619952495e-05,
+      "loss": 3.168,
+      "step": 3231
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.676365795724466e-05,
+      "loss": 3.3694,
+      "step": 3232
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.675771971496437e-05,
+      "loss": 3.1501,
+      "step": 3233
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.675178147268409e-05,
+      "loss": 3.4991,
+      "step": 3234
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.674584323040381e-05,
+      "loss": 3.256,
+      "step": 3235
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.673990498812353e-05,
+      "loss": 3.5425,
+      "step": 3236
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.673396674584323e-05,
+      "loss": 3.5567,
+      "step": 3237
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.672802850356295e-05,
+      "loss": 3.2162,
+      "step": 3238
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.672209026128266e-05,
+      "loss": 3.2016,
+      "step": 3239
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.671615201900238e-05,
+      "loss": 3.2473,
+      "step": 3240
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.671021377672209e-05,
+      "loss": 3.2036,
+      "step": 3241
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.670427553444181e-05,
+      "loss": 3.3664,
+      "step": 3242
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.669833729216153e-05,
+      "loss": 3.1038,
+      "step": 3243
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.669239904988124e-05,
+      "loss": 3.3225,
+      "step": 3244
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.668646080760096e-05,
+      "loss": 3.0442,
+      "step": 3245
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.668052256532066e-05,
+      "loss": 3.386,
+      "step": 3246
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.667458432304038e-05,
+      "loss": 3.1868,
+      "step": 3247
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.666864608076009e-05,
+      "loss": 3.113,
+      "step": 3248
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.666270783847982e-05,
+      "loss": 3.2142,
+      "step": 3249
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.665676959619953e-05,
+      "loss": 3.3358,
+      "step": 3250
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.665083135391925e-05,
+      "loss": 3.2142,
+      "step": 3251
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.664489311163896e-05,
+      "loss": 3.1457,
+      "step": 3252
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.663895486935868e-05,
+      "loss": 3.3472,
+      "step": 3253
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.663301662707839e-05,
+      "loss": 3.098,
+      "step": 3254
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.66270783847981e-05,
+      "loss": 3.3017,
+      "step": 3255
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.662114014251783e-05,
+      "loss": 3.0854,
+      "step": 3256
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.661520190023753e-05,
+      "loss": 3.1198,
+      "step": 3257
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.660926365795725e-05,
+      "loss": 3.2436,
+      "step": 3258
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.660332541567696e-05,
+      "loss": 3.3195,
+      "step": 3259
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.659738717339668e-05,
+      "loss": 3.2944,
+      "step": 3260
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.659144893111639e-05,
+      "loss": 3.0967,
+      "step": 3261
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.658551068883611e-05,
+      "loss": 2.9612,
+      "step": 3262
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.657957244655583e-05,
+      "loss": 3.1036,
+      "step": 3263
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.657363420427553e-05,
+      "loss": 3.0666,
+      "step": 3264
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.656769596199525e-05,
+      "loss": 3.3814,
+      "step": 3265
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.656175771971496e-05,
+      "loss": 2.8585,
+      "step": 3266
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.655581947743468e-05,
+      "loss": 3.2235,
+      "step": 3267
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.654988123515439e-05,
+      "loss": 2.6724,
+      "step": 3268
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.654394299287412e-05,
+      "loss": 3.1989,
+      "step": 3269
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.653800475059383e-05,
+      "loss": 3.1641,
+      "step": 3270
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.653206650831355e-05,
+      "loss": 3.192,
+      "step": 3271
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.652612826603326e-05,
+      "loss": 2.445,
+      "step": 3272
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.652019002375298e-05,
+      "loss": 2.6671,
+      "step": 3273
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.651425178147268e-05,
+      "loss": 2.6082,
+      "step": 3274
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.65083135391924e-05,
+      "loss": 2.6512,
+      "step": 3275
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.650237529691212e-05,
+      "loss": 2.6169,
+      "step": 3276
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.649643705463183e-05,
+      "loss": 3.6169,
+      "step": 3277
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.649049881235155e-05,
+      "loss": 3.3187,
+      "step": 3278
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.648456057007126e-05,
+      "loss": 3.2649,
+      "step": 3279
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.647862232779098e-05,
+      "loss": 3.2419,
+      "step": 3280
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.647268408551069e-05,
+      "loss": 3.2714,
+      "step": 3281
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.64667458432304e-05,
+      "loss": 3.2475,
+      "step": 3282
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.646080760095013e-05,
+      "loss": 3.2959,
+      "step": 3283
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.645486935866985e-05,
+      "loss": 3.3907,
+      "step": 3284
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.644893111638955e-05,
+      "loss": 3.3693,
+      "step": 3285
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.644299287410926e-05,
+      "loss": 3.4156,
+      "step": 3286
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.643705463182898e-05,
+      "loss": 3.3359,
+      "step": 3287
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.643111638954869e-05,
+      "loss": 3.2883,
+      "step": 3288
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.642517814726842e-05,
+      "loss": 3.1075,
+      "step": 3289
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.641923990498813e-05,
+      "loss": 3.2317,
+      "step": 3290
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.641330166270785e-05,
+      "loss": 2.8561,
+      "step": 3291
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.640736342042756e-05,
+      "loss": 3.3517,
+      "step": 3292
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.640142517814728e-05,
+      "loss": 3.154,
+      "step": 3293
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.639548693586698e-05,
+      "loss": 3.1671,
+      "step": 3294
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.63895486935867e-05,
+      "loss": 3.063,
+      "step": 3295
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.638361045130642e-05,
+      "loss": 3.0164,
+      "step": 3296
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.637767220902613e-05,
+      "loss": 3.0601,
+      "step": 3297
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.637173396674585e-05,
+      "loss": 3.2516,
+      "step": 3298
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.636579572446556e-05,
+      "loss": 3.0544,
+      "step": 3299
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.635985748218528e-05,
+      "loss": 2.81,
+      "step": 3300
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.635391923990498e-05,
+      "loss": 3.056,
+      "step": 3301
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.63479809976247e-05,
+      "loss": 3.0986,
+      "step": 3302
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.634204275534443e-05,
+      "loss": 3.4555,
+      "step": 3303
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.633610451306415e-05,
+      "loss": 3.061,
+      "step": 3304
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.633016627078385e-05,
+      "loss": 3.2802,
+      "step": 3305
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.632422802850356e-05,
+      "loss": 2.9572,
+      "step": 3306
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.631828978622328e-05,
+      "loss": 3.1222,
+      "step": 3307
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.631235154394299e-05,
+      "loss": 3.029,
+      "step": 3308
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.630641330166272e-05,
+      "loss": 2.8752,
+      "step": 3309
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.630047505938243e-05,
+      "loss": 2.8594,
+      "step": 3310
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.629453681710215e-05,
+      "loss": 3.0473,
+      "step": 3311
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.628859857482185e-05,
+      "loss": 3.0585,
+      "step": 3312
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.628266033254157e-05,
+      "loss": 3.2896,
+      "step": 3313
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.627672209026128e-05,
+      "loss": 2.9034,
+      "step": 3314
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.6270783847981e-05,
+      "loss": 3.0503,
+      "step": 3315
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.626484560570072e-05,
+      "loss": 2.9783,
+      "step": 3316
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.625890736342043e-05,
+      "loss": 2.826,
+      "step": 3317
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.625296912114015e-05,
+      "loss": 2.9257,
+      "step": 3318
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.624703087885986e-05,
+      "loss": 2.949,
+      "step": 3319
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.624109263657958e-05,
+      "loss": 2.6416,
+      "step": 3320
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.623515439429928e-05,
+      "loss": 2.5738,
+      "step": 3321
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.6229216152019e-05,
+      "loss": 3.2357,
+      "step": 3322
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.622327790973872e-05,
+      "loss": 2.3767,
+      "step": 3323
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.621733966745844e-05,
+      "loss": 2.6625,
+      "step": 3324
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.621140142517815e-05,
+      "loss": 3.0138,
+      "step": 3325
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.620546318289786e-05,
+      "loss": 2.4358,
+      "step": 3326
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.619952494061758e-05,
+      "loss": 3.5266,
+      "step": 3327
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.61935866983373e-05,
+      "loss": 3.4138,
+      "step": 3328
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.618764845605702e-05,
+      "loss": 3.4992,
+      "step": 3329
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.618171021377673e-05,
+      "loss": 3.3488,
+      "step": 3330
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.617577197149645e-05,
+      "loss": 3.226,
+      "step": 3331
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.616983372921615e-05,
+      "loss": 3.1805,
+      "step": 3332
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.616389548693587e-05,
+      "loss": 3.1381,
+      "step": 3333
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.615795724465558e-05,
+      "loss": 3.4073,
+      "step": 3334
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.61520190023753e-05,
+      "loss": 3.1656,
+      "step": 3335
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.614608076009502e-05,
+      "loss": 3.428,
+      "step": 3336
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.614014251781473e-05,
+      "loss": 3.2806,
+      "step": 3337
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.613420427553445e-05,
+      "loss": 3.2801,
+      "step": 3338
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.612826603325415e-05,
+      "loss": 3.2903,
+      "step": 3339
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.612232779097388e-05,
+      "loss": 3.0084,
+      "step": 3340
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.611638954869358e-05,
+      "loss": 3.2485,
+      "step": 3341
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.61104513064133e-05,
+      "loss": 2.9348,
+      "step": 3342
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.610451306413302e-05,
+      "loss": 2.9912,
+      "step": 3343
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.609857482185274e-05,
+      "loss": 2.9561,
+      "step": 3344
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.609263657957245e-05,
+      "loss": 3.094,
+      "step": 3345
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.608669833729217e-05,
+      "loss": 2.9526,
+      "step": 3346
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.608076009501188e-05,
+      "loss": 2.9681,
+      "step": 3347
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.608076009501188e-05,
+      "loss": 2.8836,
+      "step": 3348
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.60748218527316e-05,
+      "loss": 2.8508,
+      "step": 3349
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.606888361045132e-05,
+      "loss": 2.9888,
+      "step": 3350
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.606294536817102e-05,
+      "loss": 3.009,
+      "step": 3351
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.605700712589074e-05,
+      "loss": 3.3546,
+      "step": 3352
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.605106888361045e-05,
+      "loss": 3.0337,
+      "step": 3353
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.604513064133017e-05,
+      "loss": 3.0104,
+      "step": 3354
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.603919239904988e-05,
+      "loss": 2.981,
+      "step": 3355
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.60332541567696e-05,
+      "loss": 2.8737,
+      "step": 3356
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.602731591448932e-05,
+      "loss": 3.2207,
+      "step": 3357
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.602137767220903e-05,
+      "loss": 2.7755,
+      "step": 3358
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.601543942992875e-05,
+      "loss": 2.6883,
+      "step": 3359
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.600950118764845e-05,
+      "loss": 2.8775,
+      "step": 3360
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.600356294536817e-05,
+      "loss": 2.8228,
+      "step": 3361
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.599762470308788e-05,
+      "loss": 3.1126,
+      "step": 3362
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.599168646080761e-05,
+      "loss": 3.0441,
+      "step": 3363
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.598574821852732e-05,
+      "loss": 2.8341,
+      "step": 3364
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.597980997624704e-05,
+      "loss": 2.8948,
+      "step": 3365
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.597387173396675e-05,
+      "loss": 2.9437,
+      "step": 3366
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.596793349168647e-05,
+      "loss": 3.2628,
+      "step": 3367
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.596199524940618e-05,
+      "loss": 2.7974,
+      "step": 3368
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.59560570071259e-05,
+      "loss": 2.8952,
+      "step": 3369
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.595011876484562e-05,
+      "loss": 2.6277,
+      "step": 3370
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.594418052256532e-05,
+      "loss": 2.3717,
+      "step": 3371
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.593824228028504e-05,
+      "loss": 2.3035,
+      "step": 3372
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.593230403800475e-05,
+      "loss": 2.7329,
+      "step": 3373
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.592636579572447e-05,
+      "loss": 2.3283,
+      "step": 3374
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.592042755344418e-05,
+      "loss": 2.4421,
+      "step": 3375
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.59144893111639e-05,
+      "loss": 2.3844,
+      "step": 3376
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.590855106888362e-05,
+      "loss": 3.7217,
+      "step": 3377
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.590261282660334e-05,
+      "loss": 3.5637,
+      "step": 3378
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.589667458432305e-05,
+      "loss": 3.3074,
+      "step": 3379
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.589073634204275e-05,
+      "loss": 3.51,
+      "step": 3380
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.588479809976247e-05,
+      "loss": 3.1286,
+      "step": 3381
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.587885985748218e-05,
+      "loss": 3.2966,
+      "step": 3382
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.587292161520191e-05,
+      "loss": 3.3683,
+      "step": 3383
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.586698337292162e-05,
+      "loss": 3.1211,
+      "step": 3384
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.586104513064134e-05,
+      "loss": 3.5628,
+      "step": 3385
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.585510688836105e-05,
+      "loss": 3.0731,
+      "step": 3386
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.584916864608077e-05,
+      "loss": 3.2853,
+      "step": 3387
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.584323040380047e-05,
+      "loss": 3.0485,
+      "step": 3388
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.58372921615202e-05,
+      "loss": 3.0673,
+      "step": 3389
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.583135391923992e-05,
+      "loss": 2.9513,
+      "step": 3390
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.582541567695962e-05,
+      "loss": 3.0531,
+      "step": 3391
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.581947743467934e-05,
+      "loss": 3.0133,
+      "step": 3392
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.581353919239905e-05,
+      "loss": 2.9639,
+      "step": 3393
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.580760095011877e-05,
+      "loss": 2.8455,
+      "step": 3394
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.580166270783848e-05,
+      "loss": 2.9956,
+      "step": 3395
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.57957244655582e-05,
+      "loss": 3.2842,
+      "step": 3396
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.578978622327792e-05,
+      "loss": 2.8683,
+      "step": 3397
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.578384798099764e-05,
+      "loss": 2.975,
+      "step": 3398
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.577790973871734e-05,
+      "loss": 2.9303,
+      "step": 3399
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.577197149643705e-05,
+      "loss": 2.8766,
+      "step": 3400
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.576603325415677e-05,
+      "loss": 3.0882,
+      "step": 3401
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.576009501187648e-05,
+      "loss": 2.8033,
+      "step": 3402
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.575415676959621e-05,
+      "loss": 2.8837,
+      "step": 3403
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.574821852731592e-05,
+      "loss": 2.6325,
+      "step": 3404
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.574228028503564e-05,
+      "loss": 3.0806,
+      "step": 3405
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.573634204275535e-05,
+      "loss": 2.8872,
+      "step": 3406
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.573040380047507e-05,
+      "loss": 2.8087,
+      "step": 3407
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.572446555819477e-05,
+      "loss": 2.8829,
+      "step": 3408
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.57185273159145e-05,
+      "loss": 2.7788,
+      "step": 3409
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.571258907363421e-05,
+      "loss": 2.7338,
+      "step": 3410
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.570665083135392e-05,
+      "loss": 2.494,
+      "step": 3411
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.570071258907364e-05,
+      "loss": 2.6753,
+      "step": 3412
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.569477434679335e-05,
+      "loss": 2.7749,
+      "step": 3413
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.568883610451307e-05,
+      "loss": 3.0594,
+      "step": 3414
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.568289786223277e-05,
+      "loss": 2.5387,
+      "step": 3415
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.56769596199525e-05,
+      "loss": 2.6433,
+      "step": 3416
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.567102137767222e-05,
+      "loss": 2.4267,
+      "step": 3417
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.566508313539194e-05,
+      "loss": 2.5889,
+      "step": 3418
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.565914489311164e-05,
+      "loss": 2.7328,
+      "step": 3419
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.565320665083136e-05,
+      "loss": 2.5311,
+      "step": 3420
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.564726840855107e-05,
+      "loss": 2.6127,
+      "step": 3421
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.564133016627078e-05,
+      "loss": 2.7248,
+      "step": 3422
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.563539192399051e-05,
+      "loss": 2.8201,
+      "step": 3423
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.562945368171022e-05,
+      "loss": 2.4334,
+      "step": 3424
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.562351543942994e-05,
+      "loss": 2.4673,
+      "step": 3425
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.561757719714964e-05,
+      "loss": 2.1912,
+      "step": 3426
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.561163895486936e-05,
+      "loss": 3.5265,
+      "step": 3427
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.560570071258907e-05,
+      "loss": 3.565,
+      "step": 3428
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.559976247030879e-05,
+      "loss": 3.1929,
+      "step": 3429
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.559382422802851e-05,
+      "loss": 3.1062,
+      "step": 3430
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.558788598574822e-05,
+      "loss": 2.9491,
+      "step": 3431
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.558194774346794e-05,
+      "loss": 3.1582,
+      "step": 3432
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.557600950118765e-05,
+      "loss": 3.0029,
+      "step": 3433
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.557007125890737e-05,
+      "loss": 3.1672,
+      "step": 3434
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.556413301662707e-05,
+      "loss": 3.1855,
+      "step": 3435
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.55581947743468e-05,
+      "loss": 3.0513,
+      "step": 3436
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.555225653206651e-05,
+      "loss": 2.8753,
+      "step": 3437
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.554631828978623e-05,
+      "loss": 3.0546,
+      "step": 3438
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.554038004750594e-05,
+      "loss": 2.9551,
+      "step": 3439
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.553444180522566e-05,
+      "loss": 3.0088,
+      "step": 3440
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.552850356294537e-05,
+      "loss": 2.9132,
+      "step": 3441
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.552256532066509e-05,
+      "loss": 2.9092,
+      "step": 3442
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.551662707838481e-05,
+      "loss": 3.1295,
+      "step": 3443
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.551068883610452e-05,
+      "loss": 2.9819,
+      "step": 3444
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.550475059382424e-05,
+      "loss": 2.8782,
+      "step": 3445
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.549881235154394e-05,
+      "loss": 2.7847,
+      "step": 3446
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.549287410926366e-05,
+      "loss": 2.781,
+      "step": 3447
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.548693586698337e-05,
+      "loss": 2.8067,
+      "step": 3448
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.548099762470309e-05,
+      "loss": 2.8855,
+      "step": 3449
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.547505938242281e-05,
+      "loss": 2.7451,
+      "step": 3450
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.546912114014253e-05,
+      "loss": 2.8269,
+      "step": 3451
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.546318289786224e-05,
+      "loss": 2.6903,
+      "step": 3452
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.545724465558195e-05,
+      "loss": 2.5421,
+      "step": 3453
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.545130641330167e-05,
+      "loss": 2.649,
+      "step": 3454
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.544536817102137e-05,
+      "loss": 3.1045,
+      "step": 3455
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.543942992874109e-05,
+      "loss": 3.2965,
+      "step": 3456
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.543349168646081e-05,
+      "loss": 2.6676,
+      "step": 3457
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.542755344418053e-05,
+      "loss": 2.7134,
+      "step": 3458
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.542161520190024e-05,
+      "loss": 2.5688,
+      "step": 3459
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.541567695961996e-05,
+      "loss": 2.7639,
+      "step": 3460
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.540973871733967e-05,
+      "loss": 2.6758,
+      "step": 3461
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.540380047505939e-05,
+      "loss": 2.6792,
+      "step": 3462
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.539786223277911e-05,
+      "loss": 2.598,
+      "step": 3463
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.539192399049881e-05,
+      "loss": 2.7108,
+      "step": 3464
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.538598574821854e-05,
+      "loss": 2.5259,
+      "step": 3465
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.538004750593824e-05,
+      "loss": 2.5384,
+      "step": 3466
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.537410926365796e-05,
+      "loss": 2.6535,
+      "step": 3467
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.536817102137767e-05,
+      "loss": 2.8111,
+      "step": 3468
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.536223277909739e-05,
+      "loss": 2.2073,
+      "step": 3469
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.535629453681711e-05,
+      "loss": 2.5657,
+      "step": 3470
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.535035629453683e-05,
+      "loss": 2.4401,
+      "step": 3471
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.534441805225654e-05,
+      "loss": 2.3504,
+      "step": 3472
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.533847980997624e-05,
+      "loss": 2.3644,
+      "step": 3473
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.533254156769596e-05,
+      "loss": 2.2042,
+      "step": 3474
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.532660332541567e-05,
+      "loss": 1.9551,
+      "step": 3475
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.532066508313539e-05,
+      "loss": 1.8598,
+      "step": 3476
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.531472684085511e-05,
+      "loss": 4.0322,
+      "step": 3477
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.530878859857483e-05,
+      "loss": 3.7034,
+      "step": 3478
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.530285035629454e-05,
+      "loss": 3.4359,
+      "step": 3479
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.529691211401426e-05,
+      "loss": 3.2901,
+      "step": 3480
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.529097387173397e-05,
+      "loss": 3.1519,
+      "step": 3481
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.528503562945369e-05,
+      "loss": 2.9172,
+      "step": 3482
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.527909738717341e-05,
+      "loss": 2.8183,
+      "step": 3483
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.527315914489311e-05,
+      "loss": 3.2359,
+      "step": 3484
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.526722090261283e-05,
+      "loss": 3.1471,
+      "step": 3485
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.526128266033254e-05,
+      "loss": 2.9599,
+      "step": 3486
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.525534441805226e-05,
+      "loss": 3.1736,
+      "step": 3487
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.524940617577197e-05,
+      "loss": 2.917,
+      "step": 3488
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.524346793349169e-05,
+      "loss": 2.897,
+      "step": 3489
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.523752969121141e-05,
+      "loss": 3.3828,
+      "step": 3490
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.523159144893113e-05,
+      "loss": 3.0808,
+      "step": 3491
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.522565320665084e-05,
+      "loss": 3.0148,
+      "step": 3492
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.521971496437054e-05,
+      "loss": 3.0375,
+      "step": 3493
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.521377672209026e-05,
+      "loss": 2.7356,
+      "step": 3494
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.520783847980997e-05,
+      "loss": 2.8226,
+      "step": 3495
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.52019002375297e-05,
+      "loss": 2.6743,
+      "step": 3496
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.519596199524941e-05,
+      "loss": 2.4098,
+      "step": 3497
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.519002375296913e-05,
+      "loss": 2.793,
+      "step": 3498
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.518408551068884e-05,
+      "loss": 2.7424,
+      "step": 3499
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.517814726840856e-05,
+      "loss": 2.7961,
+      "step": 3500
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.517220902612826e-05,
+      "loss": 2.6919,
+      "step": 3501
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.516627078384799e-05,
+      "loss": 2.8772,
+      "step": 3502
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.51603325415677e-05,
+      "loss": 2.6462,
+      "step": 3503
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.515439429928741e-05,
+      "loss": 2.5027,
+      "step": 3504
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.514845605700713e-05,
+      "loss": 2.6591,
+      "step": 3505
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.514251781472684e-05,
+      "loss": 2.6642,
+      "step": 3506
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.513657957244656e-05,
+      "loss": 2.6106,
+      "step": 3507
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.513064133016627e-05,
+      "loss": 2.4673,
+      "step": 3508
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.512470308788599e-05,
+      "loss": 2.4694,
+      "step": 3509
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.511876484560571e-05,
+      "loss": 2.7382,
+      "step": 3510
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.511282660332543e-05,
+      "loss": 2.6261,
+      "step": 3511
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.510688836104513e-05,
+      "loss": 2.4733,
+      "step": 3512
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.510095011876485e-05,
+      "loss": 2.5531,
+      "step": 3513
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.509501187648456e-05,
+      "loss": 2.3398,
+      "step": 3514
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.508907363420427e-05,
+      "loss": 2.7515,
+      "step": 3515
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.5083135391924e-05,
+      "loss": 2.4227,
+      "step": 3516
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.507719714964371e-05,
+      "loss": 2.4204,
+      "step": 3517
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.507125890736343e-05,
+      "loss": 2.4116,
+      "step": 3518
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.506532066508314e-05,
+      "loss": 2.4377,
+      "step": 3519
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.505938242280286e-05,
+      "loss": 2.207,
+      "step": 3520
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.505344418052256e-05,
+      "loss": 2.2748,
+      "step": 3521
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.504750593824228e-05,
+      "loss": 2.1876,
+      "step": 3522
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.5041567695962e-05,
+      "loss": 1.8954,
+      "step": 3523
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.503562945368171e-05,
+      "loss": 1.9216,
+      "step": 3524
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.502969121140143e-05,
+      "loss": 1.8038,
+      "step": 3525
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.502375296912114e-05,
+      "loss": 1.9483,
+      "step": 3526
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.501781472684086e-05,
+      "loss": 3.7875,
+      "step": 3527
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.501187648456057e-05,
+      "loss": 3.1874,
+      "step": 3528
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.500593824228029e-05,
+      "loss": 3.1134,
+      "step": 3529
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.5e-05,
+      "loss": 3.2084,
+      "step": 3530
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.499406175771973e-05,
+      "loss": 2.7196,
+      "step": 3531
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.498812351543943e-05,
+      "loss": 2.7955,
+      "step": 3532
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.498218527315915e-05,
+      "loss": 3.0201,
+      "step": 3533
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.497624703087886e-05,
+      "loss": 2.8766,
+      "step": 3534
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.497030878859857e-05,
+      "loss": 2.9001,
+      "step": 3535
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.49643705463183e-05,
+      "loss": 2.8819,
+      "step": 3536
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.495843230403801e-05,
+      "loss": 2.832,
+      "step": 3537
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.495249406175773e-05,
+      "loss": 2.7069,
+      "step": 3538
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.494655581947744e-05,
+      "loss": 2.7505,
+      "step": 3539
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.494061757719716e-05,
+      "loss": 2.5129,
+      "step": 3540
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.493467933491686e-05,
+      "loss": 2.4902,
+      "step": 3541
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.492874109263658e-05,
+      "loss": 2.7427,
+      "step": 3542
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.49228028503563e-05,
+      "loss": 2.893,
+      "step": 3543
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.491686460807602e-05,
+      "loss": 2.694,
+      "step": 3544
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.491092636579573e-05,
+      "loss": 2.7305,
+      "step": 3545
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.490498812351544e-05,
+      "loss": 2.6622,
+      "step": 3546
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.489904988123516e-05,
+      "loss": 2.5383,
+      "step": 3547
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.489311163895486e-05,
+      "loss": 2.4428,
+      "step": 3548
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.488717339667458e-05,
+      "loss": 2.5852,
+      "step": 3549
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.48812351543943e-05,
+      "loss": 2.4318,
+      "step": 3550
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.487529691211403e-05,
+      "loss": 2.7357,
+      "step": 3551
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.486935866983373e-05,
+      "loss": 2.865,
+      "step": 3552
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.486342042755345e-05,
+      "loss": 2.3868,
+      "step": 3553
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.485748218527316e-05,
+      "loss": 2.5475,
+      "step": 3554
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.485154394299287e-05,
+      "loss": 2.5365,
+      "step": 3555
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.48456057007126e-05,
+      "loss": 2.4199,
+      "step": 3556
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.48396674584323e-05,
+      "loss": 2.378,
+      "step": 3557
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.483372921615203e-05,
+      "loss": 2.1603,
+      "step": 3558
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.482779097387173e-05,
+      "loss": 2.1863,
+      "step": 3559
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.482185273159145e-05,
+      "loss": 2.3804,
+      "step": 3560
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.481591448931116e-05,
+      "loss": 2.4639,
+      "step": 3561
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.480997624703088e-05,
+      "loss": 2.297,
+      "step": 3562
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.48040380047506e-05,
+      "loss": 1.9272,
+      "step": 3563
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.479809976247032e-05,
+      "loss": 2.2299,
+      "step": 3564
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.479216152019003e-05,
+      "loss": 2.8037,
+      "step": 3565
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.478622327790974e-05,
+      "loss": 1.9726,
+      "step": 3566
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.478028503562946e-05,
+      "loss": 1.9293,
+      "step": 3567
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.477434679334916e-05,
+      "loss": 1.6995,
+      "step": 3568
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.476840855106888e-05,
+      "loss": 3.3296,
+      "step": 3569
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.47624703087886e-05,
+      "loss": 3.066,
+      "step": 3570
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.475653206650832e-05,
+      "loss": 2.6969,
+      "step": 3571
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.475059382422803e-05,
+      "loss": 2.6551,
+      "step": 3572
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.474465558194775e-05,
+      "loss": 2.5139,
+      "step": 3573
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.473871733966746e-05,
+      "loss": 2.5065,
+      "step": 3574
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.473277909738718e-05,
+      "loss": 2.4086,
+      "step": 3575
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.47268408551069e-05,
+      "loss": 2.4127,
+      "step": 3576
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.47209026128266e-05,
+      "loss": 2.4001,
+      "step": 3577
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.471496437054633e-05,
+      "loss": 2.2814,
+      "step": 3578
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.470902612826603e-05,
+      "loss": 2.226,
+      "step": 3579
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.470308788598575e-05,
+      "loss": 2.1747,
+      "step": 3580
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 8.469714964370546e-05,
+      "loss": 2.2762,
+      "step": 3581
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.469121140142518e-05,
+      "loss": 2.2762,
+      "step": 3582
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.46852731591449e-05,
+      "loss": 2.4042,
+      "step": 3583
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.467933491686462e-05,
+      "loss": 2.3795,
+      "step": 3584
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.467339667458433e-05,
+      "loss": 2.356,
+      "step": 3585
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.466745843230403e-05,
+      "loss": 2.341,
+      "step": 3586
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.466152019002375e-05,
+      "loss": 2.1252,
+      "step": 3587
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.465558194774346e-05,
+      "loss": 2.0263,
+      "step": 3588
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.464964370546318e-05,
+      "loss": 2.22,
+      "step": 3589
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.46437054631829e-05,
+      "loss": 2.2887,
+      "step": 3590
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 8.463776722090262e-05,
+      "loss": 2.2493,
+      "step": 3591
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 8.463182897862233e-05,
+      "loss": 2.1681,
+      "step": 3592
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 8.462589073634205e-05,
+      "loss": 2.0634,
+      "step": 3593
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 8.461995249406176e-05,
+      "loss": 1.9875,
+      "step": 3594
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 8.461401425178148e-05,
+      "loss": 2.0823,
+      "step": 3595
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 8.46080760095012e-05,
+      "loss": 1.8606,
+      "step": 3596
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 8.46021377672209e-05,
+      "loss": 2.3018,
+      "step": 3597
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 8.459619952494062e-05,
+      "loss": 2.0987,
+      "step": 3598
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 8.459026128266033e-05,
+      "loss": 2.0448,
+      "step": 3599
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.458432304038005e-05,
+      "loss": 2.1432,
+      "step": 3600
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.457838479809976e-05,
+      "loss": 2.0613,
+      "step": 3601
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.457244655581948e-05,
+      "loss": 2.1262,
+      "step": 3602
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.45665083135392e-05,
+      "loss": 2.1267,
+      "step": 3603
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.456057007125892e-05,
+      "loss": 2.0241,
+      "step": 3604
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.455463182897863e-05,
+      "loss": 1.8819,
+      "step": 3605
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.454869358669835e-05,
+      "loss": 1.9084,
+      "step": 3606
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.454275534441805e-05,
+      "loss": 1.8479,
+      "step": 3607
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.453681710213776e-05,
+      "loss": 1.9602,
+      "step": 3608
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.45308788598575e-05,
+      "loss": 1.6797,
+      "step": 3609
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.45249406175772e-05,
+      "loss": 2.0296,
+      "step": 3610
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.451900237529692e-05,
+      "loss": 1.6795,
+      "step": 3611
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.451306413301663e-05,
+      "loss": 1.6757,
+      "step": 3612
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.450712589073635e-05,
+      "loss": 1.6244,
+      "step": 3613
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.450118764845606e-05,
+      "loss": 1.7553,
+      "step": 3614
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.449524940617578e-05,
+      "loss": 1.4149,
+      "step": 3615
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.44893111638955e-05,
+      "loss": 1.4585,
+      "step": 3616
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.448337292161522e-05,
+      "loss": 1.5718,
+      "step": 3617
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 8.447743467933492e-05,
+      "loss": 1.1665,
+      "step": 3618
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 8.447149643705463e-05,
+      "loss": 3.594,
+      "step": 3619
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 8.446555819477435e-05,
+      "loss": 3.8226,
+      "step": 3620
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 8.445961995249406e-05,
+      "loss": 3.5529,
+      "step": 3621
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 8.445368171021378e-05,
+      "loss": 2.9142,
+      "step": 3622
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 8.44477434679335e-05,
+      "loss": 2.4766,
+      "step": 3623
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 8.444180522565322e-05,
+      "loss": 2.3595,
+      "step": 3624
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 8.443586698337293e-05,
+      "loss": 2.3751,
+      "step": 3625
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 8.442992874109265e-05,
+      "loss": 2.3434,
+      "step": 3626
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 8.442399049881235e-05,
+      "loss": 2.7149,
+      "step": 3627
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 8.441805225653206e-05,
+      "loss": 2.5439,
+      "step": 3628
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 8.441211401425179e-05,
+      "loss": 2.3255,
+      "step": 3629
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 8.44061757719715e-05,
+      "loss": 2.3972,
+      "step": 3630
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 8.440023752969122e-05,
+      "loss": 2.1868,
+      "step": 3631
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 8.439429928741093e-05,
+      "loss": 2.8473,
+      "step": 3632
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 8.438836104513065e-05,
+      "loss": 2.2593,
+      "step": 3633
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 8.438242280285035e-05,
+      "loss": 2.3058,
+      "step": 3634
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 8.437648456057007e-05,
+      "loss": 2.1439,
+      "step": 3635
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 8.43705463182898e-05,
+      "loss": 2.0612,
+      "step": 3636
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 8.436460807600952e-05,
+      "loss": 2.1662,
+      "step": 3637
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 8.435866983372922e-05,
+      "loss": 2.0582,
+      "step": 3638
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 8.435273159144893e-05,
+      "loss": 2.0944,
+      "step": 3639
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 8.434679334916865e-05,
+      "loss": 2.1032,
+      "step": 3640
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 8.434085510688836e-05,
+      "loss": 2.2207,
+      "step": 3641
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 8.433491686460808e-05,
+      "loss": 2.0573,
+      "step": 3642
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 8.43289786223278e-05,
+      "loss": 2.0047,
+      "step": 3643
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 8.432304038004752e-05,
+      "loss": 2.0163,
+      "step": 3644
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 8.431710213776722e-05,
+      "loss": 1.8925,
+      "step": 3645
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 8.431116389548694e-05,
+      "loss": 1.964,
+      "step": 3646
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 8.430522565320665e-05,
+      "loss": 2.0601,
+      "step": 3647
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 8.429928741092637e-05,
+      "loss": 1.8932,
+      "step": 3648
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 8.429334916864609e-05,
+      "loss": 1.8688,
+      "step": 3649
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 8.42874109263658e-05,
+      "loss": 2.0251,
+      "step": 3650
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 8.428147268408552e-05,
+      "loss": 1.9371,
+      "step": 3651
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 8.427553444180523e-05,
+      "loss": 1.9051,
+      "step": 3652
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.426959619952495e-05,
+      "loss": 1.819,
+      "step": 3653
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.426365795724465e-05,
+      "loss": 1.9104,
+      "step": 3654
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.425771971496437e-05,
+      "loss": 1.7251,
+      "step": 3655
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.42517814726841e-05,
+      "loss": 1.9386,
+      "step": 3656
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.424584323040381e-05,
+      "loss": 1.8045,
+      "step": 3657
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.423990498812352e-05,
+      "loss": 1.7382,
+      "step": 3658
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.423396674584323e-05,
+      "loss": 1.7987,
+      "step": 3659
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.422802850356295e-05,
+      "loss": 1.8596,
+      "step": 3660
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.422209026128265e-05,
+      "loss": 1.7146,
+      "step": 3661
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.421615201900237e-05,
+      "loss": 1.673,
+      "step": 3662
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.42102137767221e-05,
+      "loss": 1.6514,
+      "step": 3663
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.420427553444182e-05,
+      "loss": 1.3969,
+      "step": 3664
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.419833729216152e-05,
+      "loss": 1.3225,
+      "step": 3665
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.419239904988124e-05,
+      "loss": 1.3385,
+      "step": 3666
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.418646080760095e-05,
+      "loss": 1.5055,
+      "step": 3667
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.418052256532067e-05,
+      "loss": 1.2312,
+      "step": 3668
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.417458432304039e-05,
+      "loss": 3.6112,
+      "step": 3669
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.41686460807601e-05,
+      "loss": 3.5603,
+      "step": 3670
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 8.416270783847982e-05,
+      "loss": 3.4031,
+      "step": 3671
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 8.415676959619952e-05,
+      "loss": 2.8445,
+      "step": 3672
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 8.415083135391924e-05,
+      "loss": 2.5218,
+      "step": 3673
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 8.414489311163895e-05,
+      "loss": 2.3981,
+      "step": 3674
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 8.413895486935867e-05,
+      "loss": 2.2556,
+      "step": 3675
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 8.413301662707839e-05,
+      "loss": 2.264,
+      "step": 3676
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 8.412707838479811e-05,
+      "loss": 2.3136,
+      "step": 3677
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 8.412114014251782e-05,
+      "loss": 2.3497,
+      "step": 3678
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 8.411520190023754e-05,
+      "loss": 2.467,
+      "step": 3679
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 8.410926365795725e-05,
+      "loss": 2.2849,
+      "step": 3680
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 8.410332541567695e-05,
+      "loss": 2.1293,
+      "step": 3681
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 8.409738717339667e-05,
+      "loss": 2.1905,
+      "step": 3682
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 8.40914489311164e-05,
+      "loss": 2.1005,
+      "step": 3683
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 8.408551068883611e-05,
+      "loss": 2.0174,
+      "step": 3684
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 8.407957244655582e-05,
+      "loss": 1.9678,
+      "step": 3685
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 8.407363420427554e-05,
+      "loss": 2.2681,
+      "step": 3686
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 8.406769596199525e-05,
+      "loss": 2.2417,
+      "step": 3687
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 8.406175771971497e-05,
+      "loss": 2.133,
+      "step": 3688
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 8.405581947743469e-05,
+      "loss": 1.9427,
+      "step": 3689
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 8.40498812351544e-05,
+      "loss": 1.8322,
+      "step": 3690
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 8.404394299287412e-05,
+      "loss": 2.2098,
+      "step": 3691
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 8.403800475059382e-05,
+      "loss": 1.9062,
+      "step": 3692
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 8.403206650831354e-05,
+      "loss": 1.9425,
+      "step": 3693
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 8.402612826603325e-05,
+      "loss": 1.9014,
+      "step": 3694
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 8.402019002375297e-05,
+      "loss": 1.8605,
+      "step": 3695
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 8.401425178147269e-05,
+      "loss": 1.8636,
+      "step": 3696
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 8.400831353919241e-05,
+      "loss": 1.7512,
+      "step": 3697
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 8.400237529691212e-05,
+      "loss": 2.2626,
+      "step": 3698
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 8.399643705463184e-05,
+      "loss": 2.1561,
+      "step": 3699
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 8.399049881235155e-05,
+      "loss": 1.7969,
+      "step": 3700
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 8.398456057007125e-05,
+      "loss": 1.9885,
+      "step": 3701
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 8.397862232779097e-05,
+      "loss": 1.9214,
+      "step": 3702
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 8.397268408551069e-05,
+      "loss": 1.8754,
+      "step": 3703
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 8.396674584323041e-05,
+      "loss": 1.6909,
+      "step": 3704
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 8.396080760095012e-05,
+      "loss": 2.1062,
+      "step": 3705
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 8.395486935866984e-05,
+      "loss": 1.7895,
+      "step": 3706
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 8.394893111638955e-05,
+      "loss": 1.893,
+      "step": 3707
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 8.394299287410927e-05,
+      "loss": 2.049,
+      "step": 3708
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 8.393705463182899e-05,
+      "loss": 1.5912,
+      "step": 3709
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 8.393111638954871e-05,
+      "loss": 1.8375,
+      "step": 3710
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 8.392517814726841e-05,
+      "loss": 1.5697,
+      "step": 3711
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 8.391923990498812e-05,
+      "loss": 1.7315,
+      "step": 3712
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 8.391330166270784e-05,
+      "loss": 1.6625,
+      "step": 3713
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 8.390736342042755e-05,
+      "loss": 1.5278,
+      "step": 3714
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 8.390142517814727e-05,
+      "loss": 1.2992,
+      "step": 3715
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 8.389548693586699e-05,
+      "loss": 1.1125,
+      "step": 3716
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 8.388954869358671e-05,
+      "loss": 1.3093,
+      "step": 3717
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 8.388361045130642e-05,
+      "loss": 1.0731,
+      "step": 3718
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 8.387767220902614e-05,
+      "loss": 3.462,
+      "step": 3719
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 8.387173396674584e-05,
+      "loss": 3.4051,
+      "step": 3720
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 8.386579572446555e-05,
+      "loss": 3.1275,
+      "step": 3721
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 8.385985748218527e-05,
+      "loss": 2.8619,
+      "step": 3722
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 8.385391923990499e-05,
+      "loss": 2.5321,
+      "step": 3723
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 8.384798099762471e-05,
+      "loss": 2.2237,
+      "step": 3724
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.384204275534442e-05,
+      "loss": 2.1503,
+      "step": 3725
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.383610451306414e-05,
+      "loss": 2.3759,
+      "step": 3726
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.383016627078385e-05,
+      "loss": 2.1062,
+      "step": 3727
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.382422802850357e-05,
+      "loss": 2.1574,
+      "step": 3728
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.381828978622329e-05,
+      "loss": 2.3821,
+      "step": 3729
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.381235154394301e-05,
+      "loss": 2.0678,
+      "step": 3730
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.380641330166271e-05,
+      "loss": 2.0373,
+      "step": 3731
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.380047505938242e-05,
+      "loss": 2.0002,
+      "step": 3732
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.379453681710214e-05,
+      "loss": 1.9783,
+      "step": 3733
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 8.378859857482185e-05,
+      "loss": 2.0271,
+      "step": 3734
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 8.378266033254157e-05,
+      "loss": 1.8981,
+      "step": 3735
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 8.377672209026129e-05,
+      "loss": 1.9036,
+      "step": 3736
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 8.377078384798101e-05,
+      "loss": 2.0282,
+      "step": 3737
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 8.376484560570072e-05,
+      "loss": 1.9152,
+      "step": 3738
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 8.375890736342044e-05,
+      "loss": 2.1601,
+      "step": 3739
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 8.375296912114014e-05,
+      "loss": 1.8665,
+      "step": 3740
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 8.374703087885986e-05,
+      "loss": 1.8305,
+      "step": 3741
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 8.374109263657958e-05,
+      "loss": 1.7504,
+      "step": 3742
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 8.373515439429929e-05,
+      "loss": 1.978,
+      "step": 3743
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 8.372921615201901e-05,
+      "loss": 1.8365,
+      "step": 3744
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 8.372327790973872e-05,
+      "loss": 1.7278,
+      "step": 3745
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 8.371733966745844e-05,
+      "loss": 1.5899,
+      "step": 3746
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 8.371140142517814e-05,
+      "loss": 1.6851,
+      "step": 3747
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 8.370546318289786e-05,
+      "loss": 1.8472,
+      "step": 3748
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 8.369952494061759e-05,
+      "loss": 1.5773,
+      "step": 3749
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 8.36935866983373e-05,
+      "loss": 1.8397,
+      "step": 3750
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 8.368764845605701e-05,
+      "loss": 1.6575,
+      "step": 3751
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 8.368171021377672e-05,
+      "loss": 1.9422,
+      "step": 3752
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 8.367577197149644e-05,
+      "loss": 1.6617,
+      "step": 3753
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 8.366983372921615e-05,
+      "loss": 1.7291,
+      "step": 3754
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 8.366389548693587e-05,
+      "loss": 1.661,
+      "step": 3755
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 8.365795724465559e-05,
+      "loss": 1.6038,
+      "step": 3756
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 8.365201900237531e-05,
+      "loss": 1.661,
+      "step": 3757
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 8.364608076009501e-05,
+      "loss": 1.6616,
+      "step": 3758
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 8.364014251781473e-05,
+      "loss": 1.5505,
+      "step": 3759
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 8.363420427553444e-05,
+      "loss": 1.7467,
+      "step": 3760
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 8.362826603325416e-05,
+      "loss": 1.4885,
+      "step": 3761
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 8.362232779097388e-05,
+      "loss": 1.474,
+      "step": 3762
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 8.361638954869359e-05,
+      "loss": 1.3272,
+      "step": 3763
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 8.361045130641331e-05,
+      "loss": 1.6834,
+      "step": 3764
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 8.360451306413302e-05,
+      "loss": 1.3076,
+      "step": 3765
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 8.359857482185274e-05,
+      "loss": 1.1224,
+      "step": 3766
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 8.359263657957244e-05,
+      "loss": 1.2827,
+      "step": 3767
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 8.358669833729216e-05,
+      "loss": 2.2044,
+      "step": 3768
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 8.358076009501188e-05,
+      "loss": 3.3575,
+      "step": 3769
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 8.35748218527316e-05,
+      "loss": 3.1829,
+      "step": 3770
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 8.356888361045131e-05,
+      "loss": 2.8463,
+      "step": 3771
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 8.356294536817103e-05,
+      "loss": 2.491,
+      "step": 3772
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 8.355700712589074e-05,
+      "loss": 2.1462,
+      "step": 3773
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 8.355106888361045e-05,
+      "loss": 1.8655,
+      "step": 3774
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 8.354513064133017e-05,
+      "loss": 2.3326,
+      "step": 3775
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 8.353919239904989e-05,
+      "loss": 2.2516,
+      "step": 3776
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 8.35332541567696e-05,
+      "loss": 2.1294,
+      "step": 3777
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 8.352731591448931e-05,
+      "loss": 2.3087,
+      "step": 3778
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 8.352137767220903e-05,
+      "loss": 1.8496,
+      "step": 3779
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 8.351543942992874e-05,
+      "loss": 1.9878,
+      "step": 3780
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 8.350950118764846e-05,
+      "loss": 2.1749,
+      "step": 3781
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 8.350356294536818e-05,
+      "loss": 1.9209,
+      "step": 3782
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 8.34976247030879e-05,
+      "loss": 2.0568,
+      "step": 3783
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 8.349168646080761e-05,
+      "loss": 1.8877,
+      "step": 3784
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 8.348574821852731e-05,
+      "loss": 2.0182,
+      "step": 3785
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 8.347980997624704e-05,
+      "loss": 1.855,
+      "step": 3786
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 8.347387173396674e-05,
+      "loss": 1.8689,
+      "step": 3787
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 8.346793349168646e-05,
+      "loss": 2.0135,
+      "step": 3788
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 8.346199524940618e-05,
+      "loss": 1.9146,
+      "step": 3789
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 8.34560570071259e-05,
+      "loss": 1.7468,
+      "step": 3790
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 8.345011876484561e-05,
+      "loss": 1.8061,
+      "step": 3791
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 8.344418052256533e-05,
+      "loss": 1.8607,
+      "step": 3792
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 8.343824228028504e-05,
+      "loss": 1.735,
+      "step": 3793
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 8.343230403800474e-05,
+      "loss": 1.6521,
+      "step": 3794
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 8.342636579572446e-05,
+      "loss": 1.7925,
+      "step": 3795
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 8.342042755344418e-05,
+      "loss": 1.6962,
+      "step": 3796
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 8.34144893111639e-05,
+      "loss": 1.6563,
+      "step": 3797
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 8.340855106888361e-05,
+      "loss": 1.4645,
+      "step": 3798
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 8.340261282660333e-05,
+      "loss": 1.6394,
+      "step": 3799
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 8.339667458432304e-05,
+      "loss": 1.5934,
+      "step": 3800
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 8.339073634204276e-05,
+      "loss": 1.7526,
+      "step": 3801
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 8.338479809976248e-05,
+      "loss": 1.6875,
+      "step": 3802
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 8.33788598574822e-05,
+      "loss": 1.495,
+      "step": 3803
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 8.33729216152019e-05,
+      "loss": 1.6174,
+      "step": 3804
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 8.336698337292161e-05,
+      "loss": 1.5604,
+      "step": 3805
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 8.336104513064133e-05,
+      "loss": 1.4845,
+      "step": 3806
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 8.335510688836104e-05,
+      "loss": 1.4889,
+      "step": 3807
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 8.334916864608076e-05,
+      "loss": 1.6483,
+      "step": 3808
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 8.334323040380048e-05,
+      "loss": 1.3597,
+      "step": 3809
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 8.33372921615202e-05,
+      "loss": 1.7573,
+      "step": 3810
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 8.333135391923991e-05,
+      "loss": 1.2371,
+      "step": 3811
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 8.332541567695963e-05,
+      "loss": 1.6047,
+      "step": 3812
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 8.331947743467934e-05,
+      "loss": 1.2804,
+      "step": 3813
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 8.331353919239906e-05,
+      "loss": 1.284,
+      "step": 3814
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 8.330760095011876e-05,
+      "loss": 1.8037,
+      "step": 3815
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 8.330166270783848e-05,
+      "loss": 1.5073,
+      "step": 3816
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 8.32957244655582e-05,
+      "loss": 1.2785,
+      "step": 3817
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 8.328978622327791e-05,
+      "loss": 1.6529,
+      "step": 3818
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 8.328384798099763e-05,
+      "loss": 3.3913,
+      "step": 3819
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 8.327790973871734e-05,
+      "loss": 3.2491,
+      "step": 3820
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 8.327197149643706e-05,
+      "loss": 2.8234,
+      "step": 3821
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 8.326603325415678e-05,
+      "loss": 2.5499,
+      "step": 3822
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 8.32600950118765e-05,
+      "loss": 2.2854,
+      "step": 3823
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 8.32541567695962e-05,
+      "loss": 1.8555,
+      "step": 3824
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 8.324821852731591e-05,
+      "loss": 1.7194,
+      "step": 3825
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 8.324228028503563e-05,
+      "loss": 2.103,
+      "step": 3826
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 8.323634204275534e-05,
+      "loss": 1.9489,
+      "step": 3827
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 8.323040380047506e-05,
+      "loss": 2.0129,
+      "step": 3828
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 8.322446555819478e-05,
+      "loss": 1.8644,
+      "step": 3829
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 8.32185273159145e-05,
+      "loss": 1.9728,
+      "step": 3830
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 8.321258907363421e-05,
+      "loss": 1.8964,
+      "step": 3831
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 8.320665083135393e-05,
+      "loss": 2.0973,
+      "step": 3832
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 8.320071258907363e-05,
+      "loss": 1.8849,
+      "step": 3833
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 8.319477434679335e-05,
+      "loss": 1.8335,
+      "step": 3834
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 8.318883610451306e-05,
+      "loss": 1.855,
+      "step": 3835
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 8.318289786223278e-05,
+      "loss": 1.7426,
+      "step": 3836
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 8.31769596199525e-05,
+      "loss": 1.6763,
+      "step": 3837
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 8.317102137767221e-05,
+      "loss": 1.8054,
+      "step": 3838
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 8.316508313539193e-05,
+      "loss": 1.6979,
+      "step": 3839
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 8.315914489311164e-05,
+      "loss": 1.759,
+      "step": 3840
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 8.315320665083136e-05,
+      "loss": 1.6623,
+      "step": 3841
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 8.314726840855108e-05,
+      "loss": 1.5633,
+      "step": 3842
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 8.31413301662708e-05,
+      "loss": 1.5895,
+      "step": 3843
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 8.31353919239905e-05,
+      "loss": 1.7404,
+      "step": 3844
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 8.312945368171022e-05,
+      "loss": 1.5476,
+      "step": 3845
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 8.312351543942993e-05,
+      "loss": 1.6571,
+      "step": 3846
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 8.311757719714964e-05,
+      "loss": 1.6152,
+      "step": 3847
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 8.311163895486936e-05,
+      "loss": 1.5446,
+      "step": 3848
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 8.310570071258908e-05,
+      "loss": 1.6346,
+      "step": 3849
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 8.30997624703088e-05,
+      "loss": 1.6918,
+      "step": 3850
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 8.30938242280285e-05,
+      "loss": 1.5221,
+      "step": 3851
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 8.308788598574823e-05,
+      "loss": 1.621,
+      "step": 3852
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 8.308194774346793e-05,
+      "loss": 1.7346,
+      "step": 3853
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 8.307600950118765e-05,
+      "loss": 1.4957,
+      "step": 3854
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 8.307007125890737e-05,
+      "loss": 1.4177,
+      "step": 3855
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 8.306413301662708e-05,
+      "loss": 1.4858,
+      "step": 3856
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 8.30581947743468e-05,
+      "loss": 1.5131,
+      "step": 3857
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 8.305225653206651e-05,
+      "loss": 1.5436,
+      "step": 3858
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 8.304631828978623e-05,
+      "loss": 1.5086,
+      "step": 3859
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 8.304038004750593e-05,
+      "loss": 1.4363,
+      "step": 3860
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 8.303444180522566e-05,
+      "loss": 1.4146,
+      "step": 3861
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 8.302850356294538e-05,
+      "loss": 1.2302,
+      "step": 3862
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 8.30225653206651e-05,
+      "loss": 1.1758,
+      "step": 3863
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 8.30166270783848e-05,
+      "loss": 1.0912,
+      "step": 3864
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 8.301068883610452e-05,
+      "loss": 2.185,
+      "step": 3865
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 8.300475059382423e-05,
+      "loss": 1.4477,
+      "step": 3866
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.299881235154394e-05,
+      "loss": 2.6773,
+      "step": 3867
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.299287410926366e-05,
+      "loss": 1.7643,
+      "step": 3868
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.298693586698338e-05,
+      "loss": 3.5387,
+      "step": 3869
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.29809976247031e-05,
+      "loss": 3.6048,
+      "step": 3870
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.29750593824228e-05,
+      "loss": 3.0913,
+      "step": 3871
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.296912114014253e-05,
+      "loss": 2.5274,
+      "step": 3872
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.296318289786223e-05,
+      "loss": 2.4207,
+      "step": 3873
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.295724465558195e-05,
+      "loss": 2.0149,
+      "step": 3874
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.295130641330167e-05,
+      "loss": 2.0949,
+      "step": 3875
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.294536817102139e-05,
+      "loss": 2.0017,
+      "step": 3876
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.29394299287411e-05,
+      "loss": 2.1804,
+      "step": 3877
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.29334916864608e-05,
+      "loss": 2.0709,
+      "step": 3878
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.292755344418053e-05,
+      "loss": 1.7946,
+      "step": 3879
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.292161520190023e-05,
+      "loss": 1.9821,
+      "step": 3880
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.291567695961995e-05,
+      "loss": 1.8254,
+      "step": 3881
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.290973871733967e-05,
+      "loss": 1.6002,
+      "step": 3882
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.29038004750594e-05,
+      "loss": 1.8563,
+      "step": 3883
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.28978622327791e-05,
+      "loss": 1.6734,
+      "step": 3884
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.289192399049882e-05,
+      "loss": 1.5325,
+      "step": 3885
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.288598574821853e-05,
+      "loss": 1.621,
+      "step": 3886
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.288004750593824e-05,
+      "loss": 1.4872,
+      "step": 3887
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.287410926365796e-05,
+      "loss": 2.1987,
+      "step": 3888
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.286817102137768e-05,
+      "loss": 1.4773,
+      "step": 3889
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.28622327790974e-05,
+      "loss": 1.6265,
+      "step": 3890
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.28562945368171e-05,
+      "loss": 1.7273,
+      "step": 3891
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.285035629453682e-05,
+      "loss": 1.7304,
+      "step": 3892
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.284441805225653e-05,
+      "loss": 1.6905,
+      "step": 3893
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.283847980997625e-05,
+      "loss": 1.6146,
+      "step": 3894
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.283254156769597e-05,
+      "loss": 1.4823,
+      "step": 3895
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.282660332541569e-05,
+      "loss": 1.4238,
+      "step": 3896
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.28206650831354e-05,
+      "loss": 1.5034,
+      "step": 3897
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.28147268408551e-05,
+      "loss": 1.6857,
+      "step": 3898
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.280878859857483e-05,
+      "loss": 1.5692,
+      "step": 3899
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.280285035629453e-05,
+      "loss": 1.3009,
+      "step": 3900
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.279691211401425e-05,
+      "loss": 1.4508,
+      "step": 3901
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.279097387173397e-05,
+      "loss": 1.4575,
+      "step": 3902
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.27850356294537e-05,
+      "loss": 1.796,
+      "step": 3903
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.27790973871734e-05,
+      "loss": 1.4679,
+      "step": 3904
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.277315914489312e-05,
+      "loss": 1.4487,
+      "step": 3905
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.276722090261283e-05,
+      "loss": 1.3559,
+      "step": 3906
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.276128266033255e-05,
+      "loss": 1.2385,
+      "step": 3907
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.275534441805225e-05,
+      "loss": 1.4226,
+      "step": 3908
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.274940617577197e-05,
+      "loss": 1.2795,
+      "step": 3909
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.27434679334917e-05,
+      "loss": 1.5988,
+      "step": 3910
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.27375296912114e-05,
+      "loss": 1.5637,
+      "step": 3911
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.273159144893112e-05,
+      "loss": 1.3669,
+      "step": 3912
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.272565320665083e-05,
+      "loss": 1.1059,
+      "step": 3913
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.271971496437055e-05,
+      "loss": 1.2343,
+      "step": 3914
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.271377672209027e-05,
+      "loss": 1.2383,
+      "step": 3915
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.270783847980999e-05,
+      "loss": 1.2383,
+      "step": 3916
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.27019002375297e-05,
+      "loss": 1.1129,
+      "step": 3917
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.26959619952494e-05,
+      "loss": 1.1538,
+      "step": 3918
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.269002375296912e-05,
+      "loss": 3.6004,
+      "step": 3919
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.268408551068883e-05,
+      "loss": 3.3121,
+      "step": 3920
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 8.267814726840855e-05,
+      "loss": 2.7901,
+      "step": 3921
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 8.267220902612827e-05,
+      "loss": 2.7205,
+      "step": 3922
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 8.266627078384799e-05,
+      "loss": 2.2774,
+      "step": 3923
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 8.26603325415677e-05,
+      "loss": 1.6183,
+      "step": 3924
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 8.265439429928742e-05,
+      "loss": 1.6513,
+      "step": 3925
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 8.264845605700713e-05,
+      "loss": 1.8864,
+      "step": 3926
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 8.264251781472685e-05,
+      "loss": 1.8262,
+      "step": 3927
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 8.263657957244655e-05,
+      "loss": 2.0229,
+      "step": 3928
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 8.263064133016627e-05,
+      "loss": 1.8431,
+      "step": 3929
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 8.2624703087886e-05,
+      "loss": 1.9438,
+      "step": 3930
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 8.26187648456057e-05,
+      "loss": 1.7178,
+      "step": 3931
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 8.261282660332542e-05,
+      "loss": 1.6943,
+      "step": 3932
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 8.260688836104513e-05,
+      "loss": 1.5911,
+      "step": 3933
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 8.260095011876485e-05,
+      "loss": 1.5369,
+      "step": 3934
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 8.259501187648457e-05,
+      "loss": 1.6283,
+      "step": 3935
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 8.258907363420429e-05,
+      "loss": 1.6804,
+      "step": 3936
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 8.2583135391924e-05,
+      "loss": 1.4525,
+      "step": 3937
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 8.257719714964372e-05,
+      "loss": 1.6613,
+      "step": 3938
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 8.257125890736342e-05,
+      "loss": 1.618,
+      "step": 3939
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 8.256532066508313e-05,
+      "loss": 1.5497,
+      "step": 3940
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 8.255938242280285e-05,
+      "loss": 1.5005,
+      "step": 3941
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 8.255344418052257e-05,
+      "loss": 1.3994,
+      "step": 3942
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 8.254750593824229e-05,
+      "loss": 1.6053,
+      "step": 3943
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 8.2541567695962e-05,
+      "loss": 1.4446,
+      "step": 3944
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 8.253562945368172e-05,
+      "loss": 1.4443,
+      "step": 3945
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 8.252969121140142e-05,
+      "loss": 1.6212,
+      "step": 3946
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 8.252375296912115e-05,
+      "loss": 1.4875,
+      "step": 3947
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 8.251781472684085e-05,
+      "loss": 1.383,
+      "step": 3948
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 8.251187648456057e-05,
+      "loss": 1.6321,
+      "step": 3949
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 8.250593824228029e-05,
+      "loss": 1.4015,
+      "step": 3950
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 8.25e-05,
+      "loss": 1.3646,
+      "step": 3951
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 8.249406175771972e-05,
+      "loss": 1.3577,
+      "step": 3952
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 8.248812351543943e-05,
+      "loss": 1.2348,
+      "step": 3953
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 8.248218527315915e-05,
+      "loss": 1.4252,
+      "step": 3954
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 8.247624703087887e-05,
+      "loss": 1.4717,
+      "step": 3955
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 8.247030878859859e-05,
+      "loss": 1.3945,
+      "step": 3956
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 8.24643705463183e-05,
+      "loss": 1.279,
+      "step": 3957
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 8.245843230403801e-05,
+      "loss": 1.1492,
+      "step": 3958
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 8.245249406175772e-05,
+      "loss": 1.634,
+      "step": 3959
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 8.244655581947743e-05,
+      "loss": 1.286,
+      "step": 3960
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 8.244061757719715e-05,
+      "loss": 1.2411,
+      "step": 3961
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 8.243467933491687e-05,
+      "loss": 1.3038,
+      "step": 3962
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 8.242874109263659e-05,
+      "loss": 1.0984,
+      "step": 3963
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 8.24228028503563e-05,
+      "loss": 1.1143,
+      "step": 3964
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.241686460807602e-05,
+      "loss": 1.129,
+      "step": 3965
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.241092636579572e-05,
+      "loss": 1.0747,
+      "step": 3966
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.240498812351544e-05,
+      "loss": 0.9844,
+      "step": 3967
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.239904988123515e-05,
+      "loss": 1.2286,
+      "step": 3968
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.239311163895488e-05,
+      "loss": 3.4992,
+      "step": 3969
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.238717339667459e-05,
+      "loss": 3.475,
+      "step": 3970
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.23812351543943e-05,
+      "loss": 3.1022,
+      "step": 3971
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.237529691211402e-05,
+      "loss": 2.4096,
+      "step": 3972
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.236935866983373e-05,
+      "loss": 2.105,
+      "step": 3973
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.236342042755345e-05,
+      "loss": 1.5589,
+      "step": 3974
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.235748218527317e-05,
+      "loss": 1.5834,
+      "step": 3975
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.235154394299289e-05,
+      "loss": 1.7367,
+      "step": 3976
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.23456057007126e-05,
+      "loss": 2.0678,
+      "step": 3977
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.233966745843231e-05,
+      "loss": 1.9474,
+      "step": 3978
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.233372921615202e-05,
+      "loss": 1.8228,
+      "step": 3979
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.232779097387174e-05,
+      "loss": 2.0427,
+      "step": 3980
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.232185273159145e-05,
+      "loss": 1.653,
+      "step": 3981
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.231591448931117e-05,
+      "loss": 1.4952,
+      "step": 3982
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.230997624703089e-05,
+      "loss": 1.6323,
+      "step": 3983
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.23040380047506e-05,
+      "loss": 1.4827,
+      "step": 3984
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.229809976247032e-05,
+      "loss": 1.451,
+      "step": 3985
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.229216152019002e-05,
+      "loss": 1.4875,
+      "step": 3986
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.228622327790974e-05,
+      "loss": 1.6336,
+      "step": 3987
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.228028503562946e-05,
+      "loss": 1.538,
+      "step": 3988
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.227434679334918e-05,
+      "loss": 1.6687,
+      "step": 3989
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.226840855106889e-05,
+      "loss": 1.4607,
+      "step": 3990
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.22624703087886e-05,
+      "loss": 1.4776,
+      "step": 3991
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.225653206650832e-05,
+      "loss": 1.3571,
+      "step": 3992
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.225059382422802e-05,
+      "loss": 1.7641,
+      "step": 3993
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.224465558194774e-05,
+      "loss": 1.4039,
+      "step": 3994
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.223871733966746e-05,
+      "loss": 1.3121,
+      "step": 3995
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.223277909738719e-05,
+      "loss": 1.3968,
+      "step": 3996
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.222684085510689e-05,
+      "loss": 1.3985,
+      "step": 3997
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.222090261282661e-05,
+      "loss": 1.2576,
+      "step": 3998
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.221496437054632e-05,
+      "loss": 1.1574,
+      "step": 3999
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.220902612826604e-05,
+      "loss": 1.2422,
+      "step": 4000
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 8.220308788598575e-05,
+      "loss": 1.2175,
+      "step": 4001
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 8.219714964370547e-05,
+      "loss": 1.2323,
+      "step": 4002
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 8.219121140142519e-05,
+      "loss": 1.44,
+      "step": 4003
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 8.21852731591449e-05,
+      "loss": 1.3838,
+      "step": 4004
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 8.217933491686461e-05,
+      "loss": 1.2947,
+      "step": 4005
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 8.217339667458432e-05,
+      "loss": 1.0295,
+      "step": 4006
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 8.216745843230404e-05,
+      "loss": 1.4377,
+      "step": 4007
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 8.216152019002376e-05,
+      "loss": 1.3834,
+      "step": 4008
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 8.215558194774348e-05,
+      "loss": 1.1608,
+      "step": 4009
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 8.214964370546319e-05,
+      "loss": 0.9569,
+      "step": 4010
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 8.214370546318291e-05,
+      "loss": 1.2606,
+      "step": 4011
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 8.213776722090262e-05,
+      "loss": 1.2535,
+      "step": 4012
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 8.213182897862232e-05,
+      "loss": 1.0547,
+      "step": 4013
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 8.212589073634204e-05,
+      "loss": 0.969,
+      "step": 4014
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 8.211995249406176e-05,
+      "loss": 1.0334,
+      "step": 4015
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 8.211401425178148e-05,
+      "loss": 1.1645,
+      "step": 4016
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 8.210807600950119e-05,
+      "loss": 0.9336,
+      "step": 4017
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 8.210213776722091e-05,
+      "loss": 1.1391,
+      "step": 4018
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 8.209619952494062e-05,
+      "loss": 3.1952,
+      "step": 4019
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 8.209026128266034e-05,
+      "loss": 3.1645,
+      "step": 4020
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 8.208432304038005e-05,
+      "loss": 3.1457,
+      "step": 4021
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 8.207838479809977e-05,
+      "loss": 2.3323,
+      "step": 4022
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 8.207244655581949e-05,
+      "loss": 1.7837,
+      "step": 4023
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 8.206650831353919e-05,
+      "loss": 1.5948,
+      "step": 4024
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 8.206057007125891e-05,
+      "loss": 1.7198,
+      "step": 4025
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 8.205463182897862e-05,
+      "loss": 1.6464,
+      "step": 4026
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 8.204869358669834e-05,
+      "loss": 1.6907,
+      "step": 4027
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 8.204275534441806e-05,
+      "loss": 1.7412,
+      "step": 4028
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 8.203681710213778e-05,
+      "loss": 1.6062,
+      "step": 4029
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 8.203087885985749e-05,
+      "loss": 1.9552,
+      "step": 4030
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 8.202494061757721e-05,
+      "loss": 1.6415,
+      "step": 4031
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 8.201900237529691e-05,
+      "loss": 1.6703,
+      "step": 4032
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 8.201306413301662e-05,
+      "loss": 1.4668,
+      "step": 4033
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 8.200712589073634e-05,
+      "loss": 1.7105,
+      "step": 4034
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 8.200118764845606e-05,
+      "loss": 1.4865,
+      "step": 4035
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 8.199524940617578e-05,
+      "loss": 1.5089,
+      "step": 4036
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 8.198931116389549e-05,
+      "loss": 1.4741,
+      "step": 4037
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 8.198337292161521e-05,
+      "loss": 1.4753,
+      "step": 4038
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 8.197743467933492e-05,
+      "loss": 1.5467,
+      "step": 4039
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 8.197149643705464e-05,
+      "loss": 1.3649,
+      "step": 4040
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 8.196555819477434e-05,
+      "loss": 1.3262,
+      "step": 4041
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 8.195961995249408e-05,
+      "loss": 1.5862,
+      "step": 4042
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 8.195368171021378e-05,
+      "loss": 1.4377,
+      "step": 4043
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 8.194774346793349e-05,
+      "loss": 1.2735,
+      "step": 4044
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 8.194180522565321e-05,
+      "loss": 1.2842,
+      "step": 4045
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 8.193586698337292e-05,
+      "loss": 1.2785,
+      "step": 4046
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 8.192992874109264e-05,
+      "loss": 1.5001,
+      "step": 4047
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 8.192399049881236e-05,
+      "loss": 1.4913,
+      "step": 4048
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 8.191805225653208e-05,
+      "loss": 1.6298,
+      "step": 4049
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 8.191211401425179e-05,
+      "loss": 1.3224,
+      "step": 4050
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 8.19061757719715e-05,
+      "loss": 1.1708,
+      "step": 4051
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 8.190023752969121e-05,
+      "loss": 1.4949,
+      "step": 4052
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 8.189429928741092e-05,
+      "loss": 1.3775,
+      "step": 4053
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 8.188836104513064e-05,
+      "loss": 0.9668,
+      "step": 4054
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 8.188242280285036e-05,
+      "loss": 1.4086,
+      "step": 4055
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 8.187648456057008e-05,
+      "loss": 1.1549,
+      "step": 4056
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 8.187054631828979e-05,
+      "loss": 1.0033,
+      "step": 4057
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 8.186460807600951e-05,
+      "loss": 1.0498,
+      "step": 4058
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 8.185866983372922e-05,
+      "loss": 1.3375,
+      "step": 4059
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 8.185273159144894e-05,
+      "loss": 1.1322,
+      "step": 4060
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 8.184679334916864e-05,
+      "loss": 1.4067,
+      "step": 4061
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 8.184085510688838e-05,
+      "loss": 1.159,
+      "step": 4062
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 8.183491686460808e-05,
+      "loss": 1.0812,
+      "step": 4063
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 8.182897862232779e-05,
+      "loss": 0.9194,
+      "step": 4064
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 8.182304038004751e-05,
+      "loss": 1.2215,
+      "step": 4065
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 8.181710213776722e-05,
+      "loss": 1.1445,
+      "step": 4066
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 8.181116389548694e-05,
+      "loss": 0.913,
+      "step": 4067
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 8.180522565320666e-05,
+      "loss": 0.7538,
+      "step": 4068
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 8.179928741092638e-05,
+      "loss": 3.3597,
+      "step": 4069
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 8.179334916864609e-05,
+      "loss": 3.2475,
+      "step": 4070
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 8.17874109263658e-05,
+      "loss": 2.9332,
+      "step": 4071
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 8.178147268408551e-05,
+      "loss": 2.5246,
+      "step": 4072
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 8.177553444180523e-05,
+      "loss": 2.0101,
+      "step": 4073
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 8.176959619952494e-05,
+      "loss": 1.6956,
+      "step": 4074
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 8.176365795724466e-05,
+      "loss": 1.4623,
+      "step": 4075
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 8.175771971496438e-05,
+      "loss": 1.5381,
+      "step": 4076
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 8.175178147268409e-05,
+      "loss": 1.6126,
+      "step": 4077
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 8.174584323040381e-05,
+      "loss": 1.51,
+      "step": 4078
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 8.173990498812351e-05,
+      "loss": 1.5531,
+      "step": 4079
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 8.173396674584323e-05,
+      "loss": 1.5452,
+      "step": 4080
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 8.172802850356294e-05,
+      "loss": 1.8381,
+      "step": 4081
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 8.172209026128268e-05,
+      "loss": 1.5012,
+      "step": 4082
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 8.171615201900238e-05,
+      "loss": 1.4286,
+      "step": 4083
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 8.171021377672209e-05,
+      "loss": 1.2688,
+      "step": 4084
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 8.170427553444181e-05,
+      "loss": 1.3161,
+      "step": 4085
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 8.169833729216152e-05,
+      "loss": 1.3821,
+      "step": 4086
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 8.169239904988124e-05,
+      "loss": 1.2325,
+      "step": 4087
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 8.168646080760096e-05,
+      "loss": 1.2194,
+      "step": 4088
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 8.168052256532068e-05,
+      "loss": 1.3854,
+      "step": 4089
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.167458432304038e-05,
+      "loss": 1.3201,
+      "step": 4090
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.16686460807601e-05,
+      "loss": 1.5176,
+      "step": 4091
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.166270783847981e-05,
+      "loss": 1.3225,
+      "step": 4092
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.165676959619953e-05,
+      "loss": 1.1002,
+      "step": 4093
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.165083135391924e-05,
+      "loss": 1.2102,
+      "step": 4094
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.164489311163896e-05,
+      "loss": 1.5748,
+      "step": 4095
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.163895486935868e-05,
+      "loss": 1.0964,
+      "step": 4096
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.163301662707839e-05,
+      "loss": 1.288,
+      "step": 4097
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.16270783847981e-05,
+      "loss": 1.2489,
+      "step": 4098
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.162114014251781e-05,
+      "loss": 1.2776,
+      "step": 4099
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.161520190023753e-05,
+      "loss": 1.0215,
+      "step": 4100
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.160926365795724e-05,
+      "loss": 1.1262,
+      "step": 4101
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.160332541567697e-05,
+      "loss": 1.1395,
+      "step": 4102
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.159738717339668e-05,
+      "loss": 1.2762,
+      "step": 4103
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.15914489311164e-05,
+      "loss": 1.2304,
+      "step": 4104
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.158551068883611e-05,
+      "loss": 0.9614,
+      "step": 4105
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.157957244655581e-05,
+      "loss": 1.0464,
+      "step": 4106
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.157363420427553e-05,
+      "loss": 1.2753,
+      "step": 4107
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.156769596199526e-05,
+      "loss": 1.1946,
+      "step": 4108
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.156175771971498e-05,
+      "loss": 1.2153,
+      "step": 4109
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.155581947743468e-05,
+      "loss": 1.3457,
+      "step": 4110
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.15498812351544e-05,
+      "loss": 1.037,
+      "step": 4111
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.154394299287411e-05,
+      "loss": 1.2481,
+      "step": 4112
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.153800475059383e-05,
+      "loss": 0.8245,
+      "step": 4113
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.153206650831354e-05,
+      "loss": 0.9145,
+      "step": 4114
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.152612826603326e-05,
+      "loss": 1.1718,
+      "step": 4115
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.152019002375298e-05,
+      "loss": 0.895,
+      "step": 4116
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 8.151425178147268e-05,
+      "loss": 0.9215,
+      "step": 4117
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 8.15083135391924e-05,
+      "loss": 0.6953,
+      "step": 4118
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 8.150237529691211e-05,
+      "loss": 2.779,
+      "step": 4119
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 8.149643705463183e-05,
+      "loss": 2.7125,
+      "step": 4120
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 8.149049881235155e-05,
+      "loss": 2.7203,
+      "step": 4121
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 8.148456057007127e-05,
+      "loss": 2.2526,
+      "step": 4122
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 8.147862232779098e-05,
+      "loss": 1.7877,
+      "step": 4123
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 8.14726840855107e-05,
+      "loss": 1.3823,
+      "step": 4124
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 8.14667458432304e-05,
+      "loss": 1.5526,
+      "step": 4125
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 8.146080760095011e-05,
+      "loss": 1.3722,
+      "step": 4126
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 8.145486935866983e-05,
+      "loss": 1.7388,
+      "step": 4127
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 8.144893111638955e-05,
+      "loss": 1.7665,
+      "step": 4128
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 8.144299287410927e-05,
+      "loss": 1.5236,
+      "step": 4129
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 8.143705463182898e-05,
+      "loss": 1.587,
+      "step": 4130
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 8.14311163895487e-05,
+      "loss": 1.2463,
+      "step": 4131
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 8.142517814726841e-05,
+      "loss": 1.3645,
+      "step": 4132
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 8.141923990498813e-05,
+      "loss": 1.3354,
+      "step": 4133
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 8.141330166270784e-05,
+      "loss": 1.4274,
+      "step": 4134
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 8.140736342042756e-05,
+      "loss": 1.2772,
+      "step": 4135
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 8.140142517814728e-05,
+      "loss": 1.359,
+      "step": 4136
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 8.139548693586698e-05,
+      "loss": 1.2259,
+      "step": 4137
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 8.13895486935867e-05,
+      "loss": 1.4839,
+      "step": 4138
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 8.138361045130641e-05,
+      "loss": 1.346,
+      "step": 4139
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 8.137767220902613e-05,
+      "loss": 1.1205,
+      "step": 4140
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 8.137173396674585e-05,
+      "loss": 1.3709,
+      "step": 4141
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 8.136579572446557e-05,
+      "loss": 1.1361,
+      "step": 4142
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 8.135985748218528e-05,
+      "loss": 1.3537,
+      "step": 4143
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 8.1353919239905e-05,
+      "loss": 1.1361,
+      "step": 4144
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 8.13479809976247e-05,
+      "loss": 1.5953,
+      "step": 4145
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 8.134204275534441e-05,
+      "loss": 1.2312,
+      "step": 4146
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 8.133610451306413e-05,
+      "loss": 1.1437,
+      "step": 4147
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 8.133016627078385e-05,
+      "loss": 1.2299,
+      "step": 4148
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 8.132422802850357e-05,
+      "loss": 1.0829,
+      "step": 4149
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 8.131828978622328e-05,
+      "loss": 1.3592,
+      "step": 4150
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 8.1312351543943e-05,
+      "loss": 1.0132,
+      "step": 4151
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 8.130641330166271e-05,
+      "loss": 0.9891,
+      "step": 4152
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 8.130047505938243e-05,
+      "loss": 1.1642,
+      "step": 4153
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 8.129453681710213e-05,
+      "loss": 1.0899,
+      "step": 4154
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 8.128859857482187e-05,
+      "loss": 1.0494,
+      "step": 4155
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 8.128266033254157e-05,
+      "loss": 1.0113,
+      "step": 4156
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 8.127672209026128e-05,
+      "loss": 1.089,
+      "step": 4157
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 8.1270783847981e-05,
+      "loss": 1.0722,
+      "step": 4158
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 8.126484560570071e-05,
+      "loss": 1.0581,
+      "step": 4159
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 8.125890736342043e-05,
+      "loss": 1.0171,
+      "step": 4160
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 8.125296912114015e-05,
+      "loss": 1.0749,
+      "step": 4161
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 8.124703087885987e-05,
+      "loss": 1.0624,
+      "step": 4162
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 8.124109263657958e-05,
+      "loss": 1.2244,
+      "step": 4163
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 8.12351543942993e-05,
+      "loss": 1.1862,
+      "step": 4164
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 8.1229216152019e-05,
+      "loss": 2.6455,
+      "step": 4165
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 8.122327790973872e-05,
+      "loss": 0.9823,
+      "step": 4166
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 8.121733966745843e-05,
+      "loss": 1.4337,
+      "step": 4167
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 8.121140142517815e-05,
+      "loss": 1.325,
+      "step": 4168
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 8.120546318289787e-05,
+      "loss": 2.9601,
+      "step": 4169
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 8.119952494061758e-05,
+      "loss": 2.4035,
+      "step": 4170
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 8.11935866983373e-05,
+      "loss": 2.1831,
+      "step": 4171
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 8.1187648456057e-05,
+      "loss": 1.8705,
+      "step": 4172
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 8.118171021377673e-05,
+      "loss": 1.6395,
+      "step": 4173
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 8.117577197149643e-05,
+      "loss": 1.4668,
+      "step": 4174
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 8.116983372921617e-05,
+      "loss": 1.3875,
+      "step": 4175
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 8.116389548693587e-05,
+      "loss": 1.4068,
+      "step": 4176
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 8.11579572446556e-05,
+      "loss": 1.5439,
+      "step": 4177
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 8.11520190023753e-05,
+      "loss": 1.7411,
+      "step": 4178
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 8.114608076009501e-05,
+      "loss": 1.5325,
+      "step": 4179
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 8.114014251781473e-05,
+      "loss": 1.5341,
+      "step": 4180
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 8.113420427553445e-05,
+      "loss": 1.4747,
+      "step": 4181
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 8.112826603325417e-05,
+      "loss": 1.3724,
+      "step": 4182
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 8.112232779097388e-05,
+      "loss": 1.5572,
+      "step": 4183
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 8.11163895486936e-05,
+      "loss": 1.6052,
+      "step": 4184
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 8.11104513064133e-05,
+      "loss": 1.2294,
+      "step": 4185
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 8.110451306413302e-05,
+      "loss": 1.2164,
+      "step": 4186
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 8.109857482185273e-05,
+      "loss": 1.4138,
+      "step": 4187
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 8.109263657957245e-05,
+      "loss": 1.4451,
+      "step": 4188
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 8.108669833729217e-05,
+      "loss": 1.0897,
+      "step": 4189
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 8.108076009501188e-05,
+      "loss": 1.0182,
+      "step": 4190
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 8.10748218527316e-05,
+      "loss": 1.1518,
+      "step": 4191
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 8.10688836104513e-05,
+      "loss": 1.1081,
+      "step": 4192
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 8.106294536817102e-05,
+      "loss": 1.2788,
+      "step": 4193
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 8.105700712589073e-05,
+      "loss": 1.2265,
+      "step": 4194
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 8.105106888361047e-05,
+      "loss": 1.3373,
+      "step": 4195
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 8.104513064133017e-05,
+      "loss": 1.3224,
+      "step": 4196
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 8.103919239904989e-05,
+      "loss": 1.1278,
+      "step": 4197
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 8.10332541567696e-05,
+      "loss": 1.3263,
+      "step": 4198
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 8.10273159144893e-05,
+      "loss": 1.2675,
+      "step": 4199
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 8.102137767220903e-05,
+      "loss": 1.2857,
+      "step": 4200
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 8.101543942992875e-05,
+      "loss": 1.0267,
+      "step": 4201
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 8.100950118764847e-05,
+      "loss": 1.1179,
+      "step": 4202
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 8.100356294536817e-05,
+      "loss": 1.1197,
+      "step": 4203
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 8.09976247030879e-05,
+      "loss": 1.1118,
+      "step": 4204
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 8.09916864608076e-05,
+      "loss": 0.9923,
+      "step": 4205
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.098574821852732e-05,
+      "loss": 1.1075,
+      "step": 4206
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.097980997624703e-05,
+      "loss": 1.0066,
+      "step": 4207
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.097387173396675e-05,
+      "loss": 1.2453,
+      "step": 4208
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.096793349168647e-05,
+      "loss": 1.1985,
+      "step": 4209
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.096199524940618e-05,
+      "loss": 1.1955,
+      "step": 4210
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.09560570071259e-05,
+      "loss": 0.9236,
+      "step": 4211
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.09501187648456e-05,
+      "loss": 0.9961,
+      "step": 4212
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.094418052256532e-05,
+      "loss": 0.8084,
+      "step": 4213
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.093824228028503e-05,
+      "loss": 0.9097,
+      "step": 4214
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.093230403800476e-05,
+      "loss": 1.2391,
+      "step": 4215
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.092636579572447e-05,
+      "loss": 1.2687,
+      "step": 4216
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.092042755344419e-05,
+      "loss": 1.1449,
+      "step": 4217
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.09144893111639e-05,
+      "loss": 0.9362,
+      "step": 4218
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.09085510688836e-05,
+      "loss": 3.1881,
+      "step": 4219
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.090261282660333e-05,
+      "loss": 2.8785,
+      "step": 4220
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.089667458432305e-05,
+      "loss": 2.5977,
+      "step": 4221
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.089073634204277e-05,
+      "loss": 2.369,
+      "step": 4222
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.088479809976247e-05,
+      "loss": 1.7029,
+      "step": 4223
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.08788598574822e-05,
+      "loss": 1.4112,
+      "step": 4224
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.08729216152019e-05,
+      "loss": 1.7911,
+      "step": 4225
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.086698337292162e-05,
+      "loss": 1.4243,
+      "step": 4226
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.086104513064133e-05,
+      "loss": 1.3944,
+      "step": 4227
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.085510688836105e-05,
+      "loss": 1.3699,
+      "step": 4228
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.084916864608077e-05,
+      "loss": 1.2663,
+      "step": 4229
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.084323040380047e-05,
+      "loss": 1.3283,
+      "step": 4230
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.08372921615202e-05,
+      "loss": 1.385,
+      "step": 4231
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.08313539192399e-05,
+      "loss": 1.1723,
+      "step": 4232
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.082541567695962e-05,
+      "loss": 1.2647,
+      "step": 4233
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.081947743467934e-05,
+      "loss": 1.0945,
+      "step": 4234
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.081353919239906e-05,
+      "loss": 1.2829,
+      "step": 4235
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.080760095011877e-05,
+      "loss": 1.2297,
+      "step": 4236
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.080166270783849e-05,
+      "loss": 1.2842,
+      "step": 4237
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.07957244655582e-05,
+      "loss": 1.3302,
+      "step": 4238
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.078978622327792e-05,
+      "loss": 1.0727,
+      "step": 4239
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.078384798099762e-05,
+      "loss": 1.0109,
+      "step": 4240
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.077790973871734e-05,
+      "loss": 1.0105,
+      "step": 4241
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.077197149643706e-05,
+      "loss": 1.213,
+      "step": 4242
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.076603325415677e-05,
+      "loss": 1.165,
+      "step": 4243
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.076009501187649e-05,
+      "loss": 1.0441,
+      "step": 4244
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.07541567695962e-05,
+      "loss": 1.3716,
+      "step": 4245
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.074821852731592e-05,
+      "loss": 1.0767,
+      "step": 4246
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.074228028503563e-05,
+      "loss": 1.007,
+      "step": 4247
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.073634204275535e-05,
+      "loss": 1.0412,
+      "step": 4248
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.073040380047507e-05,
+      "loss": 1.0282,
+      "step": 4249
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.072446555819477e-05,
+      "loss": 1.0497,
+      "step": 4250
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.07185273159145e-05,
+      "loss": 0.9927,
+      "step": 4251
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.07125890736342e-05,
+      "loss": 1.2155,
+      "step": 4252
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.070665083135392e-05,
+      "loss": 1.1202,
+      "step": 4253
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.070071258907364e-05,
+      "loss": 1.1727,
+      "step": 4254
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.069477434679336e-05,
+      "loss": 0.9666,
+      "step": 4255
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.068883610451307e-05,
+      "loss": 1.1213,
+      "step": 4256
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.068289786223279e-05,
+      "loss": 0.985,
+      "step": 4257
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.06769596199525e-05,
+      "loss": 1.0297,
+      "step": 4258
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.067102137767222e-05,
+      "loss": 1.0714,
+      "step": 4259
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.066508313539192e-05,
+      "loss": 1.0413,
+      "step": 4260
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.065914489311164e-05,
+      "loss": 1.2174,
+      "step": 4261
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.065320665083136e-05,
+      "loss": 0.7333,
+      "step": 4262
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.064726840855107e-05,
+      "loss": 1.4793,
+      "step": 4263
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.064133016627079e-05,
+      "loss": 1.1236,
+      "step": 4264
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.06353919239905e-05,
+      "loss": 0.9744,
+      "step": 4265
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.062945368171022e-05,
+      "loss": 1.2142,
+      "step": 4266
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.062351543942992e-05,
+      "loss": 0.8983,
+      "step": 4267
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.061757719714965e-05,
+      "loss": 0.8974,
+      "step": 4268
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.061163895486937e-05,
+      "loss": 2.3626,
+      "step": 4269
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.060570071258909e-05,
+      "loss": 2.4427,
+      "step": 4270
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.059976247030879e-05,
+      "loss": 2.1463,
+      "step": 4271
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.05938242280285e-05,
+      "loss": 1.7736,
+      "step": 4272
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.058788598574822e-05,
+      "loss": 1.4086,
+      "step": 4273
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.058194774346794e-05,
+      "loss": 1.3661,
+      "step": 4274
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.057600950118766e-05,
+      "loss": 1.2553,
+      "step": 4275
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.057007125890737e-05,
+      "loss": 1.4639,
+      "step": 4276
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.056413301662709e-05,
+      "loss": 1.5017,
+      "step": 4277
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.05581947743468e-05,
+      "loss": 1.4353,
+      "step": 4278
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.055225653206651e-05,
+      "loss": 1.2391,
+      "step": 4279
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.054631828978622e-05,
+      "loss": 1.3624,
+      "step": 4280
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.054038004750594e-05,
+      "loss": 1.2133,
+      "step": 4281
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.053444180522566e-05,
+      "loss": 1.3047,
+      "step": 4282
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.052850356294537e-05,
+      "loss": 1.3367,
+      "step": 4283
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.052256532066509e-05,
+      "loss": 1.2241,
+      "step": 4284
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.05166270783848e-05,
+      "loss": 1.362,
+      "step": 4285
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.051068883610452e-05,
+      "loss": 1.21,
+      "step": 4286
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.050475059382422e-05,
+      "loss": 1.223,
+      "step": 4287
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.049881235154396e-05,
+      "loss": 1.1203,
+      "step": 4288
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.049287410926366e-05,
+      "loss": 1.4955,
+      "step": 4289
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.048693586698338e-05,
+      "loss": 1.0363,
+      "step": 4290
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.048099762470309e-05,
+      "loss": 0.9705,
+      "step": 4291
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.04750593824228e-05,
+      "loss": 1.0223,
+      "step": 4292
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.046912114014252e-05,
+      "loss": 1.0448,
+      "step": 4293
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.046318289786224e-05,
+      "loss": 1.0449,
+      "step": 4294
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.045724465558196e-05,
+      "loss": 1.0203,
+      "step": 4295
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.045130641330167e-05,
+      "loss": 2.1006,
+      "step": 4296
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.044536817102139e-05,
+      "loss": 0.983,
+      "step": 4297
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.043942992874109e-05,
+      "loss": 1.173,
+      "step": 4298
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.043349168646081e-05,
+      "loss": 1.108,
+      "step": 4299
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.042755344418052e-05,
+      "loss": 0.894,
+      "step": 4300
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.042161520190024e-05,
+      "loss": 0.9051,
+      "step": 4301
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.041567695961996e-05,
+      "loss": 1.07,
+      "step": 4302
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.040973871733967e-05,
+      "loss": 1.0531,
+      "step": 4303
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.040380047505939e-05,
+      "loss": 0.9436,
+      "step": 4304
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.03978622327791e-05,
+      "loss": 0.9709,
+      "step": 4305
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.039192399049882e-05,
+      "loss": 1.0345,
+      "step": 4306
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.038598574821852e-05,
+      "loss": 0.7648,
+      "step": 4307
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.038004750593826e-05,
+      "loss": 0.9895,
+      "step": 4308
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.037410926365796e-05,
+      "loss": 0.9011,
+      "step": 4309
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.036817102137768e-05,
+      "loss": 0.7703,
+      "step": 4310
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.036223277909739e-05,
+      "loss": 1.1549,
+      "step": 4311
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.03562945368171e-05,
+      "loss": 0.9364,
+      "step": 4312
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.035035629453682e-05,
+      "loss": 0.9091,
+      "step": 4313
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.034441805225654e-05,
+      "loss": 0.9388,
+      "step": 4314
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.033847980997626e-05,
+      "loss": 0.7462,
+      "step": 4315
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.033254156769596e-05,
+      "loss": 0.7618,
+      "step": 4316
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.032660332541569e-05,
+      "loss": 0.9183,
+      "step": 4317
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.032066508313539e-05,
+      "loss": 0.7402,
+      "step": 4318
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.031472684085511e-05,
+      "loss": 2.644,
+      "step": 4319
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.030878859857482e-05,
+      "loss": 2.2853,
+      "step": 4320
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.030285035629454e-05,
+      "loss": 2.3849,
+      "step": 4321
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.029691211401426e-05,
+      "loss": 2.2644,
+      "step": 4322
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.029097387173397e-05,
+      "loss": 1.4874,
+      "step": 4323
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.028503562945369e-05,
+      "loss": 1.3303,
+      "step": 4324
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.02790973871734e-05,
+      "loss": 1.3301,
+      "step": 4325
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.027315914489311e-05,
+      "loss": 1.1793,
+      "step": 4326
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.026722090261282e-05,
+      "loss": 1.404,
+      "step": 4327
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.026128266033255e-05,
+      "loss": 1.1999,
+      "step": 4328
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.025534441805226e-05,
+      "loss": 1.796,
+      "step": 4329
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.024940617577198e-05,
+      "loss": 1.4759,
+      "step": 4330
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.024346793349169e-05,
+      "loss": 1.104,
+      "step": 4331
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.023752969121141e-05,
+      "loss": 1.2452,
+      "step": 4332
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.023159144893112e-05,
+      "loss": 1.5098,
+      "step": 4333
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.022565320665084e-05,
+      "loss": 1.1264,
+      "step": 4334
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.021971496437056e-05,
+      "loss": 1.2209,
+      "step": 4335
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.021377672209026e-05,
+      "loss": 1.0978,
+      "step": 4336
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.020783847980998e-05,
+      "loss": 1.078,
+      "step": 4337
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.020190023752969e-05,
+      "loss": 1.1215,
+      "step": 4338
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.019596199524941e-05,
+      "loss": 0.9302,
+      "step": 4339
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.019002375296912e-05,
+      "loss": 1.198,
+      "step": 4340
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.018408551068884e-05,
+      "loss": 1.3156,
+      "step": 4341
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.017814726840856e-05,
+      "loss": 1.1606,
+      "step": 4342
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.017220902612827e-05,
+      "loss": 1.0794,
+      "step": 4343
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.016627078384799e-05,
+      "loss": 1.0733,
+      "step": 4344
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.016033254156769e-05,
+      "loss": 1.1093,
+      "step": 4345
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.015439429928741e-05,
+      "loss": 1.2685,
+      "step": 4346
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.014845605700712e-05,
+      "loss": 0.9904,
+      "step": 4347
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.014251781472685e-05,
+      "loss": 1.0498,
+      "step": 4348
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.013657957244656e-05,
+      "loss": 1.3138,
+      "step": 4349
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.013064133016628e-05,
+      "loss": 0.855,
+      "step": 4350
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.012470308788599e-05,
+      "loss": 1.1168,
+      "step": 4351
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.011876484560571e-05,
+      "loss": 0.9075,
+      "step": 4352
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.011282660332541e-05,
+      "loss": 1.0645,
+      "step": 4353
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.010688836104513e-05,
+      "loss": 1.0019,
+      "step": 4354
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.010095011876486e-05,
+      "loss": 1.0279,
+      "step": 4355
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.009501187648456e-05,
+      "loss": 0.8839,
+      "step": 4356
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.008907363420428e-05,
+      "loss": 0.9694,
+      "step": 4357
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 8.008313539192399e-05,
+      "loss": 0.9334,
+      "step": 4358
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 8.007719714964371e-05,
+      "loss": 0.7814,
+      "step": 4359
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 8.007125890736342e-05,
+      "loss": 0.8591,
+      "step": 4360
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 8.006532066508314e-05,
+      "loss": 1.0541,
+      "step": 4361
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 8.005938242280286e-05,
+      "loss": 0.8407,
+      "step": 4362
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 8.005344418052258e-05,
+      "loss": 1.5599,
+      "step": 4363
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 8.004750593824228e-05,
+      "loss": 0.9154,
+      "step": 4364
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 8.004156769596199e-05,
+      "loss": 1.3114,
+      "step": 4365
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 8.003562945368171e-05,
+      "loss": 0.6372,
+      "step": 4366
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 8.002969121140143e-05,
+      "loss": 1.0592,
+      "step": 4367
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 8.002375296912115e-05,
+      "loss": 0.7896,
+      "step": 4368
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 8.001781472684086e-05,
+      "loss": 2.664,
+      "step": 4369
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 8.001187648456058e-05,
+      "loss": 2.5011,
+      "step": 4370
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 8.000593824228029e-05,
+      "loss": 2.0484,
+      "step": 4371
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 8e-05,
+      "loss": 1.7797,
+      "step": 4372
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.999406175771971e-05,
+      "loss": 1.6162,
+      "step": 4373
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.998812351543943e-05,
+      "loss": 1.6442,
+      "step": 4374
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.998218527315915e-05,
+      "loss": 1.3358,
+      "step": 4375
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.997624703087886e-05,
+      "loss": 1.1742,
+      "step": 4376
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.997030878859858e-05,
+      "loss": 1.324,
+      "step": 4377
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.996437054631829e-05,
+      "loss": 1.2332,
+      "step": 4378
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.995843230403801e-05,
+      "loss": 1.4334,
+      "step": 4379
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.995249406175772e-05,
+      "loss": 1.1179,
+      "step": 4380
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.994655581947744e-05,
+      "loss": 1.088,
+      "step": 4381
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.994061757719716e-05,
+      "loss": 1.1858,
+      "step": 4382
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.993467933491688e-05,
+      "loss": 1.0992,
+      "step": 4383
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.992874109263658e-05,
+      "loss": 1.1571,
+      "step": 4384
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.992280285035629e-05,
+      "loss": 1.1891,
+      "step": 4385
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.991686460807601e-05,
+      "loss": 1.124,
+      "step": 4386
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.991092636579573e-05,
+      "loss": 1.0441,
+      "step": 4387
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.990498812351545e-05,
+      "loss": 1.0017,
+      "step": 4388
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.989904988123516e-05,
+      "loss": 0.8941,
+      "step": 4389
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.989311163895488e-05,
+      "loss": 1.0204,
+      "step": 4390
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.988717339667458e-05,
+      "loss": 1.0288,
+      "step": 4391
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.98812351543943e-05,
+      "loss": 1.2417,
+      "step": 4392
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.987529691211401e-05,
+      "loss": 1.0036,
+      "step": 4393
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.986935866983373e-05,
+      "loss": 0.9824,
+      "step": 4394
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.986342042755345e-05,
+      "loss": 1.1253,
+      "step": 4395
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.985748218527316e-05,
+      "loss": 1.0046,
+      "step": 4396
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.985154394299288e-05,
+      "loss": 1.1085,
+      "step": 4397
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.984560570071259e-05,
+      "loss": 0.9142,
+      "step": 4398
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.983966745843231e-05,
+      "loss": 1.0373,
+      "step": 4399
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.983372921615201e-05,
+      "loss": 0.8982,
+      "step": 4400
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.982779097387175e-05,
+      "loss": 0.9966,
+      "step": 4401
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.982185273159145e-05,
+      "loss": 0.992,
+      "step": 4402
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.981591448931117e-05,
+      "loss": 0.7821,
+      "step": 4403
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.980997624703088e-05,
+      "loss": 0.9382,
+      "step": 4404
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.98040380047506e-05,
+      "loss": 0.8351,
+      "step": 4405
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.979809976247031e-05,
+      "loss": 0.8876,
+      "step": 4406
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.979216152019003e-05,
+      "loss": 0.9613,
+      "step": 4407
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.978622327790975e-05,
+      "loss": 0.9775,
+      "step": 4408
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.978028503562946e-05,
+      "loss": 0.9208,
+      "step": 4409
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.977434679334918e-05,
+      "loss": 0.902,
+      "step": 4410
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.976840855106888e-05,
+      "loss": 1.1199,
+      "step": 4411
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.97624703087886e-05,
+      "loss": 0.9107,
+      "step": 4412
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.975653206650831e-05,
+      "loss": 1.0909,
+      "step": 4413
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.975059382422803e-05,
+      "loss": 0.6941,
+      "step": 4414
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.974465558194775e-05,
+      "loss": 0.8107,
+      "step": 4415
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.973871733966746e-05,
+      "loss": 0.874,
+      "step": 4416
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.973277909738718e-05,
+      "loss": 1.0042,
+      "step": 4417
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.972684085510689e-05,
+      "loss": 0.9587,
+      "step": 4418
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.97209026128266e-05,
+      "loss": 2.2332,
+      "step": 4419
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.971496437054631e-05,
+      "loss": 2.0923,
+      "step": 4420
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.970902612826605e-05,
+      "loss": 2.2997,
+      "step": 4421
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.970308788598575e-05,
+      "loss": 1.6852,
+      "step": 4422
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.969714964370547e-05,
+      "loss": 1.415,
+      "step": 4423
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.969121140142518e-05,
+      "loss": 1.3133,
+      "step": 4424
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.96852731591449e-05,
+      "loss": 0.9623,
+      "step": 4425
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.967933491686461e-05,
+      "loss": 1.2078,
+      "step": 4426
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.967339667458433e-05,
+      "loss": 1.3066,
+      "step": 4427
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.966745843230405e-05,
+      "loss": 1.2491,
+      "step": 4428
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.966152019002376e-05,
+      "loss": 1.2748,
+      "step": 4429
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.965558194774348e-05,
+      "loss": 1.1817,
+      "step": 4430
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.964964370546318e-05,
+      "loss": 1.2555,
+      "step": 4431
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.96437054631829e-05,
+      "loss": 1.0702,
+      "step": 4432
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.963776722090261e-05,
+      "loss": 1.0649,
+      "step": 4433
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.963182897862233e-05,
+      "loss": 1.0264,
+      "step": 4434
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.962589073634205e-05,
+      "loss": 1.0271,
+      "step": 4435
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.961995249406177e-05,
+      "loss": 1.0441,
+      "step": 4436
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.961401425178148e-05,
+      "loss": 1.183,
+      "step": 4437
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.960807600950118e-05,
+      "loss": 1.0537,
+      "step": 4438
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.96021377672209e-05,
+      "loss": 0.9775,
+      "step": 4439
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.959619952494061e-05,
+      "loss": 0.9963,
+      "step": 4440
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.959026128266035e-05,
+      "loss": 1.0881,
+      "step": 4441
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.958432304038005e-05,
+      "loss": 1.1003,
+      "step": 4442
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.957838479809977e-05,
+      "loss": 0.9653,
+      "step": 4443
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.957244655581948e-05,
+      "loss": 0.9848,
+      "step": 4444
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.95665083135392e-05,
+      "loss": 0.9283,
+      "step": 4445
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.95605700712589e-05,
+      "loss": 1.1667,
+      "step": 4446
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.955463182897863e-05,
+      "loss": 1.5729,
+      "step": 4447
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.954869358669835e-05,
+      "loss": 1.2116,
+      "step": 4448
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.954275534441805e-05,
+      "loss": 0.9872,
+      "step": 4449
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.953681710213777e-05,
+      "loss": 0.8848,
+      "step": 4450
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.953087885985748e-05,
+      "loss": 0.7774,
+      "step": 4451
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.95249406175772e-05,
+      "loss": 0.9277,
+      "step": 4452
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.951900237529691e-05,
+      "loss": 1.0238,
+      "step": 4453
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.951306413301663e-05,
+      "loss": 1.0613,
+      "step": 4454
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.950712589073635e-05,
+      "loss": 0.7155,
+      "step": 4455
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.950118764845607e-05,
+      "loss": 0.8938,
+      "step": 4456
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.949524940617578e-05,
+      "loss": 0.7946,
+      "step": 4457
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.948931116389548e-05,
+      "loss": 0.6998,
+      "step": 4458
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.94833729216152e-05,
+      "loss": 0.9918,
+      "step": 4459
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.947743467933491e-05,
+      "loss": 0.6498,
+      "step": 4460
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.947149643705464e-05,
+      "loss": 1.6967,
+      "step": 4461
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.946555819477435e-05,
+      "loss": 1.5408,
+      "step": 4462
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.945961995249407e-05,
+      "loss": 1.5445,
+      "step": 4463
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.945368171021378e-05,
+      "loss": 1.349,
+      "step": 4464
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 7.94477434679335e-05,
+      "loss": 1.2371,
+      "step": 4465
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 7.94418052256532e-05,
+      "loss": 0.9193,
+      "step": 4466
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 7.943586698337293e-05,
+      "loss": 0.8691,
+      "step": 4467
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 7.942992874109265e-05,
+      "loss": 0.9121,
+      "step": 4468
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 7.942399049881235e-05,
+      "loss": 0.8365,
+      "step": 4469
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 7.941805225653207e-05,
+      "loss": 1.0294,
+      "step": 4470
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 7.941211401425178e-05,
+      "loss": 0.9333,
+      "step": 4471
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 7.94061757719715e-05,
+      "loss": 0.9234,
+      "step": 4472
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 7.940023752969121e-05,
+      "loss": 0.9325,
+      "step": 4473
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.939429928741093e-05,
+      "loss": 0.8012,
+      "step": 4474
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.938836104513065e-05,
+      "loss": 0.741,
+      "step": 4475
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.938242280285037e-05,
+      "loss": 0.8792,
+      "step": 4476
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.937648456057007e-05,
+      "loss": 0.8121,
+      "step": 4477
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.937054631828978e-05,
+      "loss": 0.8372,
+      "step": 4478
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.93646080760095e-05,
+      "loss": 0.8204,
+      "step": 4479
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.935866983372922e-05,
+      "loss": 0.8254,
+      "step": 4480
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.935273159144894e-05,
+      "loss": 0.8864,
+      "step": 4481
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.934679334916865e-05,
+      "loss": 0.7111,
+      "step": 4482
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.934085510688837e-05,
+      "loss": 0.6399,
+      "step": 4483
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.933491686460808e-05,
+      "loss": 0.8835,
+      "step": 4484
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.93289786223278e-05,
+      "loss": 0.7005,
+      "step": 4485
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.93230403800475e-05,
+      "loss": 0.7826,
+      "step": 4486
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.931710213776722e-05,
+      "loss": 0.7326,
+      "step": 4487
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.931116389548694e-05,
+      "loss": 0.6366,
+      "step": 4488
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.930522565320665e-05,
+      "loss": 0.742,
+      "step": 4489
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.929928741092637e-05,
+      "loss": 0.7526,
+      "step": 4490
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.929334916864608e-05,
+      "loss": 0.8204,
+      "step": 4491
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.92874109263658e-05,
+      "loss": 0.9882,
+      "step": 4492
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.92814726840855e-05,
+      "loss": 0.8899,
+      "step": 4493
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.927553444180523e-05,
+      "loss": 0.7064,
+      "step": 4494
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.926959619952495e-05,
+      "loss": 0.7038,
+      "step": 4495
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.926365795724467e-05,
+      "loss": 0.7272,
+      "step": 4496
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.925771971496437e-05,
+      "loss": 0.5373,
+      "step": 4497
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.92517814726841e-05,
+      "loss": 0.8449,
+      "step": 4498
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.92458432304038e-05,
+      "loss": 0.7017,
+      "step": 4499
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.923990498812352e-05,
+      "loss": 0.7354,
+      "step": 4500
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.923396674584324e-05,
+      "loss": 0.6347,
+      "step": 4501
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.922802850356295e-05,
+      "loss": 0.6731,
+      "step": 4502
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.922209026128267e-05,
+      "loss": 0.6478,
+      "step": 4503
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.921615201900238e-05,
+      "loss": 0.7043,
+      "step": 4504
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.92102137767221e-05,
+      "loss": 0.5097,
+      "step": 4505
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.92042755344418e-05,
+      "loss": 0.4767,
+      "step": 4506
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.919833729216152e-05,
+      "loss": 0.6248,
+      "step": 4507
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.919239904988124e-05,
+      "loss": 0.6105,
+      "step": 4508
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.918646080760095e-05,
+      "loss": 0.6078,
+      "step": 4509
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.918052256532067e-05,
+      "loss": 0.6659,
+      "step": 4510
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.917458432304038e-05,
+      "loss": 2.0271,
+      "step": 4511
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.91686460807601e-05,
+      "loss": 1.7379,
+      "step": 4512
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.91627078384798e-05,
+      "loss": 1.6647,
+      "step": 4513
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.915676959619952e-05,
+      "loss": 1.3209,
+      "step": 4514
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.915083135391925e-05,
+      "loss": 1.0693,
+      "step": 4515
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.914489311163897e-05,
+      "loss": 0.9966,
+      "step": 4516
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.913895486935867e-05,
+      "loss": 0.9853,
+      "step": 4517
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.913301662707839e-05,
+      "loss": 1.0885,
+      "step": 4518
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.91270783847981e-05,
+      "loss": 1.0378,
+      "step": 4519
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.912114014251782e-05,
+      "loss": 1.1656,
+      "step": 4520
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.911520190023754e-05,
+      "loss": 1.1132,
+      "step": 4521
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.910926365795725e-05,
+      "loss": 0.9937,
+      "step": 4522
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.910332541567697e-05,
+      "loss": 0.9752,
+      "step": 4523
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.909738717339667e-05,
+      "loss": 0.7915,
+      "step": 4524
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.90914489311164e-05,
+      "loss": 0.7752,
+      "step": 4525
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.90855106888361e-05,
+      "loss": 0.7056,
+      "step": 4526
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.907957244655582e-05,
+      "loss": 0.8359,
+      "step": 4527
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.907363420427554e-05,
+      "loss": 0.5888,
+      "step": 4528
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.906769596199526e-05,
+      "loss": 0.727,
+      "step": 4529
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.906175771971497e-05,
+      "loss": 0.8889,
+      "step": 4530
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.905581947743468e-05,
+      "loss": 0.7293,
+      "step": 4531
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.90498812351544e-05,
+      "loss": 0.7622,
+      "step": 4532
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.90439429928741e-05,
+      "loss": 0.8291,
+      "step": 4533
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.903800475059384e-05,
+      "loss": 0.7986,
+      "step": 4534
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.903206650831354e-05,
+      "loss": 0.7463,
+      "step": 4535
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.902612826603326e-05,
+      "loss": 0.7612,
+      "step": 4536
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.902019002375297e-05,
+      "loss": 0.719,
+      "step": 4537
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.901425178147269e-05,
+      "loss": 0.7116,
+      "step": 4538
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.90083135391924e-05,
+      "loss": 0.6164,
+      "step": 4539
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.900237529691212e-05,
+      "loss": 0.82,
+      "step": 4540
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.899643705463184e-05,
+      "loss": 0.7439,
+      "step": 4541
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.899049881235155e-05,
+      "loss": 0.6578,
+      "step": 4542
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.898456057007127e-05,
+      "loss": 0.5822,
+      "step": 4543
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.897862232779097e-05,
+      "loss": 0.6493,
+      "step": 4544
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.897268408551069e-05,
+      "loss": 0.7222,
+      "step": 4545
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.89667458432304e-05,
+      "loss": 0.6886,
+      "step": 4546
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.896080760095012e-05,
+      "loss": 0.693,
+      "step": 4547
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.895486935866984e-05,
+      "loss": 0.672,
+      "step": 4548
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.894893111638956e-05,
+      "loss": 0.548,
+      "step": 4549
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.894299287410927e-05,
+      "loss": 0.6081,
+      "step": 4550
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.893705463182897e-05,
+      "loss": 0.6811,
+      "step": 4551
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.89311163895487e-05,
+      "loss": 0.5292,
+      "step": 4552
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.89251781472684e-05,
+      "loss": 0.7117,
+      "step": 4553
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.891923990498814e-05,
+      "loss": 0.5184,
+      "step": 4554
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.891330166270784e-05,
+      "loss": 0.5501,
+      "step": 4555
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.890736342042756e-05,
+      "loss": 0.3648,
+      "step": 4556
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.890142517814727e-05,
+      "loss": 0.5508,
+      "step": 4557
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.889548693586699e-05,
+      "loss": 0.6222,
+      "step": 4558
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.88895486935867e-05,
+      "loss": 0.5176,
+      "step": 4559
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.888361045130642e-05,
+      "loss": 0.9792,
+      "step": 4560
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.887767220902614e-05,
+      "loss": 2.2705,
+      "step": 4561
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.887173396674584e-05,
+      "loss": 1.729,
+      "step": 4562
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.886579572446556e-05,
+      "loss": 1.6706,
+      "step": 4563
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.885985748218527e-05,
+      "loss": 1.2773,
+      "step": 4564
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.885391923990499e-05,
+      "loss": 1.2422,
+      "step": 4565
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.88479809976247e-05,
+      "loss": 1.035,
+      "step": 4566
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.884204275534442e-05,
+      "loss": 0.8105,
+      "step": 4567
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.883610451306414e-05,
+      "loss": 1.0175,
+      "step": 4568
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.883016627078386e-05,
+      "loss": 1.0488,
+      "step": 4569
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.882422802850357e-05,
+      "loss": 1.0312,
+      "step": 4570
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.881828978622329e-05,
+      "loss": 0.8659,
+      "step": 4571
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.8812351543943e-05,
+      "loss": 1.0447,
+      "step": 4572
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.88064133016627e-05,
+      "loss": 0.9434,
+      "step": 4573
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.880047505938243e-05,
+      "loss": 0.8119,
+      "step": 4574
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.879453681710214e-05,
+      "loss": 0.8702,
+      "step": 4575
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.878859857482186e-05,
+      "loss": 0.6835,
+      "step": 4576
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.878266033254157e-05,
+      "loss": 0.7219,
+      "step": 4577
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.877672209026129e-05,
+      "loss": 0.7219,
+      "step": 4578
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.8770783847981e-05,
+      "loss": 0.6618,
+      "step": 4579
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.876484560570072e-05,
+      "loss": 1.0247,
+      "step": 4580
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.875890736342044e-05,
+      "loss": 0.9374,
+      "step": 4581
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.875296912114014e-05,
+      "loss": 0.7835,
+      "step": 4582
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.874703087885986e-05,
+      "loss": 0.7715,
+      "step": 4583
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.874109263657957e-05,
+      "loss": 0.7601,
+      "step": 4584
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.873515439429929e-05,
+      "loss": 0.7416,
+      "step": 4585
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.8729216152019e-05,
+      "loss": 0.7071,
+      "step": 4586
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.872327790973872e-05,
+      "loss": 0.7933,
+      "step": 4587
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.871733966745844e-05,
+      "loss": 0.812,
+      "step": 4588
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.871140142517816e-05,
+      "loss": 0.7611,
+      "step": 4589
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.870546318289787e-05,
+      "loss": 0.89,
+      "step": 4590
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.869952494061759e-05,
+      "loss": 0.7888,
+      "step": 4591
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.869358669833729e-05,
+      "loss": 0.6019,
+      "step": 4592
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.8687648456057e-05,
+      "loss": 0.6739,
+      "step": 4593
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.868171021377673e-05,
+      "loss": 0.6234,
+      "step": 4594
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.867577197149644e-05,
+      "loss": 0.685,
+      "step": 4595
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.866983372921616e-05,
+      "loss": 0.5935,
+      "step": 4596
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.866389548693587e-05,
+      "loss": 0.8221,
+      "step": 4597
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.865795724465559e-05,
+      "loss": 0.6981,
+      "step": 4598
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.86520190023753e-05,
+      "loss": 0.6434,
+      "step": 4599
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.864608076009501e-05,
+      "loss": 0.5341,
+      "step": 4600
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.864014251781474e-05,
+      "loss": 0.6261,
+      "step": 4601
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.863420427553446e-05,
+      "loss": 0.8928,
+      "step": 4602
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.862826603325416e-05,
+      "loss": 0.5891,
+      "step": 4603
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.862232779097387e-05,
+      "loss": 0.564,
+      "step": 4604
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.861638954869359e-05,
+      "loss": 0.5577,
+      "step": 4605
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.86104513064133e-05,
+      "loss": 0.5587,
+      "step": 4606
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.860451306413302e-05,
+      "loss": 0.5085,
+      "step": 4607
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.859857482185274e-05,
+      "loss": 0.8857,
+      "step": 4608
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.859263657957246e-05,
+      "loss": 0.6481,
+      "step": 4609
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.858669833729216e-05,
+      "loss": 0.4945,
+      "step": 4610
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.858076009501188e-05,
+      "loss": 1.7738,
+      "step": 4611
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.857482185273159e-05,
+      "loss": 1.5641,
+      "step": 4612
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.856888361045131e-05,
+      "loss": 1.3064,
+      "step": 4613
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.856294536817103e-05,
+      "loss": 1.2306,
+      "step": 4614
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.855700712589074e-05,
+      "loss": 0.9185,
+      "step": 4615
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.855106888361046e-05,
+      "loss": 1.0558,
+      "step": 4616
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.854513064133017e-05,
+      "loss": 0.8216,
+      "step": 4617
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.853919239904989e-05,
+      "loss": 0.8625,
+      "step": 4618
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.853325415676959e-05,
+      "loss": 1.2552,
+      "step": 4619
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.852731591448931e-05,
+      "loss": 0.8743,
+      "step": 4620
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.852137767220903e-05,
+      "loss": 0.8925,
+      "step": 4621
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.851543942992875e-05,
+      "loss": 0.8166,
+      "step": 4622
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.850950118764846e-05,
+      "loss": 0.8096,
+      "step": 4623
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.850356294536817e-05,
+      "loss": 0.8806,
+      "step": 4624
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 7.849762470308789e-05,
+      "loss": 0.7788,
+      "step": 4625
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 7.84916864608076e-05,
+      "loss": 0.7864,
+      "step": 4626
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 7.848574821852732e-05,
+      "loss": 0.7141,
+      "step": 4627
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 7.847980997624704e-05,
+      "loss": 0.7279,
+      "step": 4628
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 7.847387173396676e-05,
+      "loss": 0.5736,
+      "step": 4629
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 7.846793349168646e-05,
+      "loss": 0.7591,
+      "step": 4630
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 7.846199524940618e-05,
+      "loss": 0.7153,
+      "step": 4631
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 7.845605700712589e-05,
+      "loss": 0.657,
+      "step": 4632
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 7.845011876484561e-05,
+      "loss": 0.9526,
+      "step": 4633
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.844418052256533e-05,
+      "loss": 0.7024,
+      "step": 4634
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.843824228028504e-05,
+      "loss": 0.5886,
+      "step": 4635
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.843230403800476e-05,
+      "loss": 0.7006,
+      "step": 4636
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.842636579572446e-05,
+      "loss": 0.6329,
+      "step": 4637
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.842042755344418e-05,
+      "loss": 0.6912,
+      "step": 4638
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.841448931116389e-05,
+      "loss": 0.6733,
+      "step": 4639
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.840855106888361e-05,
+      "loss": 0.5197,
+      "step": 4640
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.840261282660333e-05,
+      "loss": 0.6598,
+      "step": 4641
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.839667458432305e-05,
+      "loss": 0.5601,
+      "step": 4642
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.839073634204276e-05,
+      "loss": 0.6086,
+      "step": 4643
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.838479809976247e-05,
+      "loss": 0.7813,
+      "step": 4644
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.837885985748219e-05,
+      "loss": 0.624,
+      "step": 4645
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.83729216152019e-05,
+      "loss": 0.6178,
+      "step": 4646
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.836698337292163e-05,
+      "loss": 0.483,
+      "step": 4647
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.836104513064133e-05,
+      "loss": 0.5887,
+      "step": 4648
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.835510688836105e-05,
+      "loss": 0.5937,
+      "step": 4649
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.834916864608076e-05,
+      "loss": 0.7163,
+      "step": 4650
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 7.834323040380048e-05,
+      "loss": 0.555,
+      "step": 4651
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.833729216152019e-05,
+      "loss": 0.7016,
+      "step": 4652
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.833135391923991e-05,
+      "loss": 0.5651,
+      "step": 4653
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.832541567695963e-05,
+      "loss": 0.4805,
+      "step": 4654
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.831947743467934e-05,
+      "loss": 0.4801,
+      "step": 4655
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.831353919239906e-05,
+      "loss": 0.5383,
+      "step": 4656
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.830760095011876e-05,
+      "loss": 0.9634,
+      "step": 4657
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.830166270783848e-05,
+      "loss": 0.4757,
+      "step": 4658
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.829572446555819e-05,
+      "loss": 0.5355,
+      "step": 4659
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.828978622327791e-05,
+      "loss": 0.7397,
+      "step": 4660
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.828384798099763e-05,
+      "loss": 1.8765,
+      "step": 4661
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.827790973871735e-05,
+      "loss": 1.7028,
+      "step": 4662
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.827197149643706e-05,
+      "loss": 1.4712,
+      "step": 4663
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.826603325415678e-05,
+      "loss": 1.3033,
+      "step": 4664
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.826009501187649e-05,
+      "loss": 0.9895,
+      "step": 4665
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.825415676959619e-05,
+      "loss": 1.1071,
+      "step": 4666
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.824821852731593e-05,
+      "loss": 0.7024,
+      "step": 4667
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.824228028503563e-05,
+      "loss": 1.0086,
+      "step": 4668
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.823634204275535e-05,
+      "loss": 1.227,
+      "step": 4669
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7.823040380047506e-05,
+      "loss": 1.2835,
+      "step": 4670
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7.822446555819478e-05,
+      "loss": 0.8393,
+      "step": 4671
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7.821852731591449e-05,
+      "loss": 1.0012,
+      "step": 4672
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7.821258907363421e-05,
+      "loss": 1.0188,
+      "step": 4673
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7.820665083135393e-05,
+      "loss": 0.8597,
+      "step": 4674
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7.820071258907363e-05,
+      "loss": 0.5489,
+      "step": 4675
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7.819477434679336e-05,
+      "loss": 0.6044,
+      "step": 4676
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7.818883610451306e-05,
+      "loss": 0.5991,
+      "step": 4677
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7.818289786223278e-05,
+      "loss": 0.685,
+      "step": 4678
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.817695961995249e-05,
+      "loss": 0.7185,
+      "step": 4679
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.817102137767221e-05,
+      "loss": 0.6277,
+      "step": 4680
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.816508313539193e-05,
+      "loss": 0.6476,
+      "step": 4681
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.815914489311165e-05,
+      "loss": 0.8078,
+      "step": 4682
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.815320665083136e-05,
+      "loss": 0.723,
+      "step": 4683
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.814726840855108e-05,
+      "loss": 0.5774,
+      "step": 4684
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.814133016627078e-05,
+      "loss": 0.6522,
+      "step": 4685
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.813539192399049e-05,
+      "loss": 0.5909,
+      "step": 4686
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.812945368171022e-05,
+      "loss": 0.6642,
+      "step": 4687
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.812351543942993e-05,
+      "loss": 0.6627,
+      "step": 4688
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.811757719714965e-05,
+      "loss": 0.7796,
+      "step": 4689
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.811163895486936e-05,
+      "loss": 0.5896,
+      "step": 4690
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.810570071258908e-05,
+      "loss": 0.7099,
+      "step": 4691
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.809976247030879e-05,
+      "loss": 0.7055,
+      "step": 4692
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.80938242280285e-05,
+      "loss": 0.6108,
+      "step": 4693
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.808788598574823e-05,
+      "loss": 0.6414,
+      "step": 4694
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.808194774346795e-05,
+      "loss": 0.627,
+      "step": 4695
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.807600950118765e-05,
+      "loss": 0.6105,
+      "step": 4696
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.807007125890736e-05,
+      "loss": 0.677,
+      "step": 4697
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.806413301662708e-05,
+      "loss": 0.6199,
+      "step": 4698
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.805819477434679e-05,
+      "loss": 0.6119,
+      "step": 4699
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.805225653206651e-05,
+      "loss": 0.641,
+      "step": 4700
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.804631828978623e-05,
+      "loss": 0.6146,
+      "step": 4701
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.804038004750595e-05,
+      "loss": 0.5106,
+      "step": 4702
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.803444180522566e-05,
+      "loss": 0.6077,
+      "step": 4703
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.802850356294538e-05,
+      "loss": 0.6976,
+      "step": 4704
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.802256532066508e-05,
+      "loss": 0.5706,
+      "step": 4705
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.801662707838479e-05,
+      "loss": 0.4838,
+      "step": 4706
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.801068883610452e-05,
+      "loss": 0.5706,
+      "step": 4707
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.800475059382423e-05,
+      "loss": 0.5432,
+      "step": 4708
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.799881235154395e-05,
+      "loss": 0.6045,
+      "step": 4709
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.799287410926366e-05,
+      "loss": 0.5052,
+      "step": 4710
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.798693586698338e-05,
+      "loss": 1.6935,
+      "step": 4711
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.798099762470308e-05,
+      "loss": 1.5023,
+      "step": 4712
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.79750593824228e-05,
+      "loss": 1.5105,
+      "step": 4713
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.796912114014253e-05,
+      "loss": 1.3067,
+      "step": 4714
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.796318289786225e-05,
+      "loss": 0.85,
+      "step": 4715
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.795724465558195e-05,
+      "loss": 0.773,
+      "step": 4716
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.795130641330166e-05,
+      "loss": 0.9686,
+      "step": 4717
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.794536817102138e-05,
+      "loss": 0.7351,
+      "step": 4718
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.793942992874109e-05,
+      "loss": 0.9602,
+      "step": 4719
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.793349168646081e-05,
+      "loss": 1.084,
+      "step": 4720
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.792755344418053e-05,
+      "loss": 0.9455,
+      "step": 4721
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.792161520190025e-05,
+      "loss": 0.985,
+      "step": 4722
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.791567695961995e-05,
+      "loss": 0.8932,
+      "step": 4723
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.790973871733967e-05,
+      "loss": 0.7195,
+      "step": 4724
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.790380047505938e-05,
+      "loss": 0.758,
+      "step": 4725
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.78978622327791e-05,
+      "loss": 0.6823,
+      "step": 4726
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.789192399049882e-05,
+      "loss": 0.8307,
+      "step": 4727
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.788598574821853e-05,
+      "loss": 0.6471,
+      "step": 4728
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.788004750593825e-05,
+      "loss": 0.7492,
+      "step": 4729
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.787410926365796e-05,
+      "loss": 0.7038,
+      "step": 4730
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.786817102137768e-05,
+      "loss": 0.8223,
+      "step": 4731
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.786223277909738e-05,
+      "loss": 0.6315,
+      "step": 4732
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.78562945368171e-05,
+      "loss": 0.6236,
+      "step": 4733
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.785035629453682e-05,
+      "loss": 0.7828,
+      "step": 4734
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.784441805225654e-05,
+      "loss": 0.6479,
+      "step": 4735
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.783847980997625e-05,
+      "loss": 0.7112,
+      "step": 4736
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.783254156769597e-05,
+      "loss": 0.6322,
+      "step": 4737
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.782660332541568e-05,
+      "loss": 0.6797,
+      "step": 4738
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.782066508313539e-05,
+      "loss": 0.6639,
+      "step": 4739
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.78147268408551e-05,
+      "loss": 0.6837,
+      "step": 4740
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 7.780878859857483e-05,
+      "loss": 0.9256,
+      "step": 4741
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 7.780285035629455e-05,
+      "loss": 0.7576,
+      "step": 4742
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 7.779691211401425e-05,
+      "loss": 0.5856,
+      "step": 4743
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 7.779097387173397e-05,
+      "loss": 0.7765,
+      "step": 4744
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 7.778503562945368e-05,
+      "loss": 0.692,
+      "step": 4745
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 7.77790973871734e-05,
+      "loss": 0.6812,
+      "step": 4746
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 7.777315914489312e-05,
+      "loss": 0.6445,
+      "step": 4747
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 7.776722090261283e-05,
+      "loss": 0.6441,
+      "step": 4748
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 7.776128266033255e-05,
+      "loss": 0.7282,
+      "step": 4749
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.775534441805226e-05,
+      "loss": 0.5241,
+      "step": 4750
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.774940617577198e-05,
+      "loss": 0.6345,
+      "step": 4751
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.774346793349168e-05,
+      "loss": 0.4409,
+      "step": 4752
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.77375296912114e-05,
+      "loss": 0.524,
+      "step": 4753
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.773159144893112e-05,
+      "loss": 0.5498,
+      "step": 4754
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.772565320665084e-05,
+      "loss": 0.5537,
+      "step": 4755
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.771971496437055e-05,
+      "loss": 0.5242,
+      "step": 4756
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.771377672209027e-05,
+      "loss": 0.536,
+      "step": 4757
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.770783847980998e-05,
+      "loss": 0.4892,
+      "step": 4758
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.770190023752968e-05,
+      "loss": 0.3922,
+      "step": 4759
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.76959619952494e-05,
+      "loss": 0.2496,
+      "step": 4760
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.769002375296912e-05,
+      "loss": 1.554,
+      "step": 4761
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.768408551068885e-05,
+      "loss": 1.7138,
+      "step": 4762
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.767814726840855e-05,
+      "loss": 1.3567,
+      "step": 4763
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.767220902612827e-05,
+      "loss": 1.2607,
+      "step": 4764
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.766627078384798e-05,
+      "loss": 0.8187,
+      "step": 4765
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.76603325415677e-05,
+      "loss": 0.8453,
+      "step": 4766
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.765439429928742e-05,
+      "loss": 0.7431,
+      "step": 4767
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 7.764845605700714e-05,
+      "loss": 0.7964,
+      "step": 4768
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 7.764251781472685e-05,
+      "loss": 0.7758,
+      "step": 4769
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 7.763657957244655e-05,
+      "loss": 0.9259,
+      "step": 4770
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 7.763064133016627e-05,
+      "loss": 0.7975,
+      "step": 4771
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 7.762470308788598e-05,
+      "loss": 0.7631,
+      "step": 4772
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 7.76187648456057e-05,
+      "loss": 0.8871,
+      "step": 4773
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 7.761282660332542e-05,
+      "loss": 0.7097,
+      "step": 4774
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 7.760688836104514e-05,
+      "loss": 0.6378,
+      "step": 4775
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 7.760095011876485e-05,
+      "loss": 0.9234,
+      "step": 4776
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.759501187648457e-05,
+      "loss": 0.7821,
+      "step": 4777
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.758907363420428e-05,
+      "loss": 0.663,
+      "step": 4778
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.758313539192398e-05,
+      "loss": 0.8156,
+      "step": 4779
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.757719714964372e-05,
+      "loss": 0.8812,
+      "step": 4780
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.757125890736342e-05,
+      "loss": 0.748,
+      "step": 4781
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.756532066508314e-05,
+      "loss": 0.7829,
+      "step": 4782
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.755938242280285e-05,
+      "loss": 0.5773,
+      "step": 4783
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.755344418052257e-05,
+      "loss": 0.5403,
+      "step": 4784
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.754750593824228e-05,
+      "loss": 0.6415,
+      "step": 4785
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.7541567695962e-05,
+      "loss": 0.6253,
+      "step": 4786
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.753562945368172e-05,
+      "loss": 0.6335,
+      "step": 4787
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.752969121140144e-05,
+      "loss": 0.5925,
+      "step": 4788
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.752375296912115e-05,
+      "loss": 0.5303,
+      "step": 4789
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.751781472684085e-05,
+      "loss": 0.6616,
+      "step": 4790
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.751187648456057e-05,
+      "loss": 0.6585,
+      "step": 4791
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.750593824228028e-05,
+      "loss": 0.5591,
+      "step": 4792
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.75e-05,
+      "loss": 0.6527,
+      "step": 4793
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.749406175771972e-05,
+      "loss": 0.4539,
+      "step": 4794
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.748812351543944e-05,
+      "loss": 0.7251,
+      "step": 4795
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.748218527315915e-05,
+      "loss": 0.5745,
+      "step": 4796
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.747624703087887e-05,
+      "loss": 0.5576,
+      "step": 4797
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.747030878859857e-05,
+      "loss": 0.4427,
+      "step": 4798
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.74643705463183e-05,
+      "loss": 0.6776,
+      "step": 4799
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.745843230403802e-05,
+      "loss": 0.417,
+      "step": 4800
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.745249406175772e-05,
+      "loss": 0.6103,
+      "step": 4801
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.744655581947744e-05,
+      "loss": 0.5117,
+      "step": 4802
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.744061757719715e-05,
+      "loss": 0.4934,
+      "step": 4803
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.743467933491687e-05,
+      "loss": 0.5388,
+      "step": 4804
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.742874109263658e-05,
+      "loss": 0.7462,
+      "step": 4805
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.74228028503563e-05,
+      "loss": 0.4935,
+      "step": 4806
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.741686460807602e-05,
+      "loss": 0.7648,
+      "step": 4807
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.741092636579574e-05,
+      "loss": 0.9027,
+      "step": 4808
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.740498812351544e-05,
+      "loss": 0.5381,
+      "step": 4809
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.739904988123515e-05,
+      "loss": 0.3809,
+      "step": 4810
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.739311163895487e-05,
+      "loss": 1.5663,
+      "step": 4811
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.738717339667458e-05,
+      "loss": 1.294,
+      "step": 4812
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.73812351543943e-05,
+      "loss": 1.3836,
+      "step": 4813
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.737529691211402e-05,
+      "loss": 1.1928,
+      "step": 4814
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.736935866983374e-05,
+      "loss": 0.7996,
+      "step": 4815
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.736342042755345e-05,
+      "loss": 0.7284,
+      "step": 4816
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.735748218527317e-05,
+      "loss": 0.9098,
+      "step": 4817
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.735154394299287e-05,
+      "loss": 0.6878,
+      "step": 4818
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.73456057007126e-05,
+      "loss": 0.832,
+      "step": 4819
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.733966745843231e-05,
+      "loss": 1.0463,
+      "step": 4820
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.733372921615202e-05,
+      "loss": 0.8748,
+      "step": 4821
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.732779097387174e-05,
+      "loss": 0.6952,
+      "step": 4822
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.732185273159145e-05,
+      "loss": 1.1797,
+      "step": 4823
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.731591448931117e-05,
+      "loss": 0.8015,
+      "step": 4824
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.730997624703088e-05,
+      "loss": 0.6749,
+      "step": 4825
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.73040380047506e-05,
+      "loss": 0.7582,
+      "step": 4826
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.729809976247032e-05,
+      "loss": 0.66,
+      "step": 4827
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.729216152019004e-05,
+      "loss": 0.6404,
+      "step": 4828
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.728622327790974e-05,
+      "loss": 0.7106,
+      "step": 4829
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.728028503562946e-05,
+      "loss": 0.6131,
+      "step": 4830
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.727434679334917e-05,
+      "loss": 0.7647,
+      "step": 4831
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.726840855106888e-05,
+      "loss": 0.7138,
+      "step": 4832
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.72624703087886e-05,
+      "loss": 0.7288,
+      "step": 4833
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.725653206650832e-05,
+      "loss": 0.6685,
+      "step": 4834
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.725059382422804e-05,
+      "loss": 0.6055,
+      "step": 4835
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.724465558194774e-05,
+      "loss": 0.8388,
+      "step": 4836
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.723871733966747e-05,
+      "loss": 0.6239,
+      "step": 4837
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.723277909738717e-05,
+      "loss": 0.6456,
+      "step": 4838
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.722684085510689e-05,
+      "loss": 0.6585,
+      "step": 4839
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.722090261282661e-05,
+      "loss": 0.6108,
+      "step": 4840
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.721496437054632e-05,
+      "loss": 0.5952,
+      "step": 4841
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.720902612826604e-05,
+      "loss": 0.519,
+      "step": 4842
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.720308788598575e-05,
+      "loss": 0.6074,
+      "step": 4843
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.719714964370547e-05,
+      "loss": 0.5882,
+      "step": 4844
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.719121140142517e-05,
+      "loss": 0.8845,
+      "step": 4845
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.71852731591449e-05,
+      "loss": 0.6642,
+      "step": 4846
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.717933491686461e-05,
+      "loss": 0.5277,
+      "step": 4847
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.717339667458434e-05,
+      "loss": 0.556,
+      "step": 4848
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.716745843230404e-05,
+      "loss": 0.4313,
+      "step": 4849
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.716152019002376e-05,
+      "loss": 0.4528,
+      "step": 4850
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.715558194774347e-05,
+      "loss": 0.9631,
+      "step": 4851
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.714964370546318e-05,
+      "loss": 0.5339,
+      "step": 4852
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.71437054631829e-05,
+      "loss": 0.6266,
+      "step": 4853
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.713776722090262e-05,
+      "loss": 0.5406,
+      "step": 4854
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.713182897862234e-05,
+      "loss": 0.4054,
+      "step": 4855
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.712589073634204e-05,
+      "loss": 0.4908,
+      "step": 4856
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.711995249406176e-05,
+      "loss": 1.1578,
+      "step": 4857
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.711401425178147e-05,
+      "loss": 0.4622,
+      "step": 4858
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.710807600950119e-05,
+      "loss": 0.6329,
+      "step": 4859
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.710213776722091e-05,
+      "loss": 0.7352,
+      "step": 4860
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.709619952494063e-05,
+      "loss": 1.7357,
+      "step": 4861
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.709026128266034e-05,
+      "loss": 1.2011,
+      "step": 4862
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.708432304038005e-05,
+      "loss": 1.2551,
+      "step": 4863
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.707838479809977e-05,
+      "loss": 1.0063,
+      "step": 4864
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.707244655581947e-05,
+      "loss": 0.7546,
+      "step": 4865
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.706650831353919e-05,
+      "loss": 0.8514,
+      "step": 4866
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.706057007125891e-05,
+      "loss": 0.7219,
+      "step": 4867
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.705463182897863e-05,
+      "loss": 0.8204,
+      "step": 4868
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.704869358669834e-05,
+      "loss": 0.8062,
+      "step": 4869
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.704275534441806e-05,
+      "loss": 0.8819,
+      "step": 4870
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.703681710213777e-05,
+      "loss": 0.8111,
+      "step": 4871
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.703087885985747e-05,
+      "loss": 1.0094,
+      "step": 4872
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.70249406175772e-05,
+      "loss": 0.6545,
+      "step": 4873
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.701900237529692e-05,
+      "loss": 0.6442,
+      "step": 4874
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.701306413301664e-05,
+      "loss": 0.6786,
+      "step": 4875
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.700712589073634e-05,
+      "loss": 0.8413,
+      "step": 4876
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.700118764845606e-05,
+      "loss": 0.8215,
+      "step": 4877
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.699524940617577e-05,
+      "loss": 0.7251,
+      "step": 4878
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.698931116389549e-05,
+      "loss": 0.5847,
+      "step": 4879
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.698337292161521e-05,
+      "loss": 0.6632,
+      "step": 4880
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.697743467933493e-05,
+      "loss": 0.5428,
+      "step": 4881
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.697149643705464e-05,
+      "loss": 0.6013,
+      "step": 4882
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.696555819477434e-05,
+      "loss": 0.5203,
+      "step": 4883
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.695961995249406e-05,
+      "loss": 0.8905,
+      "step": 4884
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.695368171021377e-05,
+      "loss": 0.6986,
+      "step": 4885
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.694774346793349e-05,
+      "loss": 0.6096,
+      "step": 4886
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.694180522565321e-05,
+      "loss": 0.6205,
+      "step": 4887
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.693586698337293e-05,
+      "loss": 0.605,
+      "step": 4888
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.692992874109264e-05,
+      "loss": 0.6867,
+      "step": 4889
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.692399049881236e-05,
+      "loss": 0.6317,
+      "step": 4890
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.691805225653207e-05,
+      "loss": 0.6423,
+      "step": 4891
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.691211401425179e-05,
+      "loss": 0.6327,
+      "step": 4892
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.690617577197151e-05,
+      "loss": 0.6657,
+      "step": 4893
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.690023752969121e-05,
+      "loss": 0.6961,
+      "step": 4894
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.689429928741093e-05,
+      "loss": 0.5155,
+      "step": 4895
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.688836104513064e-05,
+      "loss": 0.7822,
+      "step": 4896
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.688242280285036e-05,
+      "loss": 0.4686,
+      "step": 4897
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.687648456057007e-05,
+      "loss": 0.5549,
+      "step": 4898
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.687054631828979e-05,
+      "loss": 0.4234,
+      "step": 4899
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.686460807600951e-05,
+      "loss": 0.7757,
+      "step": 4900
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.685866983372923e-05,
+      "loss": 0.6889,
+      "step": 4901
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.685273159144894e-05,
+      "loss": 0.6266,
+      "step": 4902
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.684679334916864e-05,
+      "loss": 0.6194,
+      "step": 4903
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.684085510688836e-05,
+      "loss": 0.5308,
+      "step": 4904
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.683491686460807e-05,
+      "loss": 0.3407,
+      "step": 4905
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.682897862232779e-05,
+      "loss": 0.5318,
+      "step": 4906
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.682304038004751e-05,
+      "loss": 0.3646,
+      "step": 4907
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.681710213776723e-05,
+      "loss": 0.5081,
+      "step": 4908
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.681116389548694e-05,
+      "loss": 0.3699,
+      "step": 4909
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.680522565320666e-05,
+      "loss": 0.3222,
+      "step": 4910
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.679928741092637e-05,
+      "loss": 1.7242,
+      "step": 4911
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.679334916864609e-05,
+      "loss": 1.2979,
+      "step": 4912
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.67874109263658e-05,
+      "loss": 1.1145,
+      "step": 4913
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.678147268408551e-05,
+      "loss": 1.1375,
+      "step": 4914
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.677553444180523e-05,
+      "loss": 0.8943,
+      "step": 4915
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.676959619952494e-05,
+      "loss": 0.7355,
+      "step": 4916
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.676365795724466e-05,
+      "loss": 0.7212,
+      "step": 4917
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.675771971496437e-05,
+      "loss": 0.7255,
+      "step": 4918
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.675178147268409e-05,
+      "loss": 0.7467,
+      "step": 4919
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.674584323040381e-05,
+      "loss": 0.794,
+      "step": 4920
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.673990498812353e-05,
+      "loss": 0.8499,
+      "step": 4921
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.673396674584323e-05,
+      "loss": 0.7386,
+      "step": 4922
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.672802850356296e-05,
+      "loss": 0.6752,
+      "step": 4923
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.672209026128266e-05,
+      "loss": 0.6473,
+      "step": 4924
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.671615201900237e-05,
+      "loss": 0.5446,
+      "step": 4925
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.671021377672209e-05,
+      "loss": 0.5725,
+      "step": 4926
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.670427553444181e-05,
+      "loss": 0.6837,
+      "step": 4927
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.669833729216153e-05,
+      "loss": 0.6835,
+      "step": 4928
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.669239904988124e-05,
+      "loss": 0.6041,
+      "step": 4929
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.668646080760096e-05,
+      "loss": 0.6523,
+      "step": 4930
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.668052256532066e-05,
+      "loss": 0.5901,
+      "step": 4931
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.667458432304038e-05,
+      "loss": 0.8288,
+      "step": 4932
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.66686460807601e-05,
+      "loss": 0.5303,
+      "step": 4933
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.666270783847982e-05,
+      "loss": 0.6203,
+      "step": 4934
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.665676959619953e-05,
+      "loss": 0.6989,
+      "step": 4935
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.665083135391924e-05,
+      "loss": 0.6126,
+      "step": 4936
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.664489311163896e-05,
+      "loss": 0.6785,
+      "step": 4937
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.663895486935867e-05,
+      "loss": 0.5043,
+      "step": 4938
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.663301662707839e-05,
+      "loss": 0.6006,
+      "step": 4939
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.66270783847981e-05,
+      "loss": 0.7414,
+      "step": 4940
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.662114014251783e-05,
+      "loss": 0.5692,
+      "step": 4941
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.661520190023753e-05,
+      "loss": 0.56,
+      "step": 4942
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.660926365795725e-05,
+      "loss": 0.5557,
+      "step": 4943
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.660332541567696e-05,
+      "loss": 0.5069,
+      "step": 4944
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.659738717339667e-05,
+      "loss": 0.5119,
+      "step": 4945
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.659144893111639e-05,
+      "loss": 0.4159,
+      "step": 4946
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.658551068883611e-05,
+      "loss": 0.406,
+      "step": 4947
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.657957244655583e-05,
+      "loss": 0.5558,
+      "step": 4948
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.657363420427554e-05,
+      "loss": 0.4267,
+      "step": 4949
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.656769596199526e-05,
+      "loss": 0.7039,
+      "step": 4950
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.656175771971496e-05,
+      "loss": 0.4416,
+      "step": 4951
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.655581947743468e-05,
+      "loss": 0.4528,
+      "step": 4952
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.65498812351544e-05,
+      "loss": 0.4249,
+      "step": 4953
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.654394299287412e-05,
+      "loss": 0.7221,
+      "step": 4954
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.653800475059383e-05,
+      "loss": 0.4654,
+      "step": 4955
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.653206650831354e-05,
+      "loss": 0.6177,
+      "step": 4956
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.652612826603326e-05,
+      "loss": 0.4884,
+      "step": 4957
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.652019002375296e-05,
+      "loss": 1.0012,
+      "step": 4958
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.651425178147268e-05,
+      "loss": 0.5563,
+      "step": 4959
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.65083135391924e-05,
+      "loss": 0.4538,
+      "step": 4960
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.650237529691213e-05,
+      "loss": 1.6769,
+      "step": 4961
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.649643705463183e-05,
+      "loss": 1.4679,
+      "step": 4962
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.649049881235155e-05,
+      "loss": 1.3511,
+      "step": 4963
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.648456057007126e-05,
+      "loss": 1.1056,
+      "step": 4964
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.647862232779098e-05,
+      "loss": 0.8921,
+      "step": 4965
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.647268408551069e-05,
+      "loss": 0.7675,
+      "step": 4966
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.646674584323041e-05,
+      "loss": 0.7767,
+      "step": 4967
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.646080760095013e-05,
+      "loss": 0.7668,
+      "step": 4968
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.645486935866983e-05,
+      "loss": 0.7521,
+      "step": 4969
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.644893111638955e-05,
+      "loss": 0.9517,
+      "step": 4970
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.644299287410926e-05,
+      "loss": 0.9307,
+      "step": 4971
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.643705463182898e-05,
+      "loss": 0.723,
+      "step": 4972
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.64311163895487e-05,
+      "loss": 0.8506,
+      "step": 4973
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.642517814726842e-05,
+      "loss": 0.654,
+      "step": 4974
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.641923990498813e-05,
+      "loss": 0.7094,
+      "step": 4975
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.641330166270784e-05,
+      "loss": 0.4686,
+      "step": 4976
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.640736342042756e-05,
+      "loss": 0.5098,
+      "step": 4977
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.640142517814726e-05,
+      "loss": 0.5369,
+      "step": 4978
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.639548693586698e-05,
+      "loss": 0.6283,
+      "step": 4979
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.63895486935867e-05,
+      "loss": 0.6635,
+      "step": 4980
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.638361045130642e-05,
+      "loss": 0.8237,
+      "step": 4981
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.637767220902613e-05,
+      "loss": 0.6006,
+      "step": 4982
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.637173396674585e-05,
+      "loss": 0.5443,
+      "step": 4983
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.636579572446556e-05,
+      "loss": 0.5139,
+      "step": 4984
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.635985748218528e-05,
+      "loss": 0.6805,
+      "step": 4985
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.635391923990499e-05,
+      "loss": 0.6729,
+      "step": 4986
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.63479809976247e-05,
+      "loss": 0.6289,
+      "step": 4987
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.634204275534443e-05,
+      "loss": 0.7665,
+      "step": 4988
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.633610451306413e-05,
+      "loss": 0.6437,
+      "step": 4989
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.633016627078385e-05,
+      "loss": 0.5488,
+      "step": 4990
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 7.632422802850356e-05,
+      "loss": 0.7209,
+      "step": 4991
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 7.631828978622328e-05,
+      "loss": 0.734,
+      "step": 4992
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 7.6312351543943e-05,
+      "loss": 0.5727,
+      "step": 4993
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 7.630641330166272e-05,
+      "loss": 0.6145,
+      "step": 4994
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 7.630047505938243e-05,
+      "loss": 0.5239,
+      "step": 4995
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 7.629453681710215e-05,
+      "loss": 0.5067,
+      "step": 4996
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 7.628859857482186e-05,
+      "loss": 0.5007,
+      "step": 4997
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 7.628266033254156e-05,
+      "loss": 0.5268,
+      "step": 4998
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 7.627672209026128e-05,
+      "loss": 0.3863,
+      "step": 4999
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.6270783847981e-05,
+      "loss": 0.5145,
+      "step": 5000
+    },
+    {
+      "epoch": 5.61,
+      "eval_loss": 1.8941514492034912,
+      "eval_runtime": 910.08,
+      "eval_samples_per_second": 2.903,
+      "eval_steps_per_second": 0.364,
+      "eval_wer": 1.1062871876239588,
+      "step": 5000
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.626484560570072e-05,
+      "loss": 0.5601,
+      "step": 5001
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.625890736342043e-05,
+      "loss": 0.6544,
+      "step": 5002
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.625296912114015e-05,
+      "loss": 0.5486,
+      "step": 5003
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.624703087885986e-05,
+      "loss": 0.5057,
+      "step": 5004
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.624109263657958e-05,
+      "loss": 0.5636,
+      "step": 5005
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.623515439429928e-05,
+      "loss": 0.684,
+      "step": 5006
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.6229216152019e-05,
+      "loss": 0.5771,
+      "step": 5007
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.622327790973872e-05,
+      "loss": 0.5277,
+      "step": 5008
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.621733966745843e-05,
+      "loss": 0.5048,
+      "step": 5009
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.621140142517815e-05,
+      "loss": 0.43,
+      "step": 5010
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.620546318289786e-05,
+      "loss": 1.5982,
+      "step": 5011
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.619952494061758e-05,
+      "loss": 1.1992,
+      "step": 5012
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.61935866983373e-05,
+      "loss": 1.1626,
+      "step": 5013
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.618764845605702e-05,
+      "loss": 0.9793,
+      "step": 5014
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.618171021377673e-05,
+      "loss": 1.0061,
+      "step": 5015
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.617577197149645e-05,
+      "loss": 0.7181,
+      "step": 5016
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.616983372921615e-05,
+      "loss": 0.8737,
+      "step": 5017
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 7.616389548693586e-05,
+      "loss": 0.7497,
+      "step": 5018
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 7.615795724465558e-05,
+      "loss": 0.649,
+      "step": 5019
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 7.61520190023753e-05,
+      "loss": 1.0213,
+      "step": 5020
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 7.614608076009502e-05,
+      "loss": 0.8558,
+      "step": 5021
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 7.614014251781473e-05,
+      "loss": 0.9594,
+      "step": 5022
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 7.613420427553445e-05,
+      "loss": 0.6446,
+      "step": 5023
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 7.612826603325416e-05,
+      "loss": 0.7929,
+      "step": 5024
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 7.612232779097388e-05,
+      "loss": 0.6409,
+      "step": 5025
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 7.61163895486936e-05,
+      "loss": 0.6124,
+      "step": 5026
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.611045130641332e-05,
+      "loss": 0.7524,
+      "step": 5027
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.610451306413302e-05,
+      "loss": 0.5834,
+      "step": 5028
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.609857482185273e-05,
+      "loss": 0.6123,
+      "step": 5029
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.609263657957245e-05,
+      "loss": 0.6442,
+      "step": 5030
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.608669833729216e-05,
+      "loss": 0.6027,
+      "step": 5031
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.608076009501188e-05,
+      "loss": 0.6022,
+      "step": 5032
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.60748218527316e-05,
+      "loss": 0.5875,
+      "step": 5033
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.606888361045132e-05,
+      "loss": 0.6221,
+      "step": 5034
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.606294536817103e-05,
+      "loss": 0.7222,
+      "step": 5035
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.605700712589075e-05,
+      "loss": 0.5796,
+      "step": 5036
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.605106888361045e-05,
+      "loss": 0.5849,
+      "step": 5037
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.604513064133016e-05,
+      "loss": 0.5272,
+      "step": 5038
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.603919239904988e-05,
+      "loss": 0.4848,
+      "step": 5039
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.60332541567696e-05,
+      "loss": 0.7861,
+      "step": 5040
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.602731591448932e-05,
+      "loss": 0.5054,
+      "step": 5041
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.602137767220903e-05,
+      "loss": 0.5308,
+      "step": 5042
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.601543942992875e-05,
+      "loss": 0.6574,
+      "step": 5043
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.600950118764845e-05,
+      "loss": 0.5516,
+      "step": 5044
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.600356294536817e-05,
+      "loss": 0.5306,
+      "step": 5045
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.59976247030879e-05,
+      "loss": 0.5036,
+      "step": 5046
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.599168646080762e-05,
+      "loss": 0.5408,
+      "step": 5047
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.598574821852732e-05,
+      "loss": 0.5514,
+      "step": 5048
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.597980997624703e-05,
+      "loss": 0.4788,
+      "step": 5049
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.597387173396675e-05,
+      "loss": 0.4716,
+      "step": 5050
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.596793349168646e-05,
+      "loss": 0.5342,
+      "step": 5051
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.596199524940618e-05,
+      "loss": 0.6632,
+      "step": 5052
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.59560570071259e-05,
+      "loss": 0.705,
+      "step": 5053
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.595011876484562e-05,
+      "loss": 0.4479,
+      "step": 5054
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.594418052256532e-05,
+      "loss": 0.5072,
+      "step": 5055
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.593824228028504e-05,
+      "loss": 0.7553,
+      "step": 5056
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.593230403800475e-05,
+      "loss": 0.504,
+      "step": 5057
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.592636579572447e-05,
+      "loss": 0.6814,
+      "step": 5058
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.592042755344418e-05,
+      "loss": 0.5032,
+      "step": 5059
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.59144893111639e-05,
+      "loss": 0.3332,
+      "step": 5060
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.590855106888362e-05,
+      "loss": 1.3411,
+      "step": 5061
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.590261282660333e-05,
+      "loss": 1.1858,
+      "step": 5062
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.589667458432305e-05,
+      "loss": 1.0193,
+      "step": 5063
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.589073634204275e-05,
+      "loss": 0.7988,
+      "step": 5064
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.588479809976247e-05,
+      "loss": 0.7661,
+      "step": 5065
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.58788598574822e-05,
+      "loss": 0.7839,
+      "step": 5066
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.587292161520191e-05,
+      "loss": 0.8674,
+      "step": 5067
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.586698337292162e-05,
+      "loss": 0.8874,
+      "step": 5068
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.586104513064133e-05,
+      "loss": 0.6142,
+      "step": 5069
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.585510688836105e-05,
+      "loss": 0.7377,
+      "step": 5070
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.584916864608075e-05,
+      "loss": 0.6736,
+      "step": 5071
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.584323040380048e-05,
+      "loss": 0.735,
+      "step": 5072
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.58372921615202e-05,
+      "loss": 0.6263,
+      "step": 5073
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.583135391923992e-05,
+      "loss": 0.7225,
+      "step": 5074
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.582541567695962e-05,
+      "loss": 0.698,
+      "step": 5075
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.581947743467934e-05,
+      "loss": 0.6067,
+      "step": 5076
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.581353919239905e-05,
+      "loss": 0.5796,
+      "step": 5077
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.580760095011877e-05,
+      "loss": 0.6577,
+      "step": 5078
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.580166270783848e-05,
+      "loss": 0.666,
+      "step": 5079
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.57957244655582e-05,
+      "loss": 0.6208,
+      "step": 5080
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.578978622327792e-05,
+      "loss": 0.4329,
+      "step": 5081
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.578384798099762e-05,
+      "loss": 0.569,
+      "step": 5082
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.577790973871734e-05,
+      "loss": 0.6778,
+      "step": 5083
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.577197149643705e-05,
+      "loss": 0.5929,
+      "step": 5084
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.576603325415677e-05,
+      "loss": 0.8113,
+      "step": 5085
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.576009501187649e-05,
+      "loss": 0.6481,
+      "step": 5086
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.575415676959621e-05,
+      "loss": 0.7227,
+      "step": 5087
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.574821852731592e-05,
+      "loss": 0.5508,
+      "step": 5088
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.574228028503564e-05,
+      "loss": 0.5722,
+      "step": 5089
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.573634204275535e-05,
+      "loss": 0.5103,
+      "step": 5090
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.573040380047505e-05,
+      "loss": 0.5604,
+      "step": 5091
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.572446555819477e-05,
+      "loss": 0.6192,
+      "step": 5092
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.57185273159145e-05,
+      "loss": 0.5171,
+      "step": 5093
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.571258907363421e-05,
+      "loss": 0.6843,
+      "step": 5094
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.570665083135392e-05,
+      "loss": 0.5605,
+      "step": 5095
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.570071258907364e-05,
+      "loss": 0.5338,
+      "step": 5096
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.569477434679335e-05,
+      "loss": 0.6112,
+      "step": 5097
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.568883610451307e-05,
+      "loss": 0.5692,
+      "step": 5098
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.568289786223278e-05,
+      "loss": 0.5122,
+      "step": 5099
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.56769596199525e-05,
+      "loss": 0.5494,
+      "step": 5100
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.567102137767222e-05,
+      "loss": 0.6011,
+      "step": 5101
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.566508313539192e-05,
+      "loss": 0.4709,
+      "step": 5102
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.565914489311164e-05,
+      "loss": 0.6304,
+      "step": 5103
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.565320665083135e-05,
+      "loss": 0.448,
+      "step": 5104
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.564726840855107e-05,
+      "loss": 0.6504,
+      "step": 5105
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.564133016627079e-05,
+      "loss": 0.5899,
+      "step": 5106
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.563539192399051e-05,
+      "loss": 0.3835,
+      "step": 5107
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.562945368171022e-05,
+      "loss": 0.5508,
+      "step": 5108
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.562351543942994e-05,
+      "loss": 0.6696,
+      "step": 5109
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.561757719714965e-05,
+      "loss": 0.6526,
+      "step": 5110
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.561163895486935e-05,
+      "loss": 1.2342,
+      "step": 5111
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.560570071258907e-05,
+      "loss": 1.3789,
+      "step": 5112
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.559976247030879e-05,
+      "loss": 1.2589,
+      "step": 5113
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.559382422802851e-05,
+      "loss": 0.8209,
+      "step": 5114
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.558788598574822e-05,
+      "loss": 0.8167,
+      "step": 5115
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.558194774346794e-05,
+      "loss": 0.6831,
+      "step": 5116
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.557600950118765e-05,
+      "loss": 0.6365,
+      "step": 5117
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.557007125890737e-05,
+      "loss": 0.7254,
+      "step": 5118
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.556413301662707e-05,
+      "loss": 0.6864,
+      "step": 5119
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.555819477434681e-05,
+      "loss": 0.6456,
+      "step": 5120
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.555225653206652e-05,
+      "loss": 0.6534,
+      "step": 5121
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.554631828978622e-05,
+      "loss": 0.7225,
+      "step": 5122
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.554038004750594e-05,
+      "loss": 0.5259,
+      "step": 5123
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.553444180522565e-05,
+      "loss": 0.6154,
+      "step": 5124
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.552850356294537e-05,
+      "loss": 0.607,
+      "step": 5125
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.552256532066509e-05,
+      "loss": 0.6308,
+      "step": 5126
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.551662707838481e-05,
+      "loss": 0.6074,
+      "step": 5127
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.551068883610452e-05,
+      "loss": 1.1934,
+      "step": 5128
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.550475059382424e-05,
+      "loss": 0.5307,
+      "step": 5129
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.549881235154394e-05,
+      "loss": 0.599,
+      "step": 5130
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.549287410926366e-05,
+      "loss": 0.585,
+      "step": 5131
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.548693586698337e-05,
+      "loss": 0.7206,
+      "step": 5132
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.548099762470309e-05,
+      "loss": 0.855,
+      "step": 5133
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.547505938242281e-05,
+      "loss": 0.5824,
+      "step": 5134
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.546912114014252e-05,
+      "loss": 0.5486,
+      "step": 5135
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.546318289786224e-05,
+      "loss": 0.5392,
+      "step": 5136
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.545724465558195e-05,
+      "loss": 0.5206,
+      "step": 5137
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.545130641330167e-05,
+      "loss": 0.6293,
+      "step": 5138
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.544536817102137e-05,
+      "loss": 0.6638,
+      "step": 5139
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.543942992874111e-05,
+      "loss": 0.6815,
+      "step": 5140
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.543349168646081e-05,
+      "loss": 0.4387,
+      "step": 5141
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.542755344418052e-05,
+      "loss": 0.6055,
+      "step": 5142
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.542161520190024e-05,
+      "loss": 0.6703,
+      "step": 5143
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.541567695961995e-05,
+      "loss": 0.5618,
+      "step": 5144
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.540973871733967e-05,
+      "loss": 0.7907,
+      "step": 5145
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.540380047505939e-05,
+      "loss": 0.6097,
+      "step": 5146
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.539786223277911e-05,
+      "loss": 0.5892,
+      "step": 5147
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.539192399049882e-05,
+      "loss": 0.3915,
+      "step": 5148
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.538598574821854e-05,
+      "loss": 0.5768,
+      "step": 5149
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.538004750593824e-05,
+      "loss": 0.5523,
+      "step": 5150
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.537410926365796e-05,
+      "loss": 0.6697,
+      "step": 5151
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.536817102137767e-05,
+      "loss": 0.5989,
+      "step": 5152
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.536223277909739e-05,
+      "loss": 0.5737,
+      "step": 5153
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.535629453681711e-05,
+      "loss": 0.3839,
+      "step": 5154
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.535035629453682e-05,
+      "loss": 0.4964,
+      "step": 5155
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.534441805225654e-05,
+      "loss": 0.5114,
+      "step": 5156
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.533847980997624e-05,
+      "loss": 0.4298,
+      "step": 5157
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.533254156769597e-05,
+      "loss": 0.5424,
+      "step": 5158
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.532660332541569e-05,
+      "loss": 0.4556,
+      "step": 5159
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.53206650831354e-05,
+      "loss": 0.3937,
+      "step": 5160
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.531472684085511e-05,
+      "loss": 1.5405,
+      "step": 5161
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.530878859857483e-05,
+      "loss": 1.2989,
+      "step": 5162
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.530285035629454e-05,
+      "loss": 1.0816,
+      "step": 5163
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.529691211401425e-05,
+      "loss": 1.1142,
+      "step": 5164
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.529097387173397e-05,
+      "loss": 0.9126,
+      "step": 5165
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.528503562945369e-05,
+      "loss": 0.6921,
+      "step": 5166
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.527909738717341e-05,
+      "loss": 0.7588,
+      "step": 5167
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.527315914489311e-05,
+      "loss": 0.5483,
+      "step": 5168
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.526722090261283e-05,
+      "loss": 0.8327,
+      "step": 5169
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.526128266033254e-05,
+      "loss": 0.8833,
+      "step": 5170
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.525534441805226e-05,
+      "loss": 0.6242,
+      "step": 5171
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.524940617577197e-05,
+      "loss": 0.6578,
+      "step": 5172
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.524346793349169e-05,
+      "loss": 0.6699,
+      "step": 5173
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.523752969121141e-05,
+      "loss": 0.7155,
+      "step": 5174
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.523159144893112e-05,
+      "loss": 0.6302,
+      "step": 5175
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.522565320665084e-05,
+      "loss": 0.6702,
+      "step": 5176
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.521971496437054e-05,
+      "loss": 0.6326,
+      "step": 5177
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.521377672209026e-05,
+      "loss": 0.703,
+      "step": 5178
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.520783847980998e-05,
+      "loss": 0.6193,
+      "step": 5179
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.52019002375297e-05,
+      "loss": 0.9393,
+      "step": 5180
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.519596199524941e-05,
+      "loss": 0.6419,
+      "step": 5181
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.519002375296913e-05,
+      "loss": 0.5117,
+      "step": 5182
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.518408551068884e-05,
+      "loss": 0.4841,
+      "step": 5183
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.517814726840855e-05,
+      "loss": 0.796,
+      "step": 5184
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.517220902612827e-05,
+      "loss": 0.4477,
+      "step": 5185
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.516627078384799e-05,
+      "loss": 0.5753,
+      "step": 5186
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.51603325415677e-05,
+      "loss": 0.7636,
+      "step": 5187
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.515439429928741e-05,
+      "loss": 0.5321,
+      "step": 5188
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.514845605700713e-05,
+      "loss": 0.4763,
+      "step": 5189
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.514251781472684e-05,
+      "loss": 0.4609,
+      "step": 5190
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.513657957244656e-05,
+      "loss": 0.5119,
+      "step": 5191
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.513064133016627e-05,
+      "loss": 0.4905,
+      "step": 5192
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.5124703087886e-05,
+      "loss": 0.472,
+      "step": 5193
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.511876484560571e-05,
+      "loss": 0.5649,
+      "step": 5194
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.511282660332542e-05,
+      "loss": 0.6134,
+      "step": 5195
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.510688836104514e-05,
+      "loss": 0.5883,
+      "step": 5196
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.510095011876484e-05,
+      "loss": 0.4316,
+      "step": 5197
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.509501187648456e-05,
+      "loss": 0.483,
+      "step": 5198
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.508907363420428e-05,
+      "loss": 0.5026,
+      "step": 5199
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.5083135391924e-05,
+      "loss": 0.5041,
+      "step": 5200
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.507719714964371e-05,
+      "loss": 0.4355,
+      "step": 5201
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.507125890736343e-05,
+      "loss": 0.5647,
+      "step": 5202
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.506532066508314e-05,
+      "loss": 0.4921,
+      "step": 5203
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.505938242280284e-05,
+      "loss": 0.545,
+      "step": 5204
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.505344418052256e-05,
+      "loss": 0.5666,
+      "step": 5205
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.504750593824228e-05,
+      "loss": 0.5772,
+      "step": 5206
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.5041567695962e-05,
+      "loss": 0.4785,
+      "step": 5207
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.503562945368171e-05,
+      "loss": 0.4429,
+      "step": 5208
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.502969121140143e-05,
+      "loss": 0.2962,
+      "step": 5209
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.502375296912114e-05,
+      "loss": 0.5431,
+      "step": 5210
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.501781472684086e-05,
+      "loss": 1.4804,
+      "step": 5211
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.501187648456057e-05,
+      "loss": 1.1729,
+      "step": 5212
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.50059382422803e-05,
+      "loss": 1.3138,
+      "step": 5213
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.8284,
+      "step": 5214
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.499406175771971e-05,
+      "loss": 0.6935,
+      "step": 5215
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.498812351543943e-05,
+      "loss": 0.5973,
+      "step": 5216
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.498218527315914e-05,
+      "loss": 0.5773,
+      "step": 5217
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.497624703087886e-05,
+      "loss": 0.6813,
+      "step": 5218
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.497030878859858e-05,
+      "loss": 0.5,
+      "step": 5219
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.49643705463183e-05,
+      "loss": 0.6373,
+      "step": 5220
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.495843230403801e-05,
+      "loss": 0.7112,
+      "step": 5221
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.495249406175773e-05,
+      "loss": 0.5061,
+      "step": 5222
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.494655581947744e-05,
+      "loss": 0.6239,
+      "step": 5223
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.494061757719716e-05,
+      "loss": 0.5285,
+      "step": 5224
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.493467933491686e-05,
+      "loss": 0.815,
+      "step": 5225
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.492874109263658e-05,
+      "loss": 0.5903,
+      "step": 5226
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.49228028503563e-05,
+      "loss": 0.9028,
+      "step": 5227
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.491686460807601e-05,
+      "loss": 0.6336,
+      "step": 5228
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.491092636579573e-05,
+      "loss": 0.6512,
+      "step": 5229
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.490498812351544e-05,
+      "loss": 0.4674,
+      "step": 5230
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.489904988123516e-05,
+      "loss": 0.6928,
+      "step": 5231
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.489311163895486e-05,
+      "loss": 0.5858,
+      "step": 5232
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.48871733966746e-05,
+      "loss": 0.9419,
+      "step": 5233
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.48812351543943e-05,
+      "loss": 0.6438,
+      "step": 5234
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.487529691211401e-05,
+      "loss": 0.8143,
+      "step": 5235
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.486935866983373e-05,
+      "loss": 0.6465,
+      "step": 5236
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.486342042755344e-05,
+      "loss": 0.642,
+      "step": 5237
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.485748218527316e-05,
+      "loss": 0.5185,
+      "step": 5238
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.485154394299288e-05,
+      "loss": 0.554,
+      "step": 5239
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.48456057007126e-05,
+      "loss": 0.4833,
+      "step": 5240
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.483966745843231e-05,
+      "loss": 0.6684,
+      "step": 5241
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.483372921615203e-05,
+      "loss": 0.6027,
+      "step": 5242
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.482779097387173e-05,
+      "loss": 0.5221,
+      "step": 5243
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.482185273159146e-05,
+      "loss": 0.5302,
+      "step": 5244
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.481591448931116e-05,
+      "loss": 0.5055,
+      "step": 5245
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.480997624703088e-05,
+      "loss": 0.5298,
+      "step": 5246
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.48040380047506e-05,
+      "loss": 0.6459,
+      "step": 5247
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.479809976247031e-05,
+      "loss": 0.3997,
+      "step": 5248
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.479216152019003e-05,
+      "loss": 0.4382,
+      "step": 5249
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.478622327790974e-05,
+      "loss": 0.4088,
+      "step": 5250
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.478028503562946e-05,
+      "loss": 0.3372,
+      "step": 5251
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.477434679334916e-05,
+      "loss": 0.4906,
+      "step": 5252
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.47684085510689e-05,
+      "loss": 0.4173,
+      "step": 5253
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.47624703087886e-05,
+      "loss": 0.3867,
+      "step": 5254
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.475653206650832e-05,
+      "loss": 0.4273,
+      "step": 5255
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.475059382422803e-05,
+      "loss": 0.5033,
+      "step": 5256
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.474465558194774e-05,
+      "loss": 0.2788,
+      "step": 5257
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.473871733966746e-05,
+      "loss": 0.587,
+      "step": 5258
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.473277909738718e-05,
+      "loss": 0.5092,
+      "step": 5259
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.47268408551069e-05,
+      "loss": 0.6522,
+      "step": 5260
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.47209026128266e-05,
+      "loss": 1.2184,
+      "step": 5261
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.471496437054633e-05,
+      "loss": 0.9728,
+      "step": 5262
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.470902612826603e-05,
+      "loss": 0.9188,
+      "step": 5263
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.470308788598575e-05,
+      "loss": 0.7623,
+      "step": 5264
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.469714964370546e-05,
+      "loss": 0.8592,
+      "step": 5265
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.469121140142518e-05,
+      "loss": 0.6042,
+      "step": 5266
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.46852731591449e-05,
+      "loss": 0.7255,
+      "step": 5267
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.467933491686461e-05,
+      "loss": 0.7643,
+      "step": 5268
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.467339667458433e-05,
+      "loss": 0.5977,
+      "step": 5269
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.466745843230404e-05,
+      "loss": 0.8504,
+      "step": 5270
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.466152019002376e-05,
+      "loss": 0.6496,
+      "step": 5271
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.465558194774348e-05,
+      "loss": 0.5891,
+      "step": 5272
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.46496437054632e-05,
+      "loss": 0.5485,
+      "step": 5273
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.46437054631829e-05,
+      "loss": 0.5879,
+      "step": 5274
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.463776722090262e-05,
+      "loss": 0.4974,
+      "step": 5275
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.463182897862233e-05,
+      "loss": 0.6357,
+      "step": 5276
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.462589073634204e-05,
+      "loss": 0.563,
+      "step": 5277
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.461995249406176e-05,
+      "loss": 0.672,
+      "step": 5278
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.461401425178148e-05,
+      "loss": 0.5586,
+      "step": 5279
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.46080760095012e-05,
+      "loss": 0.6183,
+      "step": 5280
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.46021377672209e-05,
+      "loss": 0.6487,
+      "step": 5281
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.459619952494063e-05,
+      "loss": 0.5419,
+      "step": 5282
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.459026128266033e-05,
+      "loss": 0.5066,
+      "step": 5283
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.458432304038005e-05,
+      "loss": 0.5834,
+      "step": 5284
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.457838479809976e-05,
+      "loss": 0.5291,
+      "step": 5285
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.457244655581948e-05,
+      "loss": 0.458,
+      "step": 5286
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.45665083135392e-05,
+      "loss": 0.6394,
+      "step": 5287
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.456057007125891e-05,
+      "loss": 0.5822,
+      "step": 5288
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.455463182897863e-05,
+      "loss": 0.4132,
+      "step": 5289
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.454869358669833e-05,
+      "loss": 0.585,
+      "step": 5290
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.454275534441805e-05,
+      "loss": 0.5932,
+      "step": 5291
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.453681710213777e-05,
+      "loss": 0.4214,
+      "step": 5292
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.45308788598575e-05,
+      "loss": 0.4856,
+      "step": 5293
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.45249406175772e-05,
+      "loss": 0.4738,
+      "step": 5294
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.451900237529692e-05,
+      "loss": 0.4518,
+      "step": 5295
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.451306413301663e-05,
+      "loss": 0.4508,
+      "step": 5296
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.450712589073634e-05,
+      "loss": 0.4784,
+      "step": 5297
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.450118764845606e-05,
+      "loss": 0.3728,
+      "step": 5298
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.449524940617578e-05,
+      "loss": 0.4636,
+      "step": 5299
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.44893111638955e-05,
+      "loss": 0.5139,
+      "step": 5300
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.44833729216152e-05,
+      "loss": 0.4612,
+      "step": 5301
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.447743467933492e-05,
+      "loss": 0.4578,
+      "step": 5302
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.447149643705463e-05,
+      "loss": 0.5581,
+      "step": 5303
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.446555819477435e-05,
+      "loss": 0.4437,
+      "step": 5304
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.445961995249406e-05,
+      "loss": 0.3985,
+      "step": 5305
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.445368171021378e-05,
+      "loss": 0.3087,
+      "step": 5306
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.44477434679335e-05,
+      "loss": 0.5095,
+      "step": 5307
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.44418052256532e-05,
+      "loss": 0.4686,
+      "step": 5308
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.443586698337293e-05,
+      "loss": 0.4503,
+      "step": 5309
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.442992874109263e-05,
+      "loss": 0.8853,
+      "step": 5310
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.442399049881235e-05,
+      "loss": 1.2828,
+      "step": 5311
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.441805225653207e-05,
+      "loss": 1.0682,
+      "step": 5312
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.44121140142518e-05,
+      "loss": 0.8899,
+      "step": 5313
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.44061757719715e-05,
+      "loss": 0.7063,
+      "step": 5314
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.440023752969122e-05,
+      "loss": 0.551,
+      "step": 5315
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.439429928741093e-05,
+      "loss": 0.5757,
+      "step": 5316
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.438836104513065e-05,
+      "loss": 0.8172,
+      "step": 5317
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.438242280285035e-05,
+      "loss": 0.7568,
+      "step": 5318
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.437648456057008e-05,
+      "loss": 0.916,
+      "step": 5319
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.43705463182898e-05,
+      "loss": 0.8945,
+      "step": 5320
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.43646080760095e-05,
+      "loss": 0.7841,
+      "step": 5321
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.435866983372922e-05,
+      "loss": 0.747,
+      "step": 5322
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.435273159144893e-05,
+      "loss": 0.8363,
+      "step": 5323
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.434679334916865e-05,
+      "loss": 0.6863,
+      "step": 5324
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.434085510688836e-05,
+      "loss": 0.5545,
+      "step": 5325
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.433491686460809e-05,
+      "loss": 0.4924,
+      "step": 5326
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.43289786223278e-05,
+      "loss": 0.4946,
+      "step": 5327
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.432304038004752e-05,
+      "loss": 0.5836,
+      "step": 5328
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.431710213776722e-05,
+      "loss": 0.5996,
+      "step": 5329
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.431116389548693e-05,
+      "loss": 0.5316,
+      "step": 5330
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.430522565320665e-05,
+      "loss": 0.4868,
+      "step": 5331
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.429928741092637e-05,
+      "loss": 0.4934,
+      "step": 5332
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.429334916864609e-05,
+      "loss": 0.4469,
+      "step": 5333
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.42874109263658e-05,
+      "loss": 0.3637,
+      "step": 5334
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.428147268408552e-05,
+      "loss": 0.4922,
+      "step": 5335
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.427553444180523e-05,
+      "loss": 0.5513,
+      "step": 5336
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.426959619952495e-05,
+      "loss": 0.3814,
+      "step": 5337
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.426365795724465e-05,
+      "loss": 0.7932,
+      "step": 5338
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.425771971496437e-05,
+      "loss": 0.5937,
+      "step": 5339
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.42517814726841e-05,
+      "loss": 0.5278,
+      "step": 5340
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.42458432304038e-05,
+      "loss": 0.507,
+      "step": 5341
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.423990498812352e-05,
+      "loss": 0.4492,
+      "step": 5342
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.423396674584323e-05,
+      "loss": 0.4245,
+      "step": 5343
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.422802850356295e-05,
+      "loss": 0.5536,
+      "step": 5344
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.422209026128266e-05,
+      "loss": 0.5872,
+      "step": 5345
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.421615201900239e-05,
+      "loss": 0.6312,
+      "step": 5346
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.42102137767221e-05,
+      "loss": 0.5689,
+      "step": 5347
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.420427553444182e-05,
+      "loss": 0.4778,
+      "step": 5348
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.419833729216152e-05,
+      "loss": 0.5774,
+      "step": 5349
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.419239904988123e-05,
+      "loss": 0.3538,
+      "step": 5350
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.418646080760095e-05,
+      "loss": 0.331,
+      "step": 5351
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.418052256532067e-05,
+      "loss": 0.7481,
+      "step": 5352
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.417458432304039e-05,
+      "loss": 1.352,
+      "step": 5353
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.41686460807601e-05,
+      "loss": 1.1296,
+      "step": 5354
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.416270783847982e-05,
+      "loss": 0.7127,
+      "step": 5355
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.415676959619953e-05,
+      "loss": 0.6301,
+      "step": 5356
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.415083135391925e-05,
+      "loss": 0.7728,
+      "step": 5357
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.414489311163895e-05,
+      "loss": 0.544,
+      "step": 5358
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.413895486935867e-05,
+      "loss": 0.5514,
+      "step": 5359
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.413301662707839e-05,
+      "loss": 0.6222,
+      "step": 5360
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.41270783847981e-05,
+      "loss": 0.6206,
+      "step": 5361
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.412114014251782e-05,
+      "loss": 0.6668,
+      "step": 5362
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.411520190023753e-05,
+      "loss": 0.5249,
+      "step": 5363
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.410926365795725e-05,
+      "loss": 0.6347,
+      "step": 5364
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.410332541567695e-05,
+      "loss": 0.6801,
+      "step": 5365
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.409738717339669e-05,
+      "loss": 0.7258,
+      "step": 5366
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.40914489311164e-05,
+      "loss": 0.4884,
+      "step": 5367
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.408551068883612e-05,
+      "loss": 0.4474,
+      "step": 5368
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.407957244655582e-05,
+      "loss": 0.4271,
+      "step": 5369
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.407363420427553e-05,
+      "loss": 0.3837,
+      "step": 5370
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.406769596199525e-05,
+      "loss": 0.3226,
+      "step": 5371
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.406175771971497e-05,
+      "loss": 0.4093,
+      "step": 5372
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.405581947743469e-05,
+      "loss": 0.3626,
+      "step": 5373
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.40498812351544e-05,
+      "loss": 0.4422,
+      "step": 5374
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 7.404394299287412e-05,
+      "loss": 0.3511,
+      "step": 5375
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 7.403800475059382e-05,
+      "loss": 0.4844,
+      "step": 5376
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 7.403206650831354e-05,
+      "loss": 0.4104,
+      "step": 5377
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 7.402612826603325e-05,
+      "loss": 0.4543,
+      "step": 5378
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 7.402019002375297e-05,
+      "loss": 0.3708,
+      "step": 5379
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 7.401425178147269e-05,
+      "loss": 0.4557,
+      "step": 5380
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 7.40083135391924e-05,
+      "loss": 0.4726,
+      "step": 5381
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 7.400237529691212e-05,
+      "loss": 0.4456,
+      "step": 5382
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 7.399643705463183e-05,
+      "loss": 0.436,
+      "step": 5383
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 7.399049881235155e-05,
+      "loss": 0.3086,
+      "step": 5384
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 7.398456057007125e-05,
+      "loss": 0.2793,
+      "step": 5385
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 7.397862232779099e-05,
+      "loss": 0.3217,
+      "step": 5386
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 7.39726840855107e-05,
+      "loss": 0.3599,
+      "step": 5387
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 7.396674584323041e-05,
+      "loss": 0.3602,
+      "step": 5388
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 7.396080760095012e-05,
+      "loss": 0.3023,
+      "step": 5389
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 7.395486935866984e-05,
+      "loss": 0.3146,
+      "step": 5390
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 7.394893111638955e-05,
+      "loss": 0.2823,
+      "step": 5391
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 7.394299287410927e-05,
+      "loss": 0.3146,
+      "step": 5392
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 7.393705463182899e-05,
+      "loss": 0.4498,
+      "step": 5393
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 7.39311163895487e-05,
+      "loss": 0.3239,
+      "step": 5394
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 7.392517814726842e-05,
+      "loss": 0.4602,
+      "step": 5395
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 7.391923990498812e-05,
+      "loss": 0.265,
+      "step": 5396
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 7.391330166270784e-05,
+      "loss": 0.2766,
+      "step": 5397
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 7.390736342042755e-05,
+      "loss": 0.5598,
+      "step": 5398
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 7.390142517814727e-05,
+      "loss": 0.3609,
+      "step": 5399
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 7.389548693586699e-05,
+      "loss": 0.2347,
+      "step": 5400
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 7.38895486935867e-05,
+      "loss": 0.2753,
+      "step": 5401
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 7.388361045130642e-05,
+      "loss": 0.4408,
+      "step": 5402
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 7.387767220902612e-05,
+      "loss": 0.9957,
+      "step": 5403
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 7.387173396674584e-05,
+      "loss": 0.9383,
+      "step": 5404
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 7.386579572446557e-05,
+      "loss": 0.9723,
+      "step": 5405
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 7.385985748218529e-05,
+      "loss": 0.7578,
+      "step": 5406
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 7.385391923990499e-05,
+      "loss": 0.5723,
+      "step": 5407
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 7.384798099762471e-05,
+      "loss": 0.4503,
+      "step": 5408
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 7.384204275534442e-05,
+      "loss": 0.5445,
+      "step": 5409
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 7.383610451306414e-05,
+      "loss": 0.4358,
+      "step": 5410
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 7.383016627078385e-05,
+      "loss": 0.4193,
+      "step": 5411
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 7.382422802850357e-05,
+      "loss": 0.5085,
+      "step": 5412
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 7.381828978622329e-05,
+      "loss": 0.5928,
+      "step": 5413
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 7.3812351543943e-05,
+      "loss": 0.5913,
+      "step": 5414
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 7.380641330166271e-05,
+      "loss": 0.4845,
+      "step": 5415
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 7.380047505938242e-05,
+      "loss": 0.3432,
+      "step": 5416
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 7.379453681710214e-05,
+      "loss": 0.4776,
+      "step": 5417
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 7.378859857482185e-05,
+      "loss": 0.4939,
+      "step": 5418
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 7.378266033254157e-05,
+      "loss": 0.3684,
+      "step": 5419
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 7.377672209026129e-05,
+      "loss": 0.3817,
+      "step": 5420
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 7.377078384798101e-05,
+      "loss": 0.3315,
+      "step": 5421
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 7.376484560570072e-05,
+      "loss": 0.4778,
+      "step": 5422
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 7.375890736342042e-05,
+      "loss": 0.3894,
+      "step": 5423
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 7.375296912114014e-05,
+      "loss": 0.4477,
+      "step": 5424
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 7.374703087885986e-05,
+      "loss": 0.5233,
+      "step": 5425
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 7.374109263657958e-05,
+      "loss": 0.3133,
+      "step": 5426
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 7.373515439429929e-05,
+      "loss": 0.3692,
+      "step": 5427
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 7.372921615201901e-05,
+      "loss": 0.3118,
+      "step": 5428
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 7.372327790973872e-05,
+      "loss": 0.2957,
+      "step": 5429
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 7.371733966745844e-05,
+      "loss": 0.3397,
+      "step": 5430
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 7.371140142517815e-05,
+      "loss": 0.3736,
+      "step": 5431
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 7.370546318289787e-05,
+      "loss": 0.3808,
+      "step": 5432
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 7.369952494061759e-05,
+      "loss": 0.4309,
+      "step": 5433
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 7.369358669833729e-05,
+      "loss": 0.3599,
+      "step": 5434
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 7.368764845605701e-05,
+      "loss": 0.481,
+      "step": 5435
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 7.368171021377672e-05,
+      "loss": 0.4084,
+      "step": 5436
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 7.367577197149644e-05,
+      "loss": 0.2964,
+      "step": 5437
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 7.366983372921615e-05,
+      "loss": 0.3129,
+      "step": 5438
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 7.366389548693588e-05,
+      "loss": 0.3312,
+      "step": 5439
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 7.365795724465559e-05,
+      "loss": 0.3437,
+      "step": 5440
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 7.365201900237531e-05,
+      "loss": 0.2705,
+      "step": 5441
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 7.364608076009502e-05,
+      "loss": 0.4749,
+      "step": 5442
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 7.364014251781472e-05,
+      "loss": 0.3439,
+      "step": 5443
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 7.363420427553444e-05,
+      "loss": 0.3194,
+      "step": 5444
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 7.362826603325416e-05,
+      "loss": 0.3223,
+      "step": 5445
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 7.362232779097388e-05,
+      "loss": 0.2154,
+      "step": 5446
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 7.361638954869359e-05,
+      "loss": 0.3533,
+      "step": 5447
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 7.361045130641331e-05,
+      "loss": 0.1753,
+      "step": 5448
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 7.360451306413302e-05,
+      "loss": 0.3853,
+      "step": 5449
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 7.359857482185274e-05,
+      "loss": 0.3838,
+      "step": 5450
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 7.359263657957244e-05,
+      "loss": 0.2745,
+      "step": 5451
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 7.358669833729216e-05,
+      "loss": 0.1901,
+      "step": 5452
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 7.358076009501188e-05,
+      "loss": 0.777,
+      "step": 5453
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 7.357482185273159e-05,
+      "loss": 0.6984,
+      "step": 5454
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 7.356888361045131e-05,
+      "loss": 0.9188,
+      "step": 5455
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 7.356294536817102e-05,
+      "loss": 0.5276,
+      "step": 5456
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 7.355700712589074e-05,
+      "loss": 0.7175,
+      "step": 5457
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 7.355106888361045e-05,
+      "loss": 0.3931,
+      "step": 5458
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 7.354513064133018e-05,
+      "loss": 0.4511,
+      "step": 5459
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 7.353919239904989e-05,
+      "loss": 0.4642,
+      "step": 5460
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 7.353325415676961e-05,
+      "loss": 0.584,
+      "step": 5461
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 7.352731591448931e-05,
+      "loss": 0.5961,
+      "step": 5462
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 7.352137767220902e-05,
+      "loss": 0.4337,
+      "step": 5463
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 7.351543942992874e-05,
+      "loss": 0.4767,
+      "step": 5464
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 7.350950118764846e-05,
+      "loss": 0.5477,
+      "step": 5465
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 7.350356294536818e-05,
+      "loss": 0.5474,
+      "step": 5466
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 7.349762470308789e-05,
+      "loss": 0.5426,
+      "step": 5467
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 7.349168646080761e-05,
+      "loss": 0.4707,
+      "step": 5468
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 7.348574821852732e-05,
+      "loss": 0.3305,
+      "step": 5469
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 7.347980997624704e-05,
+      "loss": 0.3592,
+      "step": 5470
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 7.347387173396674e-05,
+      "loss": 0.4099,
+      "step": 5471
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 7.346793349168646e-05,
+      "loss": 0.4496,
+      "step": 5472
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 7.346199524940618e-05,
+      "loss": 0.3331,
+      "step": 5473
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 7.345605700712589e-05,
+      "loss": 0.3449,
+      "step": 5474
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 7.345011876484561e-05,
+      "loss": 0.6378,
+      "step": 5475
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 7.344418052256532e-05,
+      "loss": 0.5324,
+      "step": 5476
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 7.343824228028504e-05,
+      "loss": 0.3389,
+      "step": 5477
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 7.343230403800474e-05,
+      "loss": 0.3909,
+      "step": 5478
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 7.342636579572448e-05,
+      "loss": 0.3837,
+      "step": 5479
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 7.342042755344419e-05,
+      "loss": 0.4025,
+      "step": 5480
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 7.34144893111639e-05,
+      "loss": 0.3391,
+      "step": 5481
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 7.340855106888361e-05,
+      "loss": 0.5292,
+      "step": 5482
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 7.340261282660333e-05,
+      "loss": 0.3288,
+      "step": 5483
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 7.339667458432304e-05,
+      "loss": 0.3114,
+      "step": 5484
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 7.339073634204276e-05,
+      "loss": 0.4046,
+      "step": 5485
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 7.338479809976248e-05,
+      "loss": 0.279,
+      "step": 5486
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 7.337885985748219e-05,
+      "loss": 0.4975,
+      "step": 5487
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 7.337292161520191e-05,
+      "loss": 0.2918,
+      "step": 5488
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 7.336698337292161e-05,
+      "loss": 0.3362,
+      "step": 5489
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 7.336104513064133e-05,
+      "loss": 0.2563,
+      "step": 5490
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 7.335510688836104e-05,
+      "loss": 0.3181,
+      "step": 5491
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 7.334916864608076e-05,
+      "loss": 0.3129,
+      "step": 5492
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 7.334323040380048e-05,
+      "loss": 0.4577,
+      "step": 5493
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 7.333729216152019e-05,
+      "loss": 0.3581,
+      "step": 5494
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 7.333135391923991e-05,
+      "loss": 0.2504,
+      "step": 5495
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 7.332541567695962e-05,
+      "loss": 0.3371,
+      "step": 5496
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 7.331947743467934e-05,
+      "loss": 0.4064,
+      "step": 5497
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 7.331353919239904e-05,
+      "loss": 0.4535,
+      "step": 5498
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 7.330760095011878e-05,
+      "loss": 0.244,
+      "step": 5499
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 7.330166270783848e-05,
+      "loss": 0.3708,
+      "step": 5500
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 7.32957244655582e-05,
+      "loss": 0.1475,
+      "step": 5501
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 7.328978622327791e-05,
+      "loss": 0.3873,
+      "step": 5502
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 7.328384798099763e-05,
+      "loss": 1.1341,
+      "step": 5503
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 7.327790973871734e-05,
+      "loss": 0.9934,
+      "step": 5504
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 7.327197149643706e-05,
+      "loss": 0.8032,
+      "step": 5505
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 7.326603325415678e-05,
+      "loss": 0.8252,
+      "step": 5506
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 7.326009501187649e-05,
+      "loss": 0.6792,
+      "step": 5507
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 7.32541567695962e-05,
+      "loss": 0.5022,
+      "step": 5508
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 7.324821852731591e-05,
+      "loss": 0.4538,
+      "step": 5509
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 7.324228028503563e-05,
+      "loss": 0.4291,
+      "step": 5510
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 7.323634204275534e-05,
+      "loss": 0.4472,
+      "step": 5511
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 7.323040380047506e-05,
+      "loss": 0.4881,
+      "step": 5512
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 7.322446555819478e-05,
+      "loss": 0.4368,
+      "step": 5513
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 7.32185273159145e-05,
+      "loss": 0.5646,
+      "step": 5514
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 7.321258907363421e-05,
+      "loss": 0.375,
+      "step": 5515
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 7.320665083135391e-05,
+      "loss": 0.4761,
+      "step": 5516
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 7.320071258907364e-05,
+      "loss": 0.4889,
+      "step": 5517
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 7.319477434679336e-05,
+      "loss": 0.4038,
+      "step": 5518
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 7.318883610451308e-05,
+      "loss": 0.4418,
+      "step": 5519
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 7.318289786223278e-05,
+      "loss": 0.3509,
+      "step": 5520
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 7.31769596199525e-05,
+      "loss": 0.4496,
+      "step": 5521
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 7.317102137767221e-05,
+      "loss": 0.4032,
+      "step": 5522
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 7.316508313539193e-05,
+      "loss": 0.3976,
+      "step": 5523
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 7.315914489311164e-05,
+      "loss": 0.3326,
+      "step": 5524
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 7.315320665083136e-05,
+      "loss": 0.3611,
+      "step": 5525
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 7.314726840855108e-05,
+      "loss": 0.4912,
+      "step": 5526
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 7.314133016627078e-05,
+      "loss": 0.3693,
+      "step": 5527
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 7.31353919239905e-05,
+      "loss": 0.4439,
+      "step": 5528
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 7.312945368171021e-05,
+      "loss": 0.3261,
+      "step": 5529
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 7.312351543942993e-05,
+      "loss": 0.2883,
+      "step": 5530
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 7.311757719714964e-05,
+      "loss": 0.2912,
+      "step": 5531
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 7.311163895486936e-05,
+      "loss": 0.313,
+      "step": 5532
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 7.310570071258908e-05,
+      "loss": 0.3894,
+      "step": 5533
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 7.30997624703088e-05,
+      "loss": 0.7133,
+      "step": 5534
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 7.309382422802851e-05,
+      "loss": 0.3051,
+      "step": 5535
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 7.308788598574821e-05,
+      "loss": 0.3858,
+      "step": 5536
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 7.308194774346793e-05,
+      "loss": 0.3982,
+      "step": 5537
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 7.307600950118765e-05,
+      "loss": 0.4868,
+      "step": 5538
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 7.307007125890737e-05,
+      "loss": 0.4331,
+      "step": 5539
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 7.306413301662708e-05,
+      "loss": 0.3264,
+      "step": 5540
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 7.30581947743468e-05,
+      "loss": 0.3191,
+      "step": 5541
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 7.305225653206651e-05,
+      "loss": 0.3548,
+      "step": 5542
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 7.304631828978623e-05,
+      "loss": 0.3459,
+      "step": 5543
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 7.304038004750594e-05,
+      "loss": 0.249,
+      "step": 5544
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 7.303444180522566e-05,
+      "loss": 0.3822,
+      "step": 5545
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 7.302850356294538e-05,
+      "loss": 0.2447,
+      "step": 5546
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 7.302256532066508e-05,
+      "loss": 0.2126,
+      "step": 5547
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 7.30166270783848e-05,
+      "loss": 0.2755,
+      "step": 5548
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 7.301068883610451e-05,
+      "loss": 0.5292,
+      "step": 5549
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 7.300475059382423e-05,
+      "loss": 0.2576,
+      "step": 5550
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 7.299881235154394e-05,
+      "loss": 0.3818,
+      "step": 5551
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 7.299287410926366e-05,
+      "loss": 0.2199,
+      "step": 5552
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 7.298693586698338e-05,
+      "loss": 1.167,
+      "step": 5553
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 7.29809976247031e-05,
+      "loss": 0.9126,
+      "step": 5554
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 7.29750593824228e-05,
+      "loss": 0.9918,
+      "step": 5555
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 7.296912114014253e-05,
+      "loss": 0.7085,
+      "step": 5556
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 7.296318289786223e-05,
+      "loss": 0.538,
+      "step": 5557
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 7.295724465558195e-05,
+      "loss": 0.5185,
+      "step": 5558
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 7.295130641330167e-05,
+      "loss": 0.6015,
+      "step": 5559
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 7.294536817102138e-05,
+      "loss": 0.4088,
+      "step": 5560
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 7.29394299287411e-05,
+      "loss": 0.5689,
+      "step": 5561
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 7.293349168646081e-05,
+      "loss": 0.4271,
+      "step": 5562
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 7.292755344418053e-05,
+      "loss": 0.5292,
+      "step": 5563
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 7.292161520190023e-05,
+      "loss": 0.4877,
+      "step": 5564
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 7.291567695961995e-05,
+      "loss": 0.3269,
+      "step": 5565
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 7.290973871733968e-05,
+      "loss": 0.4046,
+      "step": 5566
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 7.290380047505938e-05,
+      "loss": 0.4292,
+      "step": 5567
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 7.28978622327791e-05,
+      "loss": 0.4273,
+      "step": 5568
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 7.289192399049881e-05,
+      "loss": 0.4405,
+      "step": 5569
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 7.288598574821853e-05,
+      "loss": 0.3606,
+      "step": 5570
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 7.288004750593824e-05,
+      "loss": 0.2736,
+      "step": 5571
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 7.287410926365797e-05,
+      "loss": 0.3601,
+      "step": 5572
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 7.286817102137768e-05,
+      "loss": 0.4448,
+      "step": 5573
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 7.28622327790974e-05,
+      "loss": 0.4842,
+      "step": 5574
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 7.28562945368171e-05,
+      "loss": 0.3445,
+      "step": 5575
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 7.285035629453682e-05,
+      "loss": 0.2869,
+      "step": 5576
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 7.284441805225653e-05,
+      "loss": 0.3616,
+      "step": 5577
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 7.283847980997625e-05,
+      "loss": 0.4089,
+      "step": 5578
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 7.283254156769597e-05,
+      "loss": 0.3088,
+      "step": 5579
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 7.282660332541568e-05,
+      "loss": 0.3988,
+      "step": 5580
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 7.28206650831354e-05,
+      "loss": 0.3856,
+      "step": 5581
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 7.28147268408551e-05,
+      "loss": 0.4634,
+      "step": 5582
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 7.280878859857483e-05,
+      "loss": 0.3885,
+      "step": 5583
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 7.280285035629453e-05,
+      "loss": 0.4416,
+      "step": 5584
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 7.279691211401425e-05,
+      "loss": 0.418,
+      "step": 5585
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 7.279097387173397e-05,
+      "loss": 0.4895,
+      "step": 5586
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 7.27850356294537e-05,
+      "loss": 0.3302,
+      "step": 5587
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 7.27790973871734e-05,
+      "loss": 0.3983,
+      "step": 5588
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 7.277315914489311e-05,
+      "loss": 0.3027,
+      "step": 5589
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 7.276722090261283e-05,
+      "loss": 0.3218,
+      "step": 5590
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 7.276128266033254e-05,
+      "loss": 0.2629,
+      "step": 5591
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 7.275534441805227e-05,
+      "loss": 0.3547,
+      "step": 5592
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 7.274940617577198e-05,
+      "loss": 0.3394,
+      "step": 5593
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 7.27434679334917e-05,
+      "loss": 0.5575,
+      "step": 5594
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 7.27375296912114e-05,
+      "loss": 0.3204,
+      "step": 5595
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 7.273159144893112e-05,
+      "loss": 0.4475,
+      "step": 5596
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 7.272565320665083e-05,
+      "loss": 0.404,
+      "step": 5597
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 7.271971496437055e-05,
+      "loss": 0.6196,
+      "step": 5598
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 7.271377672209027e-05,
+      "loss": 0.396,
+      "step": 5599
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 7.270783847980998e-05,
+      "loss": 0.3085,
+      "step": 5600
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 7.27019002375297e-05,
+      "loss": 0.364,
+      "step": 5601
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 7.26959619952494e-05,
+      "loss": 0.4979,
+      "step": 5602
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 7.269002375296913e-05,
+      "loss": 0.7424,
+      "step": 5603
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 7.268408551068883e-05,
+      "loss": 0.7598,
+      "step": 5604
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 7.267814726840855e-05,
+      "loss": 0.7402,
+      "step": 5605
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 7.267220902612827e-05,
+      "loss": 0.6704,
+      "step": 5606
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 7.266627078384799e-05,
+      "loss": 0.5645,
+      "step": 5607
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 7.26603325415677e-05,
+      "loss": 0.521,
+      "step": 5608
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 7.26543942992874e-05,
+      "loss": 0.4463,
+      "step": 5609
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 7.264845605700713e-05,
+      "loss": 0.5365,
+      "step": 5610
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 7.264251781472683e-05,
+      "loss": 0.5737,
+      "step": 5611
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 7.263657957244657e-05,
+      "loss": 0.4233,
+      "step": 5612
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 7.263064133016627e-05,
+      "loss": 0.52,
+      "step": 5613
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 7.2624703087886e-05,
+      "loss": 0.4212,
+      "step": 5614
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 7.26187648456057e-05,
+      "loss": 0.4065,
+      "step": 5615
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 7.261282660332542e-05,
+      "loss": 0.4434,
+      "step": 5616
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 7.260688836104513e-05,
+      "loss": 0.2945,
+      "step": 5617
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 7.260095011876485e-05,
+      "loss": 0.4238,
+      "step": 5618
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 7.259501187648457e-05,
+      "loss": 0.3951,
+      "step": 5619
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 7.258907363420428e-05,
+      "loss": 0.3798,
+      "step": 5620
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 7.2583135391924e-05,
+      "loss": 0.4179,
+      "step": 5621
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 7.25771971496437e-05,
+      "loss": 0.3901,
+      "step": 5622
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 7.257125890736342e-05,
+      "loss": 0.3741,
+      "step": 5623
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 7.256532066508313e-05,
+      "loss": 0.3261,
+      "step": 5624
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 7.255938242280285e-05,
+      "loss": 0.4407,
+      "step": 5625
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 7.255344418052257e-05,
+      "loss": 0.4051,
+      "step": 5626
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 7.254750593824229e-05,
+      "loss": 0.3124,
+      "step": 5627
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 7.2541567695962e-05,
+      "loss": 0.4619,
+      "step": 5628
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 7.25356294536817e-05,
+      "loss": 0.3489,
+      "step": 5629
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 7.252969121140143e-05,
+      "loss": 0.2977,
+      "step": 5630
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 7.252375296912113e-05,
+      "loss": 0.3767,
+      "step": 5631
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 7.251781472684087e-05,
+      "loss": 0.417,
+      "step": 5632
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 7.251187648456057e-05,
+      "loss": 0.3626,
+      "step": 5633
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 7.25059382422803e-05,
+      "loss": 0.54,
+      "step": 5634
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 7.25e-05,
+      "loss": 0.2995,
+      "step": 5635
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 7.249406175771972e-05,
+      "loss": 0.2805,
+      "step": 5636
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 7.248812351543943e-05,
+      "loss": 0.3271,
+      "step": 5637
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 7.248218527315915e-05,
+      "loss": 0.3941,
+      "step": 5638
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 7.247624703087887e-05,
+      "loss": 0.2613,
+      "step": 5639
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 7.247030878859858e-05,
+      "loss": 0.339,
+      "step": 5640
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 7.24643705463183e-05,
+      "loss": 0.3039,
+      "step": 5641
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 7.2458432304038e-05,
+      "loss": 0.3628,
+      "step": 5642
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 7.245249406175772e-05,
+      "loss": 0.2243,
+      "step": 5643
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 7.244655581947743e-05,
+      "loss": 0.2829,
+      "step": 5644
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 7.244061757719715e-05,
+      "loss": 0.2626,
+      "step": 5645
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 7.243467933491687e-05,
+      "loss": 0.5456,
+      "step": 5646
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 7.242874109263659e-05,
+      "loss": 0.1909,
+      "step": 5647
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 7.24228028503563e-05,
+      "loss": 0.4711,
+      "step": 5648
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 7.241686460807602e-05,
+      "loss": 0.3548,
+      "step": 5649
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 7.241092636579572e-05,
+      "loss": 0.4852,
+      "step": 5650
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 7.240498812351544e-05,
+      "loss": 0.4023,
+      "step": 5651
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 7.239904988123517e-05,
+      "loss": 0.1711,
+      "step": 5652
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 7.239311163895487e-05,
+      "loss": 0.8822,
+      "step": 5653
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 7.238717339667459e-05,
+      "loss": 0.7611,
+      "step": 5654
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 7.23812351543943e-05,
+      "loss": 0.7352,
+      "step": 5655
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 7.237529691211402e-05,
+      "loss": 0.6057,
+      "step": 5656
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 7.236935866983373e-05,
+      "loss": 0.5525,
+      "step": 5657
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 7.236342042755345e-05,
+      "loss": 0.3637,
+      "step": 5658
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 7.235748218527317e-05,
+      "loss": 0.4491,
+      "step": 5659
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 7.235154394299287e-05,
+      "loss": 0.5349,
+      "step": 5660
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 7.23456057007126e-05,
+      "loss": 0.6159,
+      "step": 5661
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 7.23396674584323e-05,
+      "loss": 0.5842,
+      "step": 5662
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 7.233372921615202e-05,
+      "loss": 0.5108,
+      "step": 5663
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 7.232779097387173e-05,
+      "loss": 0.4988,
+      "step": 5664
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 7.232185273159145e-05,
+      "loss": 0.5153,
+      "step": 5665
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 7.231591448931117e-05,
+      "loss": 0.473,
+      "step": 5666
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 7.230997624703089e-05,
+      "loss": 0.4248,
+      "step": 5667
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 7.23040380047506e-05,
+      "loss": 0.2802,
+      "step": 5668
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 7.229809976247032e-05,
+      "loss": 0.3989,
+      "step": 5669
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 7.229216152019002e-05,
+      "loss": 0.4309,
+      "step": 5670
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 7.228622327790974e-05,
+      "loss": 0.3004,
+      "step": 5671
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 7.228028503562946e-05,
+      "loss": 0.3457,
+      "step": 5672
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 7.227434679334917e-05,
+      "loss": 0.3496,
+      "step": 5673
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 7.226840855106889e-05,
+      "loss": 0.3663,
+      "step": 5674
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 7.22624703087886e-05,
+      "loss": 0.3611,
+      "step": 5675
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 7.225653206650832e-05,
+      "loss": 0.3457,
+      "step": 5676
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 7.225059382422803e-05,
+      "loss": 0.3476,
+      "step": 5677
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 7.224465558194775e-05,
+      "loss": 0.3765,
+      "step": 5678
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 7.223871733966747e-05,
+      "loss": 0.436,
+      "step": 5679
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 7.223277909738719e-05,
+      "loss": 0.2896,
+      "step": 5680
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 7.222684085510689e-05,
+      "loss": 0.307,
+      "step": 5681
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 7.22209026128266e-05,
+      "loss": 0.267,
+      "step": 5682
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 7.221496437054632e-05,
+      "loss": 0.4001,
+      "step": 5683
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 7.220902612826603e-05,
+      "loss": 0.3892,
+      "step": 5684
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 7.220308788598576e-05,
+      "loss": 0.3975,
+      "step": 5685
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 7.219714964370547e-05,
+      "loss": 0.2796,
+      "step": 5686
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 7.219121140142519e-05,
+      "loss": 0.3516,
+      "step": 5687
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 7.21852731591449e-05,
+      "loss": 0.3457,
+      "step": 5688
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 7.217933491686462e-05,
+      "loss": 0.4087,
+      "step": 5689
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 7.217339667458432e-05,
+      "loss": 0.3485,
+      "step": 5690
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 7.216745843230404e-05,
+      "loss": 0.3181,
+      "step": 5691
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 7.216152019002376e-05,
+      "loss": 0.2647,
+      "step": 5692
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 7.215558194774347e-05,
+      "loss": 0.3989,
+      "step": 5693
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 7.214964370546319e-05,
+      "loss": 0.3531,
+      "step": 5694
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 7.21437054631829e-05,
+      "loss": 0.3353,
+      "step": 5695
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 7.213776722090262e-05,
+      "loss": 0.3632,
+      "step": 5696
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 7.213182897862232e-05,
+      "loss": 0.3079,
+      "step": 5697
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 7.212589073634204e-05,
+      "loss": 0.5524,
+      "step": 5698
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 7.211995249406176e-05,
+      "loss": 0.2062,
+      "step": 5699
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 7.211401425178148e-05,
+      "loss": 0.1958,
+      "step": 5700
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 7.210807600950119e-05,
+      "loss": 0.2934,
+      "step": 5701
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 7.21021377672209e-05,
+      "loss": 0.1692,
+      "step": 5702
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 7.209619952494062e-05,
+      "loss": 1.0363,
+      "step": 5703
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 7.209026128266033e-05,
+      "loss": 0.7201,
+      "step": 5704
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 7.208432304038006e-05,
+      "loss": 0.6985,
+      "step": 5705
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 7.207838479809977e-05,
+      "loss": 0.6004,
+      "step": 5706
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 7.207244655581949e-05,
+      "loss": 0.472,
+      "step": 5707
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 7.20665083135392e-05,
+      "loss": 0.4258,
+      "step": 5708
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 7.206057007125891e-05,
+      "loss": 0.3779,
+      "step": 5709
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 7.205463182897862e-05,
+      "loss": 0.4754,
+      "step": 5710
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 7.204869358669834e-05,
+      "loss": 0.449,
+      "step": 5711
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 7.204275534441806e-05,
+      "loss": 0.4999,
+      "step": 5712
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 7.203681710213777e-05,
+      "loss": 0.5067,
+      "step": 5713
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 7.203087885985749e-05,
+      "loss": 0.5011,
+      "step": 5714
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 7.20249406175772e-05,
+      "loss": 0.4432,
+      "step": 5715
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 7.201900237529692e-05,
+      "loss": 0.4643,
+      "step": 5716
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 7.201306413301662e-05,
+      "loss": 0.3201,
+      "step": 5717
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 7.200712589073634e-05,
+      "loss": 0.794,
+      "step": 5718
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 7.200118764845606e-05,
+      "loss": 0.375,
+      "step": 5719
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 7.199524940617578e-05,
+      "loss": 0.3734,
+      "step": 5720
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 7.198931116389549e-05,
+      "loss": 0.4473,
+      "step": 5721
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 7.198337292161521e-05,
+      "loss": 0.3785,
+      "step": 5722
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 7.197743467933492e-05,
+      "loss": 0.3775,
+      "step": 5723
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 7.197149643705462e-05,
+      "loss": 0.4263,
+      "step": 5724
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 7.196555819477436e-05,
+      "loss": 0.4112,
+      "step": 5725
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 7.195961995249407e-05,
+      "loss": 0.3569,
+      "step": 5726
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 7.195368171021379e-05,
+      "loss": 0.4356,
+      "step": 5727
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 7.194774346793349e-05,
+      "loss": 0.2626,
+      "step": 5728
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 7.194180522565321e-05,
+      "loss": 0.3649,
+      "step": 5729
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 7.193586698337292e-05,
+      "loss": 0.3111,
+      "step": 5730
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 7.192992874109264e-05,
+      "loss": 0.3308,
+      "step": 5731
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 7.192399049881236e-05,
+      "loss": 0.5092,
+      "step": 5732
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 7.191805225653207e-05,
+      "loss": 0.3512,
+      "step": 5733
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 7.191211401425179e-05,
+      "loss": 0.3958,
+      "step": 5734
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 7.19061757719715e-05,
+      "loss": 0.3015,
+      "step": 5735
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 7.190023752969121e-05,
+      "loss": 0.2522,
+      "step": 5736
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 7.189429928741092e-05,
+      "loss": 0.4943,
+      "step": 5737
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 7.188836104513064e-05,
+      "loss": 0.2741,
+      "step": 5738
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 7.188242280285036e-05,
+      "loss": 0.3909,
+      "step": 5739
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 7.187648456057008e-05,
+      "loss": 0.319,
+      "step": 5740
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 7.187054631828979e-05,
+      "loss": 0.2952,
+      "step": 5741
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 7.186460807600951e-05,
+      "loss": 0.2896,
+      "step": 5742
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 7.185866983372922e-05,
+      "loss": 0.2959,
+      "step": 5743
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 7.185273159144892e-05,
+      "loss": 0.2652,
+      "step": 5744
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 7.184679334916866e-05,
+      "loss": 0.1723,
+      "step": 5745
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 7.184085510688836e-05,
+      "loss": 0.4899,
+      "step": 5746
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 7.183491686460808e-05,
+      "loss": 0.2095,
+      "step": 5747
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 7.182897862232779e-05,
+      "loss": 0.2643,
+      "step": 5748
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 7.182304038004751e-05,
+      "loss": 0.2529,
+      "step": 5749
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 7.181710213776722e-05,
+      "loss": 0.4669,
+      "step": 5750
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 7.181116389548694e-05,
+      "loss": 0.3266,
+      "step": 5751
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 7.180522565320666e-05,
+      "loss": 0.5976,
+      "step": 5752
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 7.179928741092638e-05,
+      "loss": 0.7765,
+      "step": 5753
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 7.179334916864609e-05,
+      "loss": 0.8002,
+      "step": 5754
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 7.178741092636579e-05,
+      "loss": 0.6679,
+      "step": 5755
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 7.178147268408551e-05,
+      "loss": 0.6456,
+      "step": 5756
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 7.177553444180522e-05,
+      "loss": 0.5381,
+      "step": 5757
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 7.176959619952494e-05,
+      "loss": 0.4373,
+      "step": 5758
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 7.176365795724466e-05,
+      "loss": 0.5862,
+      "step": 5759
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 7.175771971496438e-05,
+      "loss": 0.4119,
+      "step": 5760
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 7.175178147268409e-05,
+      "loss": 0.5907,
+      "step": 5761
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 7.174584323040381e-05,
+      "loss": 0.555,
+      "step": 5762
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 7.173990498812351e-05,
+      "loss": 0.4682,
+      "step": 5763
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 7.173396674584324e-05,
+      "loss": 0.5628,
+      "step": 5764
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 7.172802850356296e-05,
+      "loss": 0.4216,
+      "step": 5765
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 7.172209026128266e-05,
+      "loss": 0.5001,
+      "step": 5766
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 7.171615201900238e-05,
+      "loss": 0.3992,
+      "step": 5767
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 7.171021377672209e-05,
+      "loss": 0.3168,
+      "step": 5768
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 7.170427553444181e-05,
+      "loss": 1.0102,
+      "step": 5769
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 7.169833729216152e-05,
+      "loss": 0.4635,
+      "step": 5770
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 7.169239904988124e-05,
+      "loss": 0.5096,
+      "step": 5771
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 7.168646080760096e-05,
+      "loss": 0.8593,
+      "step": 5772
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 7.168052256532068e-05,
+      "loss": 0.4366,
+      "step": 5773
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 7.167458432304038e-05,
+      "loss": 0.3934,
+      "step": 5774
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 7.166864608076009e-05,
+      "loss": 0.3532,
+      "step": 5775
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 7.166270783847981e-05,
+      "loss": 0.3033,
+      "step": 5776
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 7.165676959619952e-05,
+      "loss": 0.4623,
+      "step": 5777
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 7.165083135391924e-05,
+      "loss": 0.7442,
+      "step": 5778
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 7.164489311163896e-05,
+      "loss": 0.3718,
+      "step": 5779
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 7.163895486935868e-05,
+      "loss": 0.3574,
+      "step": 5780
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 7.163301662707839e-05,
+      "loss": 0.3455,
+      "step": 5781
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 7.162707838479811e-05,
+      "loss": 0.3786,
+      "step": 5782
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 7.162114014251781e-05,
+      "loss": 0.7266,
+      "step": 5783
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 7.161520190023753e-05,
+      "loss": 0.3533,
+      "step": 5784
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 7.160926365795725e-05,
+      "loss": 0.3896,
+      "step": 5785
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 7.160332541567696e-05,
+      "loss": 0.4131,
+      "step": 5786
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 7.159738717339668e-05,
+      "loss": 0.3941,
+      "step": 5787
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 7.159144893111639e-05,
+      "loss": 0.2723,
+      "step": 5788
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 7.158551068883611e-05,
+      "loss": 0.4004,
+      "step": 5789
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 7.157957244655582e-05,
+      "loss": 0.3324,
+      "step": 5790
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 7.157363420427554e-05,
+      "loss": 0.2935,
+      "step": 5791
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 7.156769596199526e-05,
+      "loss": 0.2817,
+      "step": 5792
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 7.156175771971498e-05,
+      "loss": 0.3505,
+      "step": 5793
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 7.155581947743468e-05,
+      "loss": 0.4826,
+      "step": 5794
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 7.154988123515439e-05,
+      "loss": 0.3033,
+      "step": 5795
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 7.154394299287411e-05,
+      "loss": 0.3711,
+      "step": 5796
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 7.153800475059382e-05,
+      "loss": 0.335,
+      "step": 5797
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 7.153206650831354e-05,
+      "loss": 0.3861,
+      "step": 5798
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 7.152612826603326e-05,
+      "loss": 0.2835,
+      "step": 5799
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 7.152019002375298e-05,
+      "loss": 0.2849,
+      "step": 5800
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 7.151425178147269e-05,
+      "loss": 0.2506,
+      "step": 5801
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 7.15083135391924e-05,
+      "loss": 0.4914,
+      "step": 5802
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 7.150237529691211e-05,
+      "loss": 0.9965,
+      "step": 5803
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 7.149643705463183e-05,
+      "loss": 0.9657,
+      "step": 5804
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 7.149049881235155e-05,
+      "loss": 0.6473,
+      "step": 5805
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 7.148456057007126e-05,
+      "loss": 0.7183,
+      "step": 5806
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 7.147862232779098e-05,
+      "loss": 0.5575,
+      "step": 5807
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 7.147268408551069e-05,
+      "loss": 0.4664,
+      "step": 5808
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 7.146674584323041e-05,
+      "loss": 0.4044,
+      "step": 5809
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 7.146080760095011e-05,
+      "loss": 0.4366,
+      "step": 5810
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 7.145486935866983e-05,
+      "loss": 0.3563,
+      "step": 5811
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 7.144893111638955e-05,
+      "loss": 0.4263,
+      "step": 5812
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 7.144299287410928e-05,
+      "loss": 0.4455,
+      "step": 5813
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 7.143705463182898e-05,
+      "loss": 0.4829,
+      "step": 5814
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 7.14311163895487e-05,
+      "loss": 0.4704,
+      "step": 5815
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 7.142517814726841e-05,
+      "loss": 0.5794,
+      "step": 5816
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 7.141923990498812e-05,
+      "loss": 0.4803,
+      "step": 5817
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 7.141330166270785e-05,
+      "loss": 0.3126,
+      "step": 5818
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 7.140736342042756e-05,
+      "loss": 0.3939,
+      "step": 5819
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 7.140142517814728e-05,
+      "loss": 0.3083,
+      "step": 5820
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 7.139548693586698e-05,
+      "loss": 0.2911,
+      "step": 5821
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 7.13895486935867e-05,
+      "loss": 0.4063,
+      "step": 5822
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 7.138361045130641e-05,
+      "loss": 0.4163,
+      "step": 5823
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 7.137767220902613e-05,
+      "loss": 0.3091,
+      "step": 5824
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 7.137173396674585e-05,
+      "loss": 0.3806,
+      "step": 5825
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 7.136579572446556e-05,
+      "loss": 0.3497,
+      "step": 5826
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 7.135985748218528e-05,
+      "loss": 0.4764,
+      "step": 5827
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 7.135391923990499e-05,
+      "loss": 0.3995,
+      "step": 5828
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 7.13479809976247e-05,
+      "loss": 0.4815,
+      "step": 5829
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 7.134204275534441e-05,
+      "loss": 0.3348,
+      "step": 5830
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 7.133610451306413e-05,
+      "loss": 0.3306,
+      "step": 5831
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 7.133016627078385e-05,
+      "loss": 0.4253,
+      "step": 5832
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 7.132422802850357e-05,
+      "loss": 0.2772,
+      "step": 5833
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 7.131828978622328e-05,
+      "loss": 0.4733,
+      "step": 5834
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 7.1312351543943e-05,
+      "loss": 0.3591,
+      "step": 5835
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 7.130641330166271e-05,
+      "loss": 0.2839,
+      "step": 5836
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 7.130047505938241e-05,
+      "loss": 0.418,
+      "step": 5837
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 7.129453681710215e-05,
+      "loss": 0.3753,
+      "step": 5838
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 7.128859857482186e-05,
+      "loss": 0.156,
+      "step": 5839
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 7.128266033254158e-05,
+      "loss": 0.2199,
+      "step": 5840
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 7.127672209026128e-05,
+      "loss": 0.2529,
+      "step": 5841
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 7.1270783847981e-05,
+      "loss": 0.2251,
+      "step": 5842
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 7.126484560570071e-05,
+      "loss": 0.4026,
+      "step": 5843
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 7.125890736342043e-05,
+      "loss": 0.3131,
+      "step": 5844
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 7.125296912114015e-05,
+      "loss": 0.3588,
+      "step": 5845
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 7.124703087885987e-05,
+      "loss": 0.3051,
+      "step": 5846
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 7.124109263657958e-05,
+      "loss": 0.2718,
+      "step": 5847
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 7.123515439429928e-05,
+      "loss": 0.2885,
+      "step": 5848
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 7.1229216152019e-05,
+      "loss": 0.2524,
+      "step": 5849
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 7.122327790973871e-05,
+      "loss": 0.2694,
+      "step": 5850
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 7.121733966745843e-05,
+      "loss": 0.3699,
+      "step": 5851
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 7.121140142517815e-05,
+      "loss": 0.4152,
+      "step": 5852
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 7.120546318289787e-05,
+      "loss": 0.6825,
+      "step": 5853
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 7.119952494061758e-05,
+      "loss": 0.8863,
+      "step": 5854
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 7.11935866983373e-05,
+      "loss": 0.6397,
+      "step": 5855
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 7.1187648456057e-05,
+      "loss": 0.5263,
+      "step": 5856
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 7.118171021377671e-05,
+      "loss": 0.5007,
+      "step": 5857
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 7.117577197149645e-05,
+      "loss": 0.5584,
+      "step": 5858
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 7.116983372921615e-05,
+      "loss": 0.4551,
+      "step": 5859
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 7.116389548693587e-05,
+      "loss": 0.4468,
+      "step": 5860
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 7.115795724465558e-05,
+      "loss": 0.4041,
+      "step": 5861
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 7.11520190023753e-05,
+      "loss": 0.4559,
+      "step": 5862
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 7.114608076009501e-05,
+      "loss": 0.6331,
+      "step": 5863
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 7.114014251781473e-05,
+      "loss": 0.4097,
+      "step": 5864
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 7.113420427553445e-05,
+      "loss": 0.2715,
+      "step": 5865
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 7.112826603325417e-05,
+      "loss": 0.4237,
+      "step": 5866
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 7.112232779097388e-05,
+      "loss": 0.3586,
+      "step": 5867
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 7.111638954869358e-05,
+      "loss": 0.258,
+      "step": 5868
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 7.11104513064133e-05,
+      "loss": 0.3172,
+      "step": 5869
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 7.110451306413301e-05,
+      "loss": 0.4382,
+      "step": 5870
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 7.109857482185273e-05,
+      "loss": 0.358,
+      "step": 5871
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 7.109263657957245e-05,
+      "loss": 0.3472,
+      "step": 5872
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 7.108669833729217e-05,
+      "loss": 0.356,
+      "step": 5873
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 7.108076009501188e-05,
+      "loss": 0.3176,
+      "step": 5874
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 7.10748218527316e-05,
+      "loss": 0.3503,
+      "step": 5875
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 7.10688836104513e-05,
+      "loss": 0.3587,
+      "step": 5876
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 7.106294536817103e-05,
+      "loss": 0.3581,
+      "step": 5877
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 7.105700712589075e-05,
+      "loss": 0.3419,
+      "step": 5878
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 7.105106888361045e-05,
+      "loss": 0.2668,
+      "step": 5879
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 7.104513064133017e-05,
+      "loss": 0.4675,
+      "step": 5880
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 7.103919239904988e-05,
+      "loss": 0.477,
+      "step": 5881
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 7.10332541567696e-05,
+      "loss": 0.364,
+      "step": 5882
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 7.102731591448931e-05,
+      "loss": 0.3745,
+      "step": 5883
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 7.102137767220903e-05,
+      "loss": 0.3014,
+      "step": 5884
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 7.101543942992875e-05,
+      "loss": 0.3754,
+      "step": 5885
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 7.100950118764847e-05,
+      "loss": 0.3463,
+      "step": 5886
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 7.100356294536818e-05,
+      "loss": 0.317,
+      "step": 5887
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 7.09976247030879e-05,
+      "loss": 0.277,
+      "step": 5888
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 7.09916864608076e-05,
+      "loss": 0.345,
+      "step": 5889
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 7.098574821852731e-05,
+      "loss": 0.2549,
+      "step": 5890
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 7.097980997624703e-05,
+      "loss": 0.28,
+      "step": 5891
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 7.097387173396675e-05,
+      "loss": 0.3122,
+      "step": 5892
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 7.096793349168647e-05,
+      "loss": 0.2841,
+      "step": 5893
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 7.096199524940618e-05,
+      "loss": 0.3127,
+      "step": 5894
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 7.09560570071259e-05,
+      "loss": 0.3737,
+      "step": 5895
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 7.09501187648456e-05,
+      "loss": 0.283,
+      "step": 5896
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 7.094418052256532e-05,
+      "loss": 0.272,
+      "step": 5897
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 7.093824228028504e-05,
+      "loss": 0.2481,
+      "step": 5898
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 7.093230403800475e-05,
+      "loss": 0.5247,
+      "step": 5899
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 7.092636579572447e-05,
+      "loss": 0.3727,
+      "step": 5900
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 7.092042755344418e-05,
+      "loss": 0.2558,
+      "step": 5901
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 7.09144893111639e-05,
+      "loss": 0.4242,
+      "step": 5902
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 7.09085510688836e-05,
+      "loss": 0.9319,
+      "step": 5903
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 7.090261282660333e-05,
+      "loss": 0.5873,
+      "step": 5904
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 7.089667458432305e-05,
+      "loss": 0.6062,
+      "step": 5905
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 7.089073634204277e-05,
+      "loss": 0.8719,
+      "step": 5906
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 7.088479809976247e-05,
+      "loss": 0.3536,
+      "step": 5907
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 7.08788598574822e-05,
+      "loss": 0.5035,
+      "step": 5908
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 7.08729216152019e-05,
+      "loss": 0.5384,
+      "step": 5909
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 7.086698337292161e-05,
+      "loss": 0.4313,
+      "step": 5910
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 7.086104513064133e-05,
+      "loss": 0.5254,
+      "step": 5911
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 7.085510688836105e-05,
+      "loss": 0.5036,
+      "step": 5912
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 7.084916864608077e-05,
+      "loss": 0.4534,
+      "step": 5913
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 7.084323040380048e-05,
+      "loss": 0.4298,
+      "step": 5914
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 7.08372921615202e-05,
+      "loss": 0.5399,
+      "step": 5915
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 7.08313539192399e-05,
+      "loss": 0.4453,
+      "step": 5916
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 7.082541567695962e-05,
+      "loss": 0.406,
+      "step": 5917
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 7.081947743467934e-05,
+      "loss": 0.281,
+      "step": 5918
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 7.081353919239906e-05,
+      "loss": 0.3855,
+      "step": 5919
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 7.080760095011877e-05,
+      "loss": 0.3639,
+      "step": 5920
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 7.080166270783848e-05,
+      "loss": 0.3461,
+      "step": 5921
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 7.07957244655582e-05,
+      "loss": 0.3433,
+      "step": 5922
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 7.07897862232779e-05,
+      "loss": 0.4398,
+      "step": 5923
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 7.078384798099763e-05,
+      "loss": 0.4976,
+      "step": 5924
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 7.077790973871735e-05,
+      "loss": 0.5547,
+      "step": 5925
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 7.077197149643707e-05,
+      "loss": 0.3025,
+      "step": 5926
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 7.076603325415677e-05,
+      "loss": 0.3611,
+      "step": 5927
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 7.076009501187649e-05,
+      "loss": 0.3124,
+      "step": 5928
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 7.07541567695962e-05,
+      "loss": 0.3182,
+      "step": 5929
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 7.07482185273159e-05,
+      "loss": 0.4121,
+      "step": 5930
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 7.074228028503563e-05,
+      "loss": 0.2931,
+      "step": 5931
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 7.073634204275535e-05,
+      "loss": 0.4458,
+      "step": 5932
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 7.073040380047507e-05,
+      "loss": 0.2925,
+      "step": 5933
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 7.072446555819477e-05,
+      "loss": 0.3337,
+      "step": 5934
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 7.07185273159145e-05,
+      "loss": 0.2687,
+      "step": 5935
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 7.07125890736342e-05,
+      "loss": 0.3223,
+      "step": 5936
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 7.070665083135392e-05,
+      "loss": 0.3354,
+      "step": 5937
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 7.070071258907364e-05,
+      "loss": 0.3088,
+      "step": 5938
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 7.069477434679336e-05,
+      "loss": 0.2882,
+      "step": 5939
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 7.068883610451307e-05,
+      "loss": 0.2575,
+      "step": 5940
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 7.068289786223278e-05,
+      "loss": 0.3057,
+      "step": 5941
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 7.06769596199525e-05,
+      "loss": 0.3524,
+      "step": 5942
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 7.06710213776722e-05,
+      "loss": 0.2307,
+      "step": 5943
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 7.066508313539192e-05,
+      "loss": 0.1779,
+      "step": 5944
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 7.065914489311164e-05,
+      "loss": 1.1179,
+      "step": 5945
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 7.065320665083136e-05,
+      "loss": 0.5325,
+      "step": 5946
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 7.064726840855107e-05,
+      "loss": 0.3448,
+      "step": 5947
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 7.064133016627079e-05,
+      "loss": 0.2695,
+      "step": 5948
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 7.06353919239905e-05,
+      "loss": 0.5446,
+      "step": 5949
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 7.062945368171022e-05,
+      "loss": 0.3259,
+      "step": 5950
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 7.062351543942994e-05,
+      "loss": 0.3361,
+      "step": 5951
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 7.061757719714965e-05,
+      "loss": 0.2886,
+      "step": 5952
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 7.061163895486937e-05,
+      "loss": 0.7092,
+      "step": 5953
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 7.060570071258907e-05,
+      "loss": 0.5577,
+      "step": 5954
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 7.05997624703088e-05,
+      "loss": 0.6648,
+      "step": 5955
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 7.05938242280285e-05,
+      "loss": 0.4877,
+      "step": 5956
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 7.058788598574822e-05,
+      "loss": 0.5085,
+      "step": 5957
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 7.058194774346794e-05,
+      "loss": 0.3374,
+      "step": 5958
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 7.057600950118766e-05,
+      "loss": 0.4216,
+      "step": 5959
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 7.057007125890737e-05,
+      "loss": 0.4158,
+      "step": 5960
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 7.056413301662707e-05,
+      "loss": 0.4534,
+      "step": 5961
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 7.05581947743468e-05,
+      "loss": 0.3532,
+      "step": 5962
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 7.05522565320665e-05,
+      "loss": 0.4903,
+      "step": 5963
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 7.054631828978622e-05,
+      "loss": 0.4031,
+      "step": 5964
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 7.054038004750594e-05,
+      "loss": 0.4743,
+      "step": 5965
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 7.053444180522566e-05,
+      "loss": 0.3587,
+      "step": 5966
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 7.052850356294537e-05,
+      "loss": 0.3351,
+      "step": 5967
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 7.052256532066509e-05,
+      "loss": 0.326,
+      "step": 5968
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 7.05166270783848e-05,
+      "loss": 0.3868,
+      "step": 5969
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 7.051068883610452e-05,
+      "loss": 0.3735,
+      "step": 5970
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 7.050475059382424e-05,
+      "loss": 0.4245,
+      "step": 5971
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 7.049881235154394e-05,
+      "loss": 0.4707,
+      "step": 5972
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 7.049287410926367e-05,
+      "loss": 0.3615,
+      "step": 5973
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 7.048693586698337e-05,
+      "loss": 0.3392,
+      "step": 5974
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 7.048099762470309e-05,
+      "loss": 0.4922,
+      "step": 5975
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 7.04750593824228e-05,
+      "loss": 0.5554,
+      "step": 5976
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 7.046912114014252e-05,
+      "loss": 0.3574,
+      "step": 5977
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 7.046318289786224e-05,
+      "loss": 0.3023,
+      "step": 5978
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 7.045724465558196e-05,
+      "loss": 0.3771,
+      "step": 5979
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 7.045130641330167e-05,
+      "loss": 0.3653,
+      "step": 5980
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 7.044536817102139e-05,
+      "loss": 0.5175,
+      "step": 5981
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 7.04394299287411e-05,
+      "loss": 0.3439,
+      "step": 5982
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 7.04334916864608e-05,
+      "loss": 0.3181,
+      "step": 5983
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 7.042755344418052e-05,
+      "loss": 0.5497,
+      "step": 5984
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 7.042161520190024e-05,
+      "loss": 0.4054,
+      "step": 5985
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 7.041567695961996e-05,
+      "loss": 0.2892,
+      "step": 5986
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 7.040973871733967e-05,
+      "loss": 0.3245,
+      "step": 5987
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 7.040380047505939e-05,
+      "loss": 0.4808,
+      "step": 5988
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 7.03978622327791e-05,
+      "loss": 0.2765,
+      "step": 5989
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 7.039192399049882e-05,
+      "loss": 0.4159,
+      "step": 5990
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 7.038598574821854e-05,
+      "loss": 0.2938,
+      "step": 5991
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 7.038004750593824e-05,
+      "loss": 0.3757,
+      "step": 5992
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 7.037410926365796e-05,
+      "loss": 0.3944,
+      "step": 5993
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 7.036817102137767e-05,
+      "loss": 0.3463,
+      "step": 5994
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 7.036223277909739e-05,
+      "loss": 0.2192,
+      "step": 5995
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 7.03562945368171e-05,
+      "loss": 0.2151,
+      "step": 5996
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 7.035035629453682e-05,
+      "loss": 0.2349,
+      "step": 5997
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 7.034441805225654e-05,
+      "loss": 0.3571,
+      "step": 5998
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 7.033847980997626e-05,
+      "loss": 0.2382,
+      "step": 5999
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 7.033254156769597e-05,
+      "loss": 0.1112,
+      "step": 6000
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 7.032660332541569e-05,
+      "loss": 0.2808,
+      "step": 6001
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 7.032066508313539e-05,
+      "loss": 0.3133,
+      "step": 6002
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 7.03147268408551e-05,
+      "loss": 1.2769,
+      "step": 6003
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 7.030878859857482e-05,
+      "loss": 0.6758,
+      "step": 6004
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 7.030285035629454e-05,
+      "loss": 0.8877,
+      "step": 6005
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 7.029691211401426e-05,
+      "loss": 0.5895,
+      "step": 6006
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 7.029097387173397e-05,
+      "loss": 0.4243,
+      "step": 6007
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 7.028503562945369e-05,
+      "loss": 0.3686,
+      "step": 6008
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 7.02790973871734e-05,
+      "loss": 0.3891,
+      "step": 6009
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 7.027315914489311e-05,
+      "loss": 0.4258,
+      "step": 6010
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 7.026722090261284e-05,
+      "loss": 0.3934,
+      "step": 6011
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 7.026128266033256e-05,
+      "loss": 0.4471,
+      "step": 6012
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 7.025534441805226e-05,
+      "loss": 0.4437,
+      "step": 6013
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 7.024940617577197e-05,
+      "loss": 0.4461,
+      "step": 6014
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 7.024346793349169e-05,
+      "loss": 0.3832,
+      "step": 6015
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 7.02375296912114e-05,
+      "loss": 0.5559,
+      "step": 6016
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 7.023159144893112e-05,
+      "loss": 0.3515,
+      "step": 6017
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 7.022565320665084e-05,
+      "loss": 0.3056,
+      "step": 6018
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 7.021971496437056e-05,
+      "loss": 0.4662,
+      "step": 6019
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 7.021377672209026e-05,
+      "loss": 0.2978,
+      "step": 6020
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 7.020783847980998e-05,
+      "loss": 0.3844,
+      "step": 6021
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 7.020190023752969e-05,
+      "loss": 0.3394,
+      "step": 6022
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 7.01959619952494e-05,
+      "loss": 0.3666,
+      "step": 6023
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 7.019002375296912e-05,
+      "loss": 0.4081,
+      "step": 6024
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 7.018408551068884e-05,
+      "loss": 0.4145,
+      "step": 6025
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 7.017814726840856e-05,
+      "loss": 0.3457,
+      "step": 6026
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 7.017220902612827e-05,
+      "loss": 0.4601,
+      "step": 6027
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 7.016627078384799e-05,
+      "loss": 0.3755,
+      "step": 6028
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 7.01603325415677e-05,
+      "loss": 0.4066,
+      "step": 6029
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 7.015439429928741e-05,
+      "loss": 0.2559,
+      "step": 6030
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 7.014845605700713e-05,
+      "loss": 0.4865,
+      "step": 6031
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 7.014251781472685e-05,
+      "loss": 0.427,
+      "step": 6032
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 7.013657957244656e-05,
+      "loss": 0.3255,
+      "step": 6033
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 7.013064133016627e-05,
+      "loss": 0.3136,
+      "step": 6034
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 7.012470308788599e-05,
+      "loss": 0.3413,
+      "step": 6035
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 7.01187648456057e-05,
+      "loss": 0.2616,
+      "step": 6036
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 7.011282660332542e-05,
+      "loss": 0.5696,
+      "step": 6037
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 7.010688836104514e-05,
+      "loss": 0.4088,
+      "step": 6038
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 7.010095011876486e-05,
+      "loss": 0.312,
+      "step": 6039
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 7.009501187648456e-05,
+      "loss": 0.3201,
+      "step": 6040
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 7.008907363420428e-05,
+      "loss": 0.4074,
+      "step": 6041
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 7.008313539192399e-05,
+      "loss": 0.3047,
+      "step": 6042
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 7.007719714964371e-05,
+      "loss": 0.3024,
+      "step": 6043
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 7.007125890736342e-05,
+      "loss": 0.4024,
+      "step": 6044
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 7.006532066508314e-05,
+      "loss": 0.2517,
+      "step": 6045
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 7.005938242280286e-05,
+      "loss": 0.2923,
+      "step": 6046
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 7.005344418052256e-05,
+      "loss": 0.152,
+      "step": 6047
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 7.004750593824229e-05,
+      "loss": 0.365,
+      "step": 6048
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 7.004156769596199e-05,
+      "loss": 0.4,
+      "step": 6049
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 7.003562945368171e-05,
+      "loss": 0.5604,
+      "step": 6050
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 7.002969121140143e-05,
+      "loss": 0.2514,
+      "step": 6051
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 7.002375296912115e-05,
+      "loss": 0.549,
+      "step": 6052
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 7.001781472684086e-05,
+      "loss": 0.924,
+      "step": 6053
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 7.001187648456057e-05,
+      "loss": 0.8341,
+      "step": 6054
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 7.000593824228029e-05,
+      "loss": 0.6688,
+      "step": 6055
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 7e-05,
+      "loss": 0.6229,
+      "step": 6056
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 6.999406175771971e-05,
+      "loss": 0.4412,
+      "step": 6057
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 6.998812351543943e-05,
+      "loss": 0.4381,
+      "step": 6058
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 6.998218527315915e-05,
+      "loss": 0.4243,
+      "step": 6059
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 6.997624703087886e-05,
+      "loss": 0.5244,
+      "step": 6060
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 6.997030878859858e-05,
+      "loss": 0.5253,
+      "step": 6061
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.996437054631829e-05,
+      "loss": 0.4751,
+      "step": 6062
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.995843230403801e-05,
+      "loss": 0.4228,
+      "step": 6063
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.995249406175773e-05,
+      "loss": 0.3885,
+      "step": 6064
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.994655581947744e-05,
+      "loss": 0.4831,
+      "step": 6065
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.994061757719716e-05,
+      "loss": 0.3378,
+      "step": 6066
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.993467933491686e-05,
+      "loss": 0.3764,
+      "step": 6067
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.992874109263658e-05,
+      "loss": 0.3628,
+      "step": 6068
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.992280285035629e-05,
+      "loss": 0.447,
+      "step": 6069
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.991686460807601e-05,
+      "loss": 0.3471,
+      "step": 6070
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.991092636579573e-05,
+      "loss": 0.3592,
+      "step": 6071
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.990498812351545e-05,
+      "loss": 0.3595,
+      "step": 6072
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.989904988123516e-05,
+      "loss": 0.4471,
+      "step": 6073
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.989311163895488e-05,
+      "loss": 0.3409,
+      "step": 6074
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.988717339667459e-05,
+      "loss": 0.3095,
+      "step": 6075
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.988123515439429e-05,
+      "loss": 0.5367,
+      "step": 6076
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.987529691211401e-05,
+      "loss": 0.3566,
+      "step": 6077
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.986935866983373e-05,
+      "loss": 0.3014,
+      "step": 6078
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.986342042755345e-05,
+      "loss": 0.4236,
+      "step": 6079
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.985748218527316e-05,
+      "loss": 0.3104,
+      "step": 6080
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.985154394299288e-05,
+      "loss": 0.3887,
+      "step": 6081
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.984560570071259e-05,
+      "loss": 0.2729,
+      "step": 6082
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.983966745843231e-05,
+      "loss": 0.2785,
+      "step": 6083
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.983372921615203e-05,
+      "loss": 0.3001,
+      "step": 6084
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.982779097387175e-05,
+      "loss": 0.3784,
+      "step": 6085
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.982185273159146e-05,
+      "loss": 0.4456,
+      "step": 6086
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.981591448931116e-05,
+      "loss": 0.2806,
+      "step": 6087
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.980997624703088e-05,
+      "loss": 0.2973,
+      "step": 6088
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.980403800475059e-05,
+      "loss": 0.2657,
+      "step": 6089
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.979809976247031e-05,
+      "loss": 0.346,
+      "step": 6090
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.979216152019003e-05,
+      "loss": 0.2713,
+      "step": 6091
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.978622327790975e-05,
+      "loss": 0.3238,
+      "step": 6092
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.978028503562946e-05,
+      "loss": 0.2843,
+      "step": 6093
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.977434679334918e-05,
+      "loss": 0.2467,
+      "step": 6094
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.976840855106888e-05,
+      "loss": 0.2426,
+      "step": 6095
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.976247030878859e-05,
+      "loss": 0.3364,
+      "step": 6096
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.975653206650831e-05,
+      "loss": 0.2572,
+      "step": 6097
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.975059382422803e-05,
+      "loss": 0.1751,
+      "step": 6098
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.974465558194775e-05,
+      "loss": 0.3096,
+      "step": 6099
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.973871733966746e-05,
+      "loss": 0.338,
+      "step": 6100
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.973277909738718e-05,
+      "loss": 0.24,
+      "step": 6101
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.972684085510689e-05,
+      "loss": 0.2414,
+      "step": 6102
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.97209026128266e-05,
+      "loss": 0.7219,
+      "step": 6103
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.971496437054633e-05,
+      "loss": 0.7599,
+      "step": 6104
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.970902612826605e-05,
+      "loss": 0.5894,
+      "step": 6105
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.970308788598575e-05,
+      "loss": 0.5306,
+      "step": 6106
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.969714964370546e-05,
+      "loss": 0.3295,
+      "step": 6107
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.969121140142518e-05,
+      "loss": 0.6137,
+      "step": 6108
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.968527315914489e-05,
+      "loss": 0.3895,
+      "step": 6109
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.967933491686461e-05,
+      "loss": 0.5618,
+      "step": 6110
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.967339667458433e-05,
+      "loss": 0.392,
+      "step": 6111
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.966745843230405e-05,
+      "loss": 0.5108,
+      "step": 6112
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.966152019002376e-05,
+      "loss": 0.4332,
+      "step": 6113
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.965558194774348e-05,
+      "loss": 0.3702,
+      "step": 6114
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.964964370546318e-05,
+      "loss": 0.4946,
+      "step": 6115
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.96437054631829e-05,
+      "loss": 0.3544,
+      "step": 6116
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.963776722090261e-05,
+      "loss": 0.3389,
+      "step": 6117
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.963182897862233e-05,
+      "loss": 0.4243,
+      "step": 6118
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.962589073634205e-05,
+      "loss": 0.3628,
+      "step": 6119
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.961995249406176e-05,
+      "loss": 0.3263,
+      "step": 6120
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.961401425178148e-05,
+      "loss": 0.3842,
+      "step": 6121
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.960807600950119e-05,
+      "loss": 0.2687,
+      "step": 6122
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.96021377672209e-05,
+      "loss": 0.3173,
+      "step": 6123
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.959619952494063e-05,
+      "loss": 0.6208,
+      "step": 6124
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.959026128266035e-05,
+      "loss": 0.3139,
+      "step": 6125
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.958432304038005e-05,
+      "loss": 0.2535,
+      "step": 6126
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.957838479809976e-05,
+      "loss": 0.2686,
+      "step": 6127
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.957244655581948e-05,
+      "loss": 0.3461,
+      "step": 6128
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.956650831353919e-05,
+      "loss": 0.34,
+      "step": 6129
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.956057007125891e-05,
+      "loss": 0.3799,
+      "step": 6130
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.955463182897863e-05,
+      "loss": 0.4147,
+      "step": 6131
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.954869358669835e-05,
+      "loss": 0.3377,
+      "step": 6132
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.954275534441805e-05,
+      "loss": 0.395,
+      "step": 6133
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.953681710213778e-05,
+      "loss": 0.34,
+      "step": 6134
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.953087885985748e-05,
+      "loss": 0.2441,
+      "step": 6135
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.95249406175772e-05,
+      "loss": 0.3099,
+      "step": 6136
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.951900237529691e-05,
+      "loss": 0.5325,
+      "step": 6137
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.951306413301663e-05,
+      "loss": 0.2168,
+      "step": 6138
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.950712589073635e-05,
+      "loss": 0.2551,
+      "step": 6139
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.950118764845606e-05,
+      "loss": 0.2795,
+      "step": 6140
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.949524940617578e-05,
+      "loss": 0.3173,
+      "step": 6141
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.948931116389548e-05,
+      "loss": 0.3046,
+      "step": 6142
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.94833729216152e-05,
+      "loss": 0.3622,
+      "step": 6143
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.947743467933492e-05,
+      "loss": 0.4061,
+      "step": 6144
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.947149643705464e-05,
+      "loss": 0.1762,
+      "step": 6145
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.946555819477435e-05,
+      "loss": 0.4674,
+      "step": 6146
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.945961995249407e-05,
+      "loss": 0.3511,
+      "step": 6147
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.945368171021378e-05,
+      "loss": 0.2315,
+      "step": 6148
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.944774346793349e-05,
+      "loss": 0.2278,
+      "step": 6149
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.94418052256532e-05,
+      "loss": 0.289,
+      "step": 6150
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.943586698337293e-05,
+      "loss": 0.3544,
+      "step": 6151
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.942992874109265e-05,
+      "loss": 0.4045,
+      "step": 6152
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.942399049881235e-05,
+      "loss": 0.8057,
+      "step": 6153
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.941805225653207e-05,
+      "loss": 0.9955,
+      "step": 6154
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.941211401425178e-05,
+      "loss": 0.7327,
+      "step": 6155
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.94061757719715e-05,
+      "loss": 0.6137,
+      "step": 6156
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.940023752969121e-05,
+      "loss": 0.4844,
+      "step": 6157
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.939429928741093e-05,
+      "loss": 0.4089,
+      "step": 6158
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.938836104513065e-05,
+      "loss": 0.4562,
+      "step": 6159
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.938242280285036e-05,
+      "loss": 0.4589,
+      "step": 6160
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.937648456057008e-05,
+      "loss": 0.4857,
+      "step": 6161
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.937054631828978e-05,
+      "loss": 0.5379,
+      "step": 6162
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.93646080760095e-05,
+      "loss": 0.4005,
+      "step": 6163
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.935866983372922e-05,
+      "loss": 0.4212,
+      "step": 6164
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.935273159144894e-05,
+      "loss": 0.4594,
+      "step": 6165
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.934679334916865e-05,
+      "loss": 0.3182,
+      "step": 6166
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.934085510688837e-05,
+      "loss": 0.413,
+      "step": 6167
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.933491686460808e-05,
+      "loss": 0.3962,
+      "step": 6168
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.932897862232778e-05,
+      "loss": 0.3346,
+      "step": 6169
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.93230403800475e-05,
+      "loss": 0.3093,
+      "step": 6170
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.931710213776723e-05,
+      "loss": 0.3386,
+      "step": 6171
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.931116389548695e-05,
+      "loss": 0.564,
+      "step": 6172
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.930522565320665e-05,
+      "loss": 0.337,
+      "step": 6173
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.929928741092637e-05,
+      "loss": 0.3804,
+      "step": 6174
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.929334916864608e-05,
+      "loss": 0.3432,
+      "step": 6175
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.92874109263658e-05,
+      "loss": 0.348,
+      "step": 6176
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.92814726840855e-05,
+      "loss": 0.484,
+      "step": 6177
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.927553444180524e-05,
+      "loss": 0.3626,
+      "step": 6178
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.926959619952495e-05,
+      "loss": 0.2548,
+      "step": 6179
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.926365795724465e-05,
+      "loss": 0.2985,
+      "step": 6180
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.925771971496437e-05,
+      "loss": 0.3131,
+      "step": 6181
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.925178147268408e-05,
+      "loss": 0.2925,
+      "step": 6182
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.92458432304038e-05,
+      "loss": 0.369,
+      "step": 6183
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.923990498812352e-05,
+      "loss": 0.3412,
+      "step": 6184
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.923396674584324e-05,
+      "loss": 0.2427,
+      "step": 6185
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.922802850356295e-05,
+      "loss": 0.3292,
+      "step": 6186
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.922209026128267e-05,
+      "loss": 0.3133,
+      "step": 6187
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.921615201900238e-05,
+      "loss": 0.3284,
+      "step": 6188
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.921021377672208e-05,
+      "loss": 0.2177,
+      "step": 6189
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.92042755344418e-05,
+      "loss": 0.2566,
+      "step": 6190
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.919833729216152e-05,
+      "loss": 0.3161,
+      "step": 6191
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.919239904988124e-05,
+      "loss": 0.3216,
+      "step": 6192
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.918646080760095e-05,
+      "loss": 0.2701,
+      "step": 6193
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.918052256532067e-05,
+      "loss": 0.414,
+      "step": 6194
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.917458432304038e-05,
+      "loss": 0.3729,
+      "step": 6195
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.91686460807601e-05,
+      "loss": 0.2647,
+      "step": 6196
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.916270783847982e-05,
+      "loss": 0.3851,
+      "step": 6197
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.915676959619954e-05,
+      "loss": 0.2326,
+      "step": 6198
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.915083135391925e-05,
+      "loss": 0.2198,
+      "step": 6199
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.914489311163895e-05,
+      "loss": 0.1989,
+      "step": 6200
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.913895486935867e-05,
+      "loss": 0.3601,
+      "step": 6201
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.913301662707838e-05,
+      "loss": 0.5107,
+      "step": 6202
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.91270783847981e-05,
+      "loss": 0.8413,
+      "step": 6203
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.912114014251782e-05,
+      "loss": 0.6691,
+      "step": 6204
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.911520190023754e-05,
+      "loss": 0.6519,
+      "step": 6205
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.910926365795725e-05,
+      "loss": 0.4261,
+      "step": 6206
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.910332541567697e-05,
+      "loss": 0.5689,
+      "step": 6207
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.909738717339667e-05,
+      "loss": 0.429,
+      "step": 6208
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.90914489311164e-05,
+      "loss": 0.3632,
+      "step": 6209
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.90855106888361e-05,
+      "loss": 0.6292,
+      "step": 6210
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.907957244655582e-05,
+      "loss": 0.3786,
+      "step": 6211
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.907363420427554e-05,
+      "loss": 0.4199,
+      "step": 6212
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.906769596199525e-05,
+      "loss": 0.4092,
+      "step": 6213
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.906175771971497e-05,
+      "loss": 0.3959,
+      "step": 6214
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.905581947743468e-05,
+      "loss": 0.3728,
+      "step": 6215
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.90498812351544e-05,
+      "loss": 0.3349,
+      "step": 6216
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.904394299287412e-05,
+      "loss": 0.4573,
+      "step": 6217
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.903800475059384e-05,
+      "loss": 0.4133,
+      "step": 6218
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.903206650831354e-05,
+      "loss": 0.2953,
+      "step": 6219
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.902612826603325e-05,
+      "loss": 0.2556,
+      "step": 6220
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.902019002375297e-05,
+      "loss": 0.2811,
+      "step": 6221
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.901425178147268e-05,
+      "loss": 0.4792,
+      "step": 6222
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.90083135391924e-05,
+      "loss": 0.2835,
+      "step": 6223
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.900237529691212e-05,
+      "loss": 0.3816,
+      "step": 6224
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.899643705463184e-05,
+      "loss": 0.3377,
+      "step": 6225
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.899049881235155e-05,
+      "loss": 0.4503,
+      "step": 6226
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.898456057007127e-05,
+      "loss": 0.3616,
+      "step": 6227
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.897862232779097e-05,
+      "loss": 0.2538,
+      "step": 6228
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.89726840855107e-05,
+      "loss": 0.3735,
+      "step": 6229
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.89667458432304e-05,
+      "loss": 0.3777,
+      "step": 6230
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.896080760095012e-05,
+      "loss": 0.2729,
+      "step": 6231
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.895486935866984e-05,
+      "loss": 0.3072,
+      "step": 6232
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.894893111638955e-05,
+      "loss": 0.4378,
+      "step": 6233
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.894299287410927e-05,
+      "loss": 0.3208,
+      "step": 6234
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.893705463182898e-05,
+      "loss": 0.2558,
+      "step": 6235
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.89311163895487e-05,
+      "loss": 0.3976,
+      "step": 6236
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.892517814726842e-05,
+      "loss": 0.2296,
+      "step": 6237
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.891923990498814e-05,
+      "loss": 0.2943,
+      "step": 6238
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.891330166270784e-05,
+      "loss": 0.2275,
+      "step": 6239
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.890736342042756e-05,
+      "loss": 0.3868,
+      "step": 6240
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.890142517814727e-05,
+      "loss": 0.2298,
+      "step": 6241
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.889548693586698e-05,
+      "loss": 0.235,
+      "step": 6242
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.88895486935867e-05,
+      "loss": 0.4357,
+      "step": 6243
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.888361045130642e-05,
+      "loss": 0.1412,
+      "step": 6244
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.887767220902614e-05,
+      "loss": 0.5093,
+      "step": 6245
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.887173396674585e-05,
+      "loss": 0.4577,
+      "step": 6246
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.886579572446557e-05,
+      "loss": 0.4357,
+      "step": 6247
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.885985748218527e-05,
+      "loss": 0.351,
+      "step": 6248
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.885391923990499e-05,
+      "loss": 0.3835,
+      "step": 6249
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.88479809976247e-05,
+      "loss": 0.3432,
+      "step": 6250
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.884204275534442e-05,
+      "loss": 0.2944,
+      "step": 6251
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.883610451306414e-05,
+      "loss": 0.301,
+      "step": 6252
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.883016627078385e-05,
+      "loss": 0.3002,
+      "step": 6253
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.882422802850357e-05,
+      "loss": 0.3464,
+      "step": 6254
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.881828978622327e-05,
+      "loss": 0.3471,
+      "step": 6255
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.8812351543943e-05,
+      "loss": 0.3834,
+      "step": 6256
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.880641330166271e-05,
+      "loss": 0.3502,
+      "step": 6257
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.880047505938244e-05,
+      "loss": 0.3173,
+      "step": 6258
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.879453681710214e-05,
+      "loss": 0.2734,
+      "step": 6259
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.878859857482186e-05,
+      "loss": 0.3792,
+      "step": 6260
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.878266033254157e-05,
+      "loss": 0.2748,
+      "step": 6261
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.877672209026128e-05,
+      "loss": 0.1846,
+      "step": 6262
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.8770783847981e-05,
+      "loss": 0.1986,
+      "step": 6263
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.876484560570072e-05,
+      "loss": 0.2593,
+      "step": 6264
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.875890736342044e-05,
+      "loss": 0.1833,
+      "step": 6265
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.875296912114014e-05,
+      "loss": 0.3183,
+      "step": 6266
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.874703087885986e-05,
+      "loss": 0.2456,
+      "step": 6267
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.874109263657957e-05,
+      "loss": 0.1961,
+      "step": 6268
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.873515439429929e-05,
+      "loss": 0.277,
+      "step": 6269
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.8729216152019e-05,
+      "loss": 0.2598,
+      "step": 6270
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.872327790973873e-05,
+      "loss": 0.2462,
+      "step": 6271
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.871733966745844e-05,
+      "loss": 0.2082,
+      "step": 6272
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.871140142517815e-05,
+      "loss": 0.1717,
+      "step": 6273
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.870546318289787e-05,
+      "loss": 0.3213,
+      "step": 6274
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.869952494061757e-05,
+      "loss": 0.1917,
+      "step": 6275
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.86935866983373e-05,
+      "loss": 0.2394,
+      "step": 6276
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.868764845605701e-05,
+      "loss": 0.2247,
+      "step": 6277
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.868171021377673e-05,
+      "loss": 0.2852,
+      "step": 6278
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.867577197149644e-05,
+      "loss": 0.1963,
+      "step": 6279
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.866983372921616e-05,
+      "loss": 0.279,
+      "step": 6280
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.866389548693587e-05,
+      "loss": 0.2916,
+      "step": 6281
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.865795724465559e-05,
+      "loss": 0.2211,
+      "step": 6282
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.86520190023753e-05,
+      "loss": 0.2524,
+      "step": 6283
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.864608076009502e-05,
+      "loss": 0.2264,
+      "step": 6284
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.864014251781474e-05,
+      "loss": 0.2818,
+      "step": 6285
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.863420427553444e-05,
+      "loss": 0.2451,
+      "step": 6286
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.862826603325416e-05,
+      "loss": 0.2113,
+      "step": 6287
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.862232779097387e-05,
+      "loss": 0.1332,
+      "step": 6288
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.861638954869359e-05,
+      "loss": 0.1365,
+      "step": 6289
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.86104513064133e-05,
+      "loss": 0.1953,
+      "step": 6290
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.860451306413303e-05,
+      "loss": 0.2504,
+      "step": 6291
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.859857482185274e-05,
+      "loss": 0.2008,
+      "step": 6292
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.859263657957244e-05,
+      "loss": 0.1708,
+      "step": 6293
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 6.858669833729216e-05,
+      "loss": 0.239,
+      "step": 6294
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 6.858076009501187e-05,
+      "loss": 0.7589,
+      "step": 6295
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 6.857482185273159e-05,
+      "loss": 0.6401,
+      "step": 6296
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 6.856888361045131e-05,
+      "loss": 0.6977,
+      "step": 6297
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 6.856294536817103e-05,
+      "loss": 0.448,
+      "step": 6298
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 6.855700712589074e-05,
+      "loss": 0.3044,
+      "step": 6299
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 6.855106888361046e-05,
+      "loss": 0.4255,
+      "step": 6300
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 6.854513064133017e-05,
+      "loss": 0.4182,
+      "step": 6301
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.853919239904989e-05,
+      "loss": 0.3258,
+      "step": 6302
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.85332541567696e-05,
+      "loss": 0.2844,
+      "step": 6303
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.852731591448931e-05,
+      "loss": 0.6085,
+      "step": 6304
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.852137767220903e-05,
+      "loss": 0.3026,
+      "step": 6305
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.851543942992874e-05,
+      "loss": 0.3248,
+      "step": 6306
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.850950118764846e-05,
+      "loss": 0.3595,
+      "step": 6307
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.850356294536817e-05,
+      "loss": 0.251,
+      "step": 6308
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.849762470308789e-05,
+      "loss": 0.3002,
+      "step": 6309
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 6.849168646080761e-05,
+      "loss": 0.2371,
+      "step": 6310
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.848574821852733e-05,
+      "loss": 0.3685,
+      "step": 6311
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.847980997624704e-05,
+      "loss": 0.2435,
+      "step": 6312
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.847387173396676e-05,
+      "loss": 0.2352,
+      "step": 6313
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.846793349168646e-05,
+      "loss": 0.2665,
+      "step": 6314
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.846199524940617e-05,
+      "loss": 0.229,
+      "step": 6315
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.845605700712589e-05,
+      "loss": 0.2778,
+      "step": 6316
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.845011876484561e-05,
+      "loss": 0.2544,
+      "step": 6317
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.844418052256533e-05,
+      "loss": 0.2395,
+      "step": 6318
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.843824228028504e-05,
+      "loss": 0.2559,
+      "step": 6319
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.843230403800476e-05,
+      "loss": 0.2911,
+      "step": 6320
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.842636579572447e-05,
+      "loss": 0.2327,
+      "step": 6321
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.842042755344419e-05,
+      "loss": 0.3415,
+      "step": 6322
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.841448931116389e-05,
+      "loss": 0.2342,
+      "step": 6323
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.840855106888361e-05,
+      "loss": 0.2525,
+      "step": 6324
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.840261282660333e-05,
+      "loss": 0.3796,
+      "step": 6325
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.839667458432304e-05,
+      "loss": 0.2104,
+      "step": 6326
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.839073634204276e-05,
+      "loss": 0.2265,
+      "step": 6327
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.838479809976247e-05,
+      "loss": 0.2211,
+      "step": 6328
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.837885985748219e-05,
+      "loss": 0.2161,
+      "step": 6329
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.837292161520191e-05,
+      "loss": 0.3156,
+      "step": 6330
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.836698337292163e-05,
+      "loss": 0.2189,
+      "step": 6331
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.836104513064134e-05,
+      "loss": 0.2064,
+      "step": 6332
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.835510688836106e-05,
+      "loss": 0.253,
+      "step": 6333
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.834916864608076e-05,
+      "loss": 0.2257,
+      "step": 6334
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.834323040380047e-05,
+      "loss": 0.1652,
+      "step": 6335
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.833729216152019e-05,
+      "loss": 0.1906,
+      "step": 6336
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 6.833135391923991e-05,
+      "loss": 0.2735,
+      "step": 6337
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.832541567695963e-05,
+      "loss": 0.2954,
+      "step": 6338
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.831947743467934e-05,
+      "loss": 0.3054,
+      "step": 6339
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.831353919239906e-05,
+      "loss": 0.154,
+      "step": 6340
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.830760095011876e-05,
+      "loss": 0.3264,
+      "step": 6341
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.830166270783848e-05,
+      "loss": 0.2088,
+      "step": 6342
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.829572446555819e-05,
+      "loss": 0.1506,
+      "step": 6343
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.828978622327791e-05,
+      "loss": 0.1297,
+      "step": 6344
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.828384798099763e-05,
+      "loss": 0.6323,
+      "step": 6345
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.827790973871734e-05,
+      "loss": 0.8003,
+      "step": 6346
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.827197149643706e-05,
+      "loss": 0.5832,
+      "step": 6347
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.826603325415677e-05,
+      "loss": 0.4733,
+      "step": 6348
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.826009501187649e-05,
+      "loss": 0.3918,
+      "step": 6349
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.82541567695962e-05,
+      "loss": 0.3645,
+      "step": 6350
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.824821852731593e-05,
+      "loss": 0.3295,
+      "step": 6351
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.824228028503563e-05,
+      "loss": 0.3179,
+      "step": 6352
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.823634204275535e-05,
+      "loss": 0.3026,
+      "step": 6353
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.823040380047506e-05,
+      "loss": 0.4088,
+      "step": 6354
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.822446555819477e-05,
+      "loss": 0.3335,
+      "step": 6355
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.821852731591449e-05,
+      "loss": 0.4453,
+      "step": 6356
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.821258907363421e-05,
+      "loss": 0.3015,
+      "step": 6357
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.820665083135393e-05,
+      "loss": 0.3315,
+      "step": 6358
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.820071258907364e-05,
+      "loss": 0.3463,
+      "step": 6359
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.819477434679336e-05,
+      "loss": 0.3447,
+      "step": 6360
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.818883610451306e-05,
+      "loss": 0.2503,
+      "step": 6361
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.818289786223278e-05,
+      "loss": 0.4119,
+      "step": 6362
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.817695961995249e-05,
+      "loss": 0.2497,
+      "step": 6363
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 6.817102137767222e-05,
+      "loss": 0.2604,
+      "step": 6364
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.816508313539193e-05,
+      "loss": 0.2621,
+      "step": 6365
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.815914489311164e-05,
+      "loss": 0.2127,
+      "step": 6366
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.815320665083136e-05,
+      "loss": 0.2879,
+      "step": 6367
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.814726840855106e-05,
+      "loss": 0.6462,
+      "step": 6368
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.814133016627079e-05,
+      "loss": 0.2119,
+      "step": 6369
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.81353919239905e-05,
+      "loss": 0.3221,
+      "step": 6370
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.812945368171023e-05,
+      "loss": 0.2188,
+      "step": 6371
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.812351543942993e-05,
+      "loss": 0.2098,
+      "step": 6372
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.811757719714965e-05,
+      "loss": 0.2783,
+      "step": 6373
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.811163895486936e-05,
+      "loss": 0.1901,
+      "step": 6374
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.810570071258908e-05,
+      "loss": 0.2252,
+      "step": 6375
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.809976247030879e-05,
+      "loss": 0.1696,
+      "step": 6376
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.809382422802851e-05,
+      "loss": 0.3162,
+      "step": 6377
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.808788598574823e-05,
+      "loss": 0.2033,
+      "step": 6378
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.808194774346793e-05,
+      "loss": 0.4264,
+      "step": 6379
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.807600950118765e-05,
+      "loss": 0.215,
+      "step": 6380
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.807007125890736e-05,
+      "loss": 0.1693,
+      "step": 6381
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 6.806413301662708e-05,
+      "loss": 0.1716,
+      "step": 6382
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.805819477434679e-05,
+      "loss": 0.2783,
+      "step": 6383
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.805225653206652e-05,
+      "loss": 0.203,
+      "step": 6384
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.804631828978623e-05,
+      "loss": 0.1737,
+      "step": 6385
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.804038004750594e-05,
+      "loss": 0.1834,
+      "step": 6386
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.803444180522566e-05,
+      "loss": 0.2221,
+      "step": 6387
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.802850356294536e-05,
+      "loss": 0.1455,
+      "step": 6388
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.802256532066508e-05,
+      "loss": 0.1188,
+      "step": 6389
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.80166270783848e-05,
+      "loss": 0.2828,
+      "step": 6390
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.801068883610452e-05,
+      "loss": 0.1802,
+      "step": 6391
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.800475059382423e-05,
+      "loss": 0.1939,
+      "step": 6392
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.799881235154395e-05,
+      "loss": 0.7387,
+      "step": 6393
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.799287410926366e-05,
+      "loss": 0.2726,
+      "step": 6394
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.798693586698338e-05,
+      "loss": 0.5741,
+      "step": 6395
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.798099762470309e-05,
+      "loss": 0.4739,
+      "step": 6396
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.79750593824228e-05,
+      "loss": 0.4952,
+      "step": 6397
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.796912114014253e-05,
+      "loss": 0.5151,
+      "step": 6398
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.796318289786223e-05,
+      "loss": 0.4976,
+      "step": 6399
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.795724465558195e-05,
+      "loss": 0.2771,
+      "step": 6400
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.795130641330166e-05,
+      "loss": 0.3531,
+      "step": 6401
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.794536817102138e-05,
+      "loss": 0.3113,
+      "step": 6402
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.793942992874109e-05,
+      "loss": 0.2701,
+      "step": 6403
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.793349168646082e-05,
+      "loss": 0.4451,
+      "step": 6404
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.792755344418053e-05,
+      "loss": 0.3214,
+      "step": 6405
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.792161520190025e-05,
+      "loss": 0.3002,
+      "step": 6406
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.791567695961996e-05,
+      "loss": 0.3176,
+      "step": 6407
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.790973871733966e-05,
+      "loss": 0.287,
+      "step": 6408
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 6.790380047505938e-05,
+      "loss": 0.2741,
+      "step": 6409
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 6.78978622327791e-05,
+      "loss": 0.2708,
+      "step": 6410
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 6.789192399049882e-05,
+      "loss": 0.299,
+      "step": 6411
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 6.788598574821853e-05,
+      "loss": 0.3099,
+      "step": 6412
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 6.788004750593825e-05,
+      "loss": 0.2337,
+      "step": 6413
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 6.787410926365796e-05,
+      "loss": 0.248,
+      "step": 6414
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 6.786817102137768e-05,
+      "loss": 0.2266,
+      "step": 6415
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 6.786223277909738e-05,
+      "loss": 0.2383,
+      "step": 6416
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 6.78562945368171e-05,
+      "loss": 0.2105,
+      "step": 6417
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.785035629453683e-05,
+      "loss": 0.265,
+      "step": 6418
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.784441805225653e-05,
+      "loss": 0.2445,
+      "step": 6419
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.783847980997625e-05,
+      "loss": 0.2723,
+      "step": 6420
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.783254156769596e-05,
+      "loss": 0.2889,
+      "step": 6421
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.782660332541568e-05,
+      "loss": 0.2481,
+      "step": 6422
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.782066508313539e-05,
+      "loss": 0.5328,
+      "step": 6423
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.781472684085512e-05,
+      "loss": 0.2961,
+      "step": 6424
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.780878859857483e-05,
+      "loss": 0.2031,
+      "step": 6425
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.780285035629455e-05,
+      "loss": 0.2114,
+      "step": 6426
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.779691211401425e-05,
+      "loss": 0.275,
+      "step": 6427
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.779097387173396e-05,
+      "loss": 0.239,
+      "step": 6428
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.778503562945368e-05,
+      "loss": 0.2484,
+      "step": 6429
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.77790973871734e-05,
+      "loss": 0.2174,
+      "step": 6430
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.777315914489312e-05,
+      "loss": 0.2867,
+      "step": 6431
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.776722090261283e-05,
+      "loss": 0.133,
+      "step": 6432
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.776128266033255e-05,
+      "loss": 0.2766,
+      "step": 6433
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.775534441805226e-05,
+      "loss": 0.2416,
+      "step": 6434
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 6.774940617577198e-05,
+      "loss": 0.221,
+      "step": 6435
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.774346793349168e-05,
+      "loss": 0.2301,
+      "step": 6436
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.77375296912114e-05,
+      "loss": 0.2342,
+      "step": 6437
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.773159144893112e-05,
+      "loss": 0.1598,
+      "step": 6438
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.772565320665083e-05,
+      "loss": 0.2658,
+      "step": 6439
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.771971496437055e-05,
+      "loss": 0.2388,
+      "step": 6440
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.771377672209026e-05,
+      "loss": 0.1357,
+      "step": 6441
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.770783847980998e-05,
+      "loss": 0.1358,
+      "step": 6442
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.77019002375297e-05,
+      "loss": 0.2338,
+      "step": 6443
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.769596199524942e-05,
+      "loss": 0.5014,
+      "step": 6444
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.769002375296913e-05,
+      "loss": 0.6918,
+      "step": 6445
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.768408551068885e-05,
+      "loss": 0.6155,
+      "step": 6446
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.767814726840855e-05,
+      "loss": 0.6499,
+      "step": 6447
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.767220902612826e-05,
+      "loss": 0.4946,
+      "step": 6448
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.766627078384798e-05,
+      "loss": 0.4634,
+      "step": 6449
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.76603325415677e-05,
+      "loss": 0.2716,
+      "step": 6450
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.765439429928742e-05,
+      "loss": 0.3507,
+      "step": 6451
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.764845605700713e-05,
+      "loss": 0.3491,
+      "step": 6452
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.764251781472685e-05,
+      "loss": 0.3328,
+      "step": 6453
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 6.763657957244655e-05,
+      "loss": 0.3225,
+      "step": 6454
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 6.763064133016627e-05,
+      "loss": 0.3687,
+      "step": 6455
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 6.762470308788598e-05,
+      "loss": 0.2813,
+      "step": 6456
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 6.76187648456057e-05,
+      "loss": 0.3189,
+      "step": 6457
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 6.761282660332542e-05,
+      "loss": 0.246,
+      "step": 6458
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 6.760688836104513e-05,
+      "loss": 0.2019,
+      "step": 6459
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 6.760095011876485e-05,
+      "loss": 0.1896,
+      "step": 6460
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 6.759501187648456e-05,
+      "loss": 0.254,
+      "step": 6461
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 6.758907363420428e-05,
+      "loss": 0.2795,
+      "step": 6462
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.7583135391924e-05,
+      "loss": 0.2432,
+      "step": 6463
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.757719714964372e-05,
+      "loss": 0.2183,
+      "step": 6464
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.757125890736342e-05,
+      "loss": 0.2182,
+      "step": 6465
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.756532066508314e-05,
+      "loss": 0.2083,
+      "step": 6466
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.755938242280285e-05,
+      "loss": 0.2091,
+      "step": 6467
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.755344418052257e-05,
+      "loss": 0.2456,
+      "step": 6468
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.754750593824228e-05,
+      "loss": 0.3259,
+      "step": 6469
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.7541567695962e-05,
+      "loss": 0.2771,
+      "step": 6470
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.753562945368172e-05,
+      "loss": 0.2651,
+      "step": 6471
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.752969121140143e-05,
+      "loss": 0.3042,
+      "step": 6472
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.752375296912115e-05,
+      "loss": 0.3092,
+      "step": 6473
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.751781472684085e-05,
+      "loss": 0.5151,
+      "step": 6474
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.751187648456057e-05,
+      "loss": 0.3123,
+      "step": 6475
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.750593824228028e-05,
+      "loss": 0.2584,
+      "step": 6476
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.750000000000001e-05,
+      "loss": 0.2618,
+      "step": 6477
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.749406175771972e-05,
+      "loss": 0.2358,
+      "step": 6478
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.748812351543944e-05,
+      "loss": 0.2083,
+      "step": 6479
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 6.748218527315915e-05,
+      "loss": 0.1709,
+      "step": 6480
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.747624703087886e-05,
+      "loss": 0.2149,
+      "step": 6481
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.747030878859858e-05,
+      "loss": 0.3034,
+      "step": 6482
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.74643705463183e-05,
+      "loss": 0.2462,
+      "step": 6483
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.745843230403802e-05,
+      "loss": 0.1543,
+      "step": 6484
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.745249406175772e-05,
+      "loss": 0.2007,
+      "step": 6485
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.744655581947744e-05,
+      "loss": 0.2158,
+      "step": 6486
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.744061757719715e-05,
+      "loss": 0.3515,
+      "step": 6487
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.743467933491687e-05,
+      "loss": 0.2443,
+      "step": 6488
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.742874109263658e-05,
+      "loss": 0.2612,
+      "step": 6489
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.74228028503563e-05,
+      "loss": 0.2322,
+      "step": 6490
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.741686460807602e-05,
+      "loss": 0.3817,
+      "step": 6491
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.741092636579572e-05,
+      "loss": 0.1729,
+      "step": 6492
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.740498812351545e-05,
+      "loss": 0.1903,
+      "step": 6493
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.739904988123515e-05,
+      "loss": 0.1527,
+      "step": 6494
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.739311163895487e-05,
+      "loss": 0.5886,
+      "step": 6495
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.738717339667458e-05,
+      "loss": 0.5219,
+      "step": 6496
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.738123515439431e-05,
+      "loss": 0.5709,
+      "step": 6497
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.737529691211402e-05,
+      "loss": 0.3901,
+      "step": 6498
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 6.736935866983374e-05,
+      "loss": 0.3391,
+      "step": 6499
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 6.736342042755345e-05,
+      "loss": 0.2634,
+      "step": 6500
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 6.735748218527315e-05,
+      "loss": 0.2636,
+      "step": 6501
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 6.735154394299287e-05,
+      "loss": 0.2895,
+      "step": 6502
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 6.73456057007126e-05,
+      "loss": 0.3035,
+      "step": 6503
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 6.733966745843231e-05,
+      "loss": 0.4372,
+      "step": 6504
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 6.733372921615202e-05,
+      "loss": 0.2931,
+      "step": 6505
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 6.732779097387174e-05,
+      "loss": 0.2875,
+      "step": 6506
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 6.732185273159145e-05,
+      "loss": 0.3667,
+      "step": 6507
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.731591448931117e-05,
+      "loss": 0.3631,
+      "step": 6508
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.730997624703088e-05,
+      "loss": 0.5488,
+      "step": 6509
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.73040380047506e-05,
+      "loss": 0.2473,
+      "step": 6510
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.729809976247032e-05,
+      "loss": 0.227,
+      "step": 6511
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.729216152019002e-05,
+      "loss": 0.2013,
+      "step": 6512
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.728622327790974e-05,
+      "loss": 0.32,
+      "step": 6513
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.728028503562945e-05,
+      "loss": 0.3036,
+      "step": 6514
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.727434679334917e-05,
+      "loss": 0.1935,
+      "step": 6515
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.726840855106888e-05,
+      "loss": 0.3289,
+      "step": 6516
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.726247030878861e-05,
+      "loss": 0.233,
+      "step": 6517
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.725653206650832e-05,
+      "loss": 0.1949,
+      "step": 6518
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.725059382422804e-05,
+      "loss": 0.2348,
+      "step": 6519
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.724465558194775e-05,
+      "loss": 0.2527,
+      "step": 6520
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.723871733966745e-05,
+      "loss": 0.2605,
+      "step": 6521
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.723277909738717e-05,
+      "loss": 0.231,
+      "step": 6522
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.72268408551069e-05,
+      "loss": 0.2433,
+      "step": 6523
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.722090261282661e-05,
+      "loss": 0.3835,
+      "step": 6524
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.721496437054632e-05,
+      "loss": 0.3105,
+      "step": 6525
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.720902612826604e-05,
+      "loss": 0.2211,
+      "step": 6526
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.720308788598575e-05,
+      "loss": 0.3025,
+      "step": 6527
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.719714964370547e-05,
+      "loss": 0.3042,
+      "step": 6528
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.719121140142517e-05,
+      "loss": 0.2489,
+      "step": 6529
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.71852731591449e-05,
+      "loss": 0.2319,
+      "step": 6530
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.717933491686462e-05,
+      "loss": 0.2318,
+      "step": 6531
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.717339667458432e-05,
+      "loss": 0.1664,
+      "step": 6532
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.716745843230404e-05,
+      "loss": 0.1598,
+      "step": 6533
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.716152019002375e-05,
+      "loss": 0.2482,
+      "step": 6534
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.715558194774347e-05,
+      "loss": 0.284,
+      "step": 6535
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.714964370546318e-05,
+      "loss": 0.2635,
+      "step": 6536
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.714370546318291e-05,
+      "loss": 0.1969,
+      "step": 6537
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.713776722090262e-05,
+      "loss": 0.238,
+      "step": 6538
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.713182897862234e-05,
+      "loss": 0.194,
+      "step": 6539
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.712589073634204e-05,
+      "loss": 0.1269,
+      "step": 6540
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.711995249406176e-05,
+      "loss": 0.2066,
+      "step": 6541
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.711401425178147e-05,
+      "loss": 0.1627,
+      "step": 6542
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.710807600950119e-05,
+      "loss": 0.2211,
+      "step": 6543
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.710213776722091e-05,
+      "loss": 0.0844,
+      "step": 6544
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.709619952494062e-05,
+      "loss": 0.738,
+      "step": 6545
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.709026128266034e-05,
+      "loss": 0.5609,
+      "step": 6546
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.708432304038005e-05,
+      "loss": 0.4127,
+      "step": 6547
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.707838479809977e-05,
+      "loss": 0.4878,
+      "step": 6548
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.707244655581947e-05,
+      "loss": 0.3529,
+      "step": 6549
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.70665083135392e-05,
+      "loss": 0.3411,
+      "step": 6550
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.706057007125891e-05,
+      "loss": 0.321,
+      "step": 6551
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.705463182897862e-05,
+      "loss": 0.4839,
+      "step": 6552
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.704869358669834e-05,
+      "loss": 0.3073,
+      "step": 6553
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.704275534441805e-05,
+      "loss": 0.2797,
+      "step": 6554
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.703681710213777e-05,
+      "loss": 0.2404,
+      "step": 6555
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.703087885985749e-05,
+      "loss": 0.3335,
+      "step": 6556
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.702494061757721e-05,
+      "loss": 0.3488,
+      "step": 6557
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.701900237529692e-05,
+      "loss": 0.3939,
+      "step": 6558
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.701306413301664e-05,
+      "loss": 0.2377,
+      "step": 6559
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.700712589073634e-05,
+      "loss": 0.3292,
+      "step": 6560
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.700118764845606e-05,
+      "loss": 0.2207,
+      "step": 6561
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.699524940617577e-05,
+      "loss": 0.3172,
+      "step": 6562
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.698931116389549e-05,
+      "loss": 0.3055,
+      "step": 6563
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.698337292161521e-05,
+      "loss": 0.1722,
+      "step": 6564
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.697743467933492e-05,
+      "loss": 0.2317,
+      "step": 6565
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.697149643705464e-05,
+      "loss": 0.3401,
+      "step": 6566
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.696555819477435e-05,
+      "loss": 0.1729,
+      "step": 6567
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.695961995249407e-05,
+      "loss": 0.2946,
+      "step": 6568
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.695368171021377e-05,
+      "loss": 0.2346,
+      "step": 6569
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.694774346793349e-05,
+      "loss": 0.2252,
+      "step": 6570
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.694180522565321e-05,
+      "loss": 0.2689,
+      "step": 6571
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.693586698337293e-05,
+      "loss": 0.207,
+      "step": 6572
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.692992874109264e-05,
+      "loss": 0.2097,
+      "step": 6573
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.692399049881235e-05,
+      "loss": 0.2058,
+      "step": 6574
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.691805225653207e-05,
+      "loss": 0.8377,
+      "step": 6575
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.691211401425179e-05,
+      "loss": 0.2999,
+      "step": 6576
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.690617577197151e-05,
+      "loss": 0.2573,
+      "step": 6577
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.690023752969121e-05,
+      "loss": 0.2095,
+      "step": 6578
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.689429928741094e-05,
+      "loss": 0.2211,
+      "step": 6579
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.688836104513064e-05,
+      "loss": 0.2486,
+      "step": 6580
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.688242280285036e-05,
+      "loss": 0.1527,
+      "step": 6581
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.687648456057007e-05,
+      "loss": 0.1703,
+      "step": 6582
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.687054631828979e-05,
+      "loss": 0.2072,
+      "step": 6583
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.686460807600951e-05,
+      "loss": 0.1754,
+      "step": 6584
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.685866983372922e-05,
+      "loss": 0.2344,
+      "step": 6585
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.685273159144894e-05,
+      "loss": 0.1569,
+      "step": 6586
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.684679334916864e-05,
+      "loss": 0.1993,
+      "step": 6587
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.684085510688836e-05,
+      "loss": 0.2432,
+      "step": 6588
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.683491686460807e-05,
+      "loss": 0.1721,
+      "step": 6589
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.682897862232779e-05,
+      "loss": 0.1118,
+      "step": 6590
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.682304038004751e-05,
+      "loss": 0.1212,
+      "step": 6591
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.681710213776723e-05,
+      "loss": 0.1232,
+      "step": 6592
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.681116389548694e-05,
+      "loss": 0.1811,
+      "step": 6593
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.680522565320665e-05,
+      "loss": 0.32,
+      "step": 6594
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.679928741092637e-05,
+      "loss": 0.6837,
+      "step": 6595
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.679334916864609e-05,
+      "loss": 0.722,
+      "step": 6596
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.67874109263658e-05,
+      "loss": 0.5612,
+      "step": 6597
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.678147268408551e-05,
+      "loss": 0.4663,
+      "step": 6598
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.677553444180523e-05,
+      "loss": 0.3934,
+      "step": 6599
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.676959619952494e-05,
+      "loss": 0.2944,
+      "step": 6600
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.676365795724466e-05,
+      "loss": 0.386,
+      "step": 6601
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.675771971496437e-05,
+      "loss": 0.3607,
+      "step": 6602
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.675178147268409e-05,
+      "loss": 0.29,
+      "step": 6603
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.674584323040381e-05,
+      "loss": 0.2545,
+      "step": 6604
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.673990498812352e-05,
+      "loss": 0.4166,
+      "step": 6605
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.673396674584324e-05,
+      "loss": 0.4022,
+      "step": 6606
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.672802850356294e-05,
+      "loss": 0.3822,
+      "step": 6607
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.672209026128266e-05,
+      "loss": 0.2526,
+      "step": 6608
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.671615201900237e-05,
+      "loss": 0.3467,
+      "step": 6609
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.67102137767221e-05,
+      "loss": 0.2794,
+      "step": 6610
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.670427553444181e-05,
+      "loss": 0.2189,
+      "step": 6611
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.669833729216153e-05,
+      "loss": 0.2146,
+      "step": 6612
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.669239904988124e-05,
+      "loss": 0.2773,
+      "step": 6613
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.668646080760094e-05,
+      "loss": 0.4007,
+      "step": 6614
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.668052256532066e-05,
+      "loss": 0.3507,
+      "step": 6615
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.667458432304039e-05,
+      "loss": 0.2197,
+      "step": 6616
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.66686460807601e-05,
+      "loss": 0.3968,
+      "step": 6617
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.666270783847981e-05,
+      "loss": 0.2707,
+      "step": 6618
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.665676959619953e-05,
+      "loss": 0.2542,
+      "step": 6619
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.665083135391924e-05,
+      "loss": 0.2252,
+      "step": 6620
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.664489311163896e-05,
+      "loss": 0.1887,
+      "step": 6621
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.663895486935867e-05,
+      "loss": 0.2277,
+      "step": 6622
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.663301662707839e-05,
+      "loss": 0.2365,
+      "step": 6623
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.662707838479811e-05,
+      "loss": 0.2449,
+      "step": 6624
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.662114014251781e-05,
+      "loss": 0.1918,
+      "step": 6625
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.661520190023753e-05,
+      "loss": 0.2785,
+      "step": 6626
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.660926365795724e-05,
+      "loss": 0.2206,
+      "step": 6627
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.660332541567696e-05,
+      "loss": 0.2544,
+      "step": 6628
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.659738717339667e-05,
+      "loss": 0.2216,
+      "step": 6629
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.65914489311164e-05,
+      "loss": 0.1893,
+      "step": 6630
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.658551068883611e-05,
+      "loss": 0.22,
+      "step": 6631
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 6.657957244655583e-05,
+      "loss": 0.1519,
+      "step": 6632
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.657363420427554e-05,
+      "loss": 0.1726,
+      "step": 6633
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.656769596199526e-05,
+      "loss": 0.2468,
+      "step": 6634
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.656175771971496e-05,
+      "loss": 0.2218,
+      "step": 6635
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.655581947743468e-05,
+      "loss": 0.4101,
+      "step": 6636
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.65498812351544e-05,
+      "loss": 0.2313,
+      "step": 6637
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.654394299287411e-05,
+      "loss": 0.1799,
+      "step": 6638
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.653800475059383e-05,
+      "loss": 0.0989,
+      "step": 6639
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.653206650831354e-05,
+      "loss": 0.1477,
+      "step": 6640
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.652612826603326e-05,
+      "loss": 0.2018,
+      "step": 6641
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.652019002375297e-05,
+      "loss": 0.0763,
+      "step": 6642
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.651425178147269e-05,
+      "loss": 0.3473,
+      "step": 6643
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.65083135391924e-05,
+      "loss": 0.1695,
+      "step": 6644
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.650237529691211e-05,
+      "loss": 0.7145,
+      "step": 6645
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.649643705463183e-05,
+      "loss": 0.6019,
+      "step": 6646
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.649049881235154e-05,
+      "loss": 0.4504,
+      "step": 6647
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.648456057007126e-05,
+      "loss": 0.4168,
+      "step": 6648
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.647862232779097e-05,
+      "loss": 0.3401,
+      "step": 6649
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.64726840855107e-05,
+      "loss": 0.3021,
+      "step": 6650
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.646674584323041e-05,
+      "loss": 0.2533,
+      "step": 6651
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.646080760095013e-05,
+      "loss": 0.2613,
+      "step": 6652
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.645486935866984e-05,
+      "loss": 0.3194,
+      "step": 6653
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.644893111638956e-05,
+      "loss": 0.4309,
+      "step": 6654
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.644299287410926e-05,
+      "loss": 0.2727,
+      "step": 6655
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.643705463182898e-05,
+      "loss": 0.3438,
+      "step": 6656
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.64311163895487e-05,
+      "loss": 0.316,
+      "step": 6657
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.642517814726841e-05,
+      "loss": 0.4015,
+      "step": 6658
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.641923990498813e-05,
+      "loss": 0.4191,
+      "step": 6659
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.641330166270784e-05,
+      "loss": 0.2126,
+      "step": 6660
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.640736342042756e-05,
+      "loss": 0.222,
+      "step": 6661
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.640142517814726e-05,
+      "loss": 0.2385,
+      "step": 6662
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.639548693586698e-05,
+      "loss": 0.1817,
+      "step": 6663
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.63895486935867e-05,
+      "loss": 0.2781,
+      "step": 6664
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.638361045130643e-05,
+      "loss": 0.289,
+      "step": 6665
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.637767220902613e-05,
+      "loss": 0.2464,
+      "step": 6666
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.637173396674584e-05,
+      "loss": 0.3298,
+      "step": 6667
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.636579572446556e-05,
+      "loss": 0.2877,
+      "step": 6668
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.635985748218527e-05,
+      "loss": 0.264,
+      "step": 6669
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.6353919239905e-05,
+      "loss": 0.2018,
+      "step": 6670
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.63479809976247e-05,
+      "loss": 0.2917,
+      "step": 6671
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.634204275534443e-05,
+      "loss": 0.3393,
+      "step": 6672
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.633610451306413e-05,
+      "loss": 0.2074,
+      "step": 6673
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.633016627078385e-05,
+      "loss": 0.2032,
+      "step": 6674
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.632422802850356e-05,
+      "loss": 0.2238,
+      "step": 6675
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.631828978622328e-05,
+      "loss": 0.1852,
+      "step": 6676
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.6312351543943e-05,
+      "loss": 0.2462,
+      "step": 6677
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.630641330166271e-05,
+      "loss": 0.18,
+      "step": 6678
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.630047505938243e-05,
+      "loss": 0.2507,
+      "step": 6679
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.629453681710214e-05,
+      "loss": 0.2046,
+      "step": 6680
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.628859857482186e-05,
+      "loss": 0.3857,
+      "step": 6681
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.628266033254156e-05,
+      "loss": 0.1682,
+      "step": 6682
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.627672209026128e-05,
+      "loss": 0.1833,
+      "step": 6683
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.6270783847981e-05,
+      "loss": 0.155,
+      "step": 6684
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.626484560570072e-05,
+      "loss": 0.1501,
+      "step": 6685
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.625890736342043e-05,
+      "loss": 0.2751,
+      "step": 6686
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.625296912114014e-05,
+      "loss": 0.1664,
+      "step": 6687
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.624703087885986e-05,
+      "loss": 0.3124,
+      "step": 6688
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.624109263657958e-05,
+      "loss": 0.2436,
+      "step": 6689
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.62351543942993e-05,
+      "loss": 0.8511,
+      "step": 6690
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.6229216152019e-05,
+      "loss": 0.2096,
+      "step": 6691
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.622327790973873e-05,
+      "loss": 0.9423,
+      "step": 6692
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.621733966745843e-05,
+      "loss": 0.1842,
+      "step": 6693
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.621140142517815e-05,
+      "loss": 0.233,
+      "step": 6694
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.620546318289786e-05,
+      "loss": 0.8173,
+      "step": 6695
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.619952494061758e-05,
+      "loss": 0.7173,
+      "step": 6696
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.61935866983373e-05,
+      "loss": 0.6126,
+      "step": 6697
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.618764845605701e-05,
+      "loss": 0.4093,
+      "step": 6698
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.618171021377673e-05,
+      "loss": 0.4782,
+      "step": 6699
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.617577197149643e-05,
+      "loss": 0.3252,
+      "step": 6700
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.616983372921615e-05,
+      "loss": 0.2623,
+      "step": 6701
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.616389548693586e-05,
+      "loss": 0.2878,
+      "step": 6702
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.615795724465558e-05,
+      "loss": 0.3205,
+      "step": 6703
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.61520190023753e-05,
+      "loss": 0.2905,
+      "step": 6704
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.614608076009502e-05,
+      "loss": 0.2725,
+      "step": 6705
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.614014251781473e-05,
+      "loss": 0.3894,
+      "step": 6706
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.613420427553445e-05,
+      "loss": 0.2942,
+      "step": 6707
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.612826603325416e-05,
+      "loss": 0.2871,
+      "step": 6708
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.612232779097388e-05,
+      "loss": 0.3532,
+      "step": 6709
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.61163895486936e-05,
+      "loss": 0.2168,
+      "step": 6710
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.61104513064133e-05,
+      "loss": 0.3992,
+      "step": 6711
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.610451306413302e-05,
+      "loss": 0.3893,
+      "step": 6712
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.609857482185273e-05,
+      "loss": 0.209,
+      "step": 6713
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.609263657957245e-05,
+      "loss": 0.2817,
+      "step": 6714
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.608669833729216e-05,
+      "loss": 0.3281,
+      "step": 6715
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.608076009501188e-05,
+      "loss": 0.2129,
+      "step": 6716
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.60748218527316e-05,
+      "loss": 0.226,
+      "step": 6717
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.60688836104513e-05,
+      "loss": 0.3286,
+      "step": 6718
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.606294536817103e-05,
+      "loss": 0.4369,
+      "step": 6719
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.605700712589073e-05,
+      "loss": 0.225,
+      "step": 6720
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.605106888361045e-05,
+      "loss": 0.2151,
+      "step": 6721
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.604513064133016e-05,
+      "loss": 0.2249,
+      "step": 6722
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.60391923990499e-05,
+      "loss": 0.1821,
+      "step": 6723
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.60332541567696e-05,
+      "loss": 0.2342,
+      "step": 6724
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.602731591448932e-05,
+      "loss": 0.2096,
+      "step": 6725
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.602137767220903e-05,
+      "loss": 0.2311,
+      "step": 6726
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.601543942992875e-05,
+      "loss": 0.2079,
+      "step": 6727
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.600950118764846e-05,
+      "loss": 0.2897,
+      "step": 6728
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.600356294536818e-05,
+      "loss": 0.2432,
+      "step": 6729
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 6.59976247030879e-05,
+      "loss": 0.2176,
+      "step": 6730
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.59916864608076e-05,
+      "loss": 0.2088,
+      "step": 6731
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.598574821852732e-05,
+      "loss": 0.2708,
+      "step": 6732
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.597980997624703e-05,
+      "loss": 0.2459,
+      "step": 6733
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.597387173396675e-05,
+      "loss": 0.2907,
+      "step": 6734
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.596793349168646e-05,
+      "loss": 0.2084,
+      "step": 6735
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.596199524940618e-05,
+      "loss": 0.2138,
+      "step": 6736
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.59560570071259e-05,
+      "loss": 0.1853,
+      "step": 6737
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.595011876484562e-05,
+      "loss": 0.1751,
+      "step": 6738
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.594418052256532e-05,
+      "loss": 0.1627,
+      "step": 6739
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.593824228028503e-05,
+      "loss": 0.166,
+      "step": 6740
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.593230403800475e-05,
+      "loss": 0.3002,
+      "step": 6741
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.592636579572446e-05,
+      "loss": 0.1632,
+      "step": 6742
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.592042755344419e-05,
+      "loss": 0.189,
+      "step": 6743
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.59144893111639e-05,
+      "loss": 0.4044,
+      "step": 6744
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.590855106888362e-05,
+      "loss": 0.7947,
+      "step": 6745
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.590261282660333e-05,
+      "loss": 0.644,
+      "step": 6746
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.589667458432305e-05,
+      "loss": 0.4977,
+      "step": 6747
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.589073634204275e-05,
+      "loss": 0.4655,
+      "step": 6748
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.588479809976247e-05,
+      "loss": 0.3563,
+      "step": 6749
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.58788598574822e-05,
+      "loss": 0.2631,
+      "step": 6750
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.58729216152019e-05,
+      "loss": 0.2497,
+      "step": 6751
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.586698337292162e-05,
+      "loss": 0.2805,
+      "step": 6752
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.586104513064133e-05,
+      "loss": 0.2804,
+      "step": 6753
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.585510688836105e-05,
+      "loss": 0.3301,
+      "step": 6754
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.584916864608076e-05,
+      "loss": 0.278,
+      "step": 6755
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 6.584323040380048e-05,
+      "loss": 0.3982,
+      "step": 6756
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.58372921615202e-05,
+      "loss": 0.3005,
+      "step": 6757
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.583135391923992e-05,
+      "loss": 0.2684,
+      "step": 6758
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.582541567695962e-05,
+      "loss": 0.3389,
+      "step": 6759
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.581947743467933e-05,
+      "loss": 0.3173,
+      "step": 6760
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.581353919239905e-05,
+      "loss": 0.3198,
+      "step": 6761
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.580760095011876e-05,
+      "loss": 0.2528,
+      "step": 6762
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.580166270783849e-05,
+      "loss": 0.2169,
+      "step": 6763
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.57957244655582e-05,
+      "loss": 0.2973,
+      "step": 6764
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.578978622327792e-05,
+      "loss": 0.3867,
+      "step": 6765
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.578384798099763e-05,
+      "loss": 0.2221,
+      "step": 6766
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.577790973871735e-05,
+      "loss": 0.2448,
+      "step": 6767
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.577197149643705e-05,
+      "loss": 0.3031,
+      "step": 6768
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.576603325415677e-05,
+      "loss": 0.2901,
+      "step": 6769
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.57600950118765e-05,
+      "loss": 0.1625,
+      "step": 6770
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.57541567695962e-05,
+      "loss": 0.2808,
+      "step": 6771
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.574821852731592e-05,
+      "loss": 0.2093,
+      "step": 6772
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.574228028503563e-05,
+      "loss": 0.2586,
+      "step": 6773
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.573634204275535e-05,
+      "loss": 0.2024,
+      "step": 6774
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 6.573040380047505e-05,
+      "loss": 0.1967,
+      "step": 6775
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 6.572446555819477e-05,
+      "loss": 0.2634,
+      "step": 6776
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 6.57185273159145e-05,
+      "loss": 0.212,
+      "step": 6777
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 6.571258907363422e-05,
+      "loss": 0.2211,
+      "step": 6778
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 6.570665083135392e-05,
+      "loss": 0.2206,
+      "step": 6779
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 6.570071258907363e-05,
+      "loss": 0.1975,
+      "step": 6780
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 6.569477434679335e-05,
+      "loss": 0.1706,
+      "step": 6781
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 6.568883610451306e-05,
+      "loss": 0.2138,
+      "step": 6782
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 6.568289786223279e-05,
+      "loss": 0.1452,
+      "step": 6783
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.56769596199525e-05,
+      "loss": 0.1792,
+      "step": 6784
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.567102137767222e-05,
+      "loss": 0.177,
+      "step": 6785
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.566508313539192e-05,
+      "loss": 0.2659,
+      "step": 6786
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.565914489311164e-05,
+      "loss": 0.2477,
+      "step": 6787
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.565320665083135e-05,
+      "loss": 0.1678,
+      "step": 6788
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.564726840855107e-05,
+      "loss": 0.1709,
+      "step": 6789
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.564133016627079e-05,
+      "loss": 0.264,
+      "step": 6790
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.56353919239905e-05,
+      "loss": 0.1968,
+      "step": 6791
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.562945368171022e-05,
+      "loss": 0.274,
+      "step": 6792
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.562351543942993e-05,
+      "loss": 0.265,
+      "step": 6793
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.561757719714965e-05,
+      "loss": 0.2132,
+      "step": 6794
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.561163895486935e-05,
+      "loss": 0.6122,
+      "step": 6795
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.560570071258907e-05,
+      "loss": 0.5151,
+      "step": 6796
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.55997624703088e-05,
+      "loss": 0.5625,
+      "step": 6797
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.559382422802851e-05,
+      "loss": 0.3805,
+      "step": 6798
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.558788598574822e-05,
+      "loss": 0.2824,
+      "step": 6799
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.558194774346794e-05,
+      "loss": 0.3206,
+      "step": 6800
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.557600950118765e-05,
+      "loss": 0.3558,
+      "step": 6801
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 6.557007125890736e-05,
+      "loss": 0.3427,
+      "step": 6802
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 6.556413301662709e-05,
+      "loss": 0.4236,
+      "step": 6803
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 6.55581947743468e-05,
+      "loss": 0.2624,
+      "step": 6804
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 6.555225653206652e-05,
+      "loss": 0.3283,
+      "step": 6805
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 6.554631828978622e-05,
+      "loss": 0.2605,
+      "step": 6806
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 6.554038004750594e-05,
+      "loss": 0.3462,
+      "step": 6807
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 6.553444180522565e-05,
+      "loss": 0.304,
+      "step": 6808
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 6.552850356294537e-05,
+      "loss": 0.2974,
+      "step": 6809
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 6.552256532066509e-05,
+      "loss": 0.2959,
+      "step": 6810
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.55166270783848e-05,
+      "loss": 0.2437,
+      "step": 6811
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.551068883610452e-05,
+      "loss": 0.2126,
+      "step": 6812
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.550475059382422e-05,
+      "loss": 0.2205,
+      "step": 6813
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.549881235154395e-05,
+      "loss": 0.2166,
+      "step": 6814
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.549287410926365e-05,
+      "loss": 0.2955,
+      "step": 6815
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.548693586698337e-05,
+      "loss": 0.255,
+      "step": 6816
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.548099762470309e-05,
+      "loss": 0.2951,
+      "step": 6817
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.547505938242281e-05,
+      "loss": 0.2583,
+      "step": 6818
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.546912114014252e-05,
+      "loss": 0.1969,
+      "step": 6819
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 6.546318289786224e-05,
+      "loss": 0.3093,
+      "step": 6820
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 6.545724465558195e-05,
+      "loss": 0.5258,
+      "step": 6821
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 6.545130641330167e-05,
+      "loss": 0.2244,
+      "step": 6822
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 6.544536817102139e-05,
+      "loss": 0.199,
+      "step": 6823
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 6.54394299287411e-05,
+      "loss": 0.1641,
+      "step": 6824
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 6.543349168646081e-05,
+      "loss": 0.3232,
+      "step": 6825
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 6.542755344418052e-05,
+      "loss": 0.2918,
+      "step": 6826
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 6.542161520190024e-05,
+      "loss": 0.1686,
+      "step": 6827
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 6.541567695961995e-05,
+      "loss": 0.2012,
+      "step": 6828
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.540973871733967e-05,
+      "loss": 0.2935,
+      "step": 6829
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.540380047505939e-05,
+      "loss": 0.1914,
+      "step": 6830
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.539786223277911e-05,
+      "loss": 0.2778,
+      "step": 6831
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.539192399049882e-05,
+      "loss": 0.1798,
+      "step": 6832
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.538598574821852e-05,
+      "loss": 0.1584,
+      "step": 6833
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.538004750593824e-05,
+      "loss": 0.2315,
+      "step": 6834
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.537410926365795e-05,
+      "loss": 0.1978,
+      "step": 6835
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.536817102137767e-05,
+      "loss": 0.2378,
+      "step": 6836
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.536223277909739e-05,
+      "loss": 0.1955,
+      "step": 6837
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.535629453681711e-05,
+      "loss": 0.1744,
+      "step": 6838
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.535035629453682e-05,
+      "loss": 0.2796,
+      "step": 6839
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.534441805225654e-05,
+      "loss": 0.176,
+      "step": 6840
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.533847980997625e-05,
+      "loss": 0.1968,
+      "step": 6841
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.533254156769597e-05,
+      "loss": 0.3253,
+      "step": 6842
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.532660332541569e-05,
+      "loss": 0.149,
+      "step": 6843
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.532066508313539e-05,
+      "loss": 0.1869,
+      "step": 6844
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.531472684085511e-05,
+      "loss": 0.5916,
+      "step": 6845
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.530878859857482e-05,
+      "loss": 0.4552,
+      "step": 6846
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.530285035629454e-05,
+      "loss": 0.5729,
+      "step": 6847
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.529691211401425e-05,
+      "loss": 0.3653,
+      "step": 6848
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.529097387173397e-05,
+      "loss": 0.2365,
+      "step": 6849
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.528503562945369e-05,
+      "loss": 0.2555,
+      "step": 6850
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.527909738717341e-05,
+      "loss": 0.2895,
+      "step": 6851
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.527315914489312e-05,
+      "loss": 0.2392,
+      "step": 6852
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.526722090261282e-05,
+      "loss": 0.2819,
+      "step": 6853
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.526128266033254e-05,
+      "loss": 0.4945,
+      "step": 6854
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.525534441805225e-05,
+      "loss": 0.2677,
+      "step": 6855
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.524940617577198e-05,
+      "loss": 0.3409,
+      "step": 6856
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.524346793349169e-05,
+      "loss": 0.2695,
+      "step": 6857
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.523752969121141e-05,
+      "loss": 0.3492,
+      "step": 6858
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.523159144893112e-05,
+      "loss": 0.3277,
+      "step": 6859
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.522565320665084e-05,
+      "loss": 0.3231,
+      "step": 6860
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.521971496437054e-05,
+      "loss": 0.3325,
+      "step": 6861
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.521377672209026e-05,
+      "loss": 0.2639,
+      "step": 6862
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.520783847980999e-05,
+      "loss": 0.2119,
+      "step": 6863
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.520190023752969e-05,
+      "loss": 0.384,
+      "step": 6864
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.519596199524941e-05,
+      "loss": 0.9148,
+      "step": 6865
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.519002375296912e-05,
+      "loss": 0.3356,
+      "step": 6866
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.518408551068884e-05,
+      "loss": 0.2432,
+      "step": 6867
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.517814726840855e-05,
+      "loss": 0.4155,
+      "step": 6868
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.517220902612827e-05,
+      "loss": 0.2726,
+      "step": 6869
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.516627078384799e-05,
+      "loss": 0.3406,
+      "step": 6870
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.516033254156771e-05,
+      "loss": 0.2097,
+      "step": 6871
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.515439429928741e-05,
+      "loss": 0.2077,
+      "step": 6872
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.514845605700713e-05,
+      "loss": 0.2562,
+      "step": 6873
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.514251781472684e-05,
+      "loss": 0.2996,
+      "step": 6874
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.513657957244655e-05,
+      "loss": 0.2925,
+      "step": 6875
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.513064133016628e-05,
+      "loss": 0.3652,
+      "step": 6876
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.512470308788599e-05,
+      "loss": 0.1955,
+      "step": 6877
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.511876484560571e-05,
+      "loss": 0.2129,
+      "step": 6878
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.511282660332542e-05,
+      "loss": 0.305,
+      "step": 6879
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.510688836104514e-05,
+      "loss": 0.1824,
+      "step": 6880
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.510095011876484e-05,
+      "loss": 0.2294,
+      "step": 6881
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.509501187648456e-05,
+      "loss": 0.1544,
+      "step": 6882
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.508907363420428e-05,
+      "loss": 0.1797,
+      "step": 6883
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.508313539192399e-05,
+      "loss": 0.1904,
+      "step": 6884
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.507719714964371e-05,
+      "loss": 0.2278,
+      "step": 6885
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.507125890736342e-05,
+      "loss": 0.1744,
+      "step": 6886
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.506532066508314e-05,
+      "loss": 0.1105,
+      "step": 6887
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.505938242280284e-05,
+      "loss": 0.1604,
+      "step": 6888
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.505344418052257e-05,
+      "loss": 0.1701,
+      "step": 6889
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.504750593824229e-05,
+      "loss": 0.1544,
+      "step": 6890
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.5041567695962e-05,
+      "loss": 0.3564,
+      "step": 6891
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.503562945368171e-05,
+      "loss": 0.2933,
+      "step": 6892
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.502969121140143e-05,
+      "loss": 0.2885,
+      "step": 6893
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.502375296912114e-05,
+      "loss": 0.3087,
+      "step": 6894
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.501781472684085e-05,
+      "loss": 0.4962,
+      "step": 6895
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.501187648456058e-05,
+      "loss": 0.4195,
+      "step": 6896
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.500593824228029e-05,
+      "loss": 0.4957,
+      "step": 6897
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 0.3871,
+      "step": 6898
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.499406175771971e-05,
+      "loss": 0.4293,
+      "step": 6899
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.498812351543944e-05,
+      "loss": 0.2629,
+      "step": 6900
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.498218527315914e-05,
+      "loss": 0.2858,
+      "step": 6901
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.497624703087886e-05,
+      "loss": 0.2309,
+      "step": 6902
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.497030878859858e-05,
+      "loss": 0.3723,
+      "step": 6903
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.49643705463183e-05,
+      "loss": 0.2743,
+      "step": 6904
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.495843230403801e-05,
+      "loss": 0.2967,
+      "step": 6905
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.495249406175772e-05,
+      "loss": 0.3049,
+      "step": 6906
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.494655581947744e-05,
+      "loss": 0.2762,
+      "step": 6907
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.494061757719714e-05,
+      "loss": 0.2646,
+      "step": 6908
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.493467933491686e-05,
+      "loss": 0.2932,
+      "step": 6909
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.492874109263658e-05,
+      "loss": 0.2442,
+      "step": 6910
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.49228028503563e-05,
+      "loss": 0.2106,
+      "step": 6911
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.491686460807601e-05,
+      "loss": 0.2313,
+      "step": 6912
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.491092636579573e-05,
+      "loss": 0.2439,
+      "step": 6913
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.490498812351544e-05,
+      "loss": 0.2375,
+      "step": 6914
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.489904988123515e-05,
+      "loss": 0.2778,
+      "step": 6915
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.489311163895488e-05,
+      "loss": 0.3126,
+      "step": 6916
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.488717339667459e-05,
+      "loss": 0.262,
+      "step": 6917
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.48812351543943e-05,
+      "loss": 0.2012,
+      "step": 6918
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.487529691211401e-05,
+      "loss": 0.3075,
+      "step": 6919
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.486935866983373e-05,
+      "loss": 0.2416,
+      "step": 6920
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.486342042755344e-05,
+      "loss": 0.3294,
+      "step": 6921
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.485748218527316e-05,
+      "loss": 0.2028,
+      "step": 6922
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.485154394299288e-05,
+      "loss": 0.2451,
+      "step": 6923
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.48456057007126e-05,
+      "loss": 0.3226,
+      "step": 6924
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.483966745843231e-05,
+      "loss": 0.1888,
+      "step": 6925
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.483372921615202e-05,
+      "loss": 0.2792,
+      "step": 6926
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.482779097387174e-05,
+      "loss": 0.2159,
+      "step": 6927
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.482185273159144e-05,
+      "loss": 0.3087,
+      "step": 6928
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.481591448931116e-05,
+      "loss": 0.1422,
+      "step": 6929
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.480997624703088e-05,
+      "loss": 0.2175,
+      "step": 6930
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.48040380047506e-05,
+      "loss": 0.1923,
+      "step": 6931
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.479809976247031e-05,
+      "loss": 0.1634,
+      "step": 6932
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.479216152019003e-05,
+      "loss": 0.2083,
+      "step": 6933
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.478622327790974e-05,
+      "loss": 0.2653,
+      "step": 6934
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.478028503562946e-05,
+      "loss": 0.167,
+      "step": 6935
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.477434679334918e-05,
+      "loss": 0.4543,
+      "step": 6936
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.476840855106888e-05,
+      "loss": 0.1385,
+      "step": 6937
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.47624703087886e-05,
+      "loss": 0.5054,
+      "step": 6938
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.475653206650831e-05,
+      "loss": 0.199,
+      "step": 6939
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.475059382422803e-05,
+      "loss": 0.1226,
+      "step": 6940
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.474465558194774e-05,
+      "loss": 0.155,
+      "step": 6941
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.473871733966746e-05,
+      "loss": 0.1638,
+      "step": 6942
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.473277909738718e-05,
+      "loss": 0.1762,
+      "step": 6943
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.47268408551069e-05,
+      "loss": 0.227,
+      "step": 6944
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.472090261282661e-05,
+      "loss": 0.562,
+      "step": 6945
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.471496437054631e-05,
+      "loss": 0.453,
+      "step": 6946
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.470902612826603e-05,
+      "loss": 0.3429,
+      "step": 6947
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.470308788598574e-05,
+      "loss": 0.4958,
+      "step": 6948
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.469714964370546e-05,
+      "loss": 0.3856,
+      "step": 6949
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.469121140142518e-05,
+      "loss": 0.3521,
+      "step": 6950
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.46852731591449e-05,
+      "loss": 0.2905,
+      "step": 6951
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.467933491686461e-05,
+      "loss": 0.3569,
+      "step": 6952
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.467339667458433e-05,
+      "loss": 0.2534,
+      "step": 6953
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.466745843230404e-05,
+      "loss": 0.3288,
+      "step": 6954
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.466152019002376e-05,
+      "loss": 0.265,
+      "step": 6955
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.465558194774348e-05,
+      "loss": 0.388,
+      "step": 6956
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.464964370546318e-05,
+      "loss": 0.2851,
+      "step": 6957
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.46437054631829e-05,
+      "loss": 0.2974,
+      "step": 6958
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.463776722090261e-05,
+      "loss": 0.2498,
+      "step": 6959
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.463182897862233e-05,
+      "loss": 0.2351,
+      "step": 6960
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.462589073634204e-05,
+      "loss": 0.2227,
+      "step": 6961
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.461995249406176e-05,
+      "loss": 0.3182,
+      "step": 6962
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.461401425178148e-05,
+      "loss": 0.2031,
+      "step": 6963
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.46080760095012e-05,
+      "loss": 0.2089,
+      "step": 6964
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.46021377672209e-05,
+      "loss": 0.2941,
+      "step": 6965
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.459619952494063e-05,
+      "loss": 0.1613,
+      "step": 6966
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.459026128266033e-05,
+      "loss": 0.2865,
+      "step": 6967
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.458432304038004e-05,
+      "loss": 0.2677,
+      "step": 6968
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.457838479809976e-05,
+      "loss": 0.2364,
+      "step": 6969
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.457244655581948e-05,
+      "loss": 0.197,
+      "step": 6970
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.45665083135392e-05,
+      "loss": 0.3285,
+      "step": 6971
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.456057007125891e-05,
+      "loss": 0.1919,
+      "step": 6972
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.455463182897863e-05,
+      "loss": 0.2142,
+      "step": 6973
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.454869358669833e-05,
+      "loss": 0.2496,
+      "step": 6974
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.454275534441806e-05,
+      "loss": 0.2133,
+      "step": 6975
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.453681710213778e-05,
+      "loss": 0.1877,
+      "step": 6976
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.453087885985748e-05,
+      "loss": 0.2564,
+      "step": 6977
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.45249406175772e-05,
+      "loss": 0.3437,
+      "step": 6978
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.451900237529691e-05,
+      "loss": 0.1799,
+      "step": 6979
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.451306413301663e-05,
+      "loss": 0.165,
+      "step": 6980
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.450712589073634e-05,
+      "loss": 0.2524,
+      "step": 6981
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.450118764845606e-05,
+      "loss": 0.1691,
+      "step": 6982
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.449524940617578e-05,
+      "loss": 0.1874,
+      "step": 6983
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.44893111638955e-05,
+      "loss": 0.1958,
+      "step": 6984
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.44833729216152e-05,
+      "loss": 0.2694,
+      "step": 6985
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.447743467933492e-05,
+      "loss": 0.2151,
+      "step": 6986
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.447149643705463e-05,
+      "loss": 0.1557,
+      "step": 6987
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.446555819477434e-05,
+      "loss": 0.1752,
+      "step": 6988
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.445961995249407e-05,
+      "loss": 0.185,
+      "step": 6989
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.445368171021378e-05,
+      "loss": 0.1518,
+      "step": 6990
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.44477434679335e-05,
+      "loss": 0.2071,
+      "step": 6991
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.44418052256532e-05,
+      "loss": 0.3569,
+      "step": 6992
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.443586698337293e-05,
+      "loss": 0.1098,
+      "step": 6993
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.442992874109263e-05,
+      "loss": 0.2001,
+      "step": 6994
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.442399049881235e-05,
+      "loss": 0.7128,
+      "step": 6995
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.441805225653207e-05,
+      "loss": 0.4087,
+      "step": 6996
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.44121140142518e-05,
+      "loss": 0.6621,
+      "step": 6997
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.44061757719715e-05,
+      "loss": 0.4543,
+      "step": 6998
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.440023752969121e-05,
+      "loss": 0.273,
+      "step": 6999
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.439429928741093e-05,
+      "loss": 0.2468,
+      "step": 7000
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.438836104513064e-05,
+      "loss": 0.2763,
+      "step": 7001
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.438242280285036e-05,
+      "loss": 0.2158,
+      "step": 7002
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.437648456057008e-05,
+      "loss": 0.3065,
+      "step": 7003
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.43705463182898e-05,
+      "loss": 0.2795,
+      "step": 7004
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.43646080760095e-05,
+      "loss": 0.277,
+      "step": 7005
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.435866983372922e-05,
+      "loss": 0.2956,
+      "step": 7006
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.435273159144893e-05,
+      "loss": 0.2868,
+      "step": 7007
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.434679334916864e-05,
+      "loss": 0.2959,
+      "step": 7008
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.434085510688837e-05,
+      "loss": 0.2452,
+      "step": 7009
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.433491686460808e-05,
+      "loss": 0.287,
+      "step": 7010
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.43289786223278e-05,
+      "loss": 0.3401,
+      "step": 7011
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.43230403800475e-05,
+      "loss": 0.2682,
+      "step": 7012
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.431710213776723e-05,
+      "loss": 0.2654,
+      "step": 7013
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.431116389548693e-05,
+      "loss": 0.2152,
+      "step": 7014
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.430522565320665e-05,
+      "loss": 0.2797,
+      "step": 7015
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.429928741092637e-05,
+      "loss": 0.3457,
+      "step": 7016
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.42933491686461e-05,
+      "loss": 0.18,
+      "step": 7017
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.42874109263658e-05,
+      "loss": 0.2561,
+      "step": 7018
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.428147268408551e-05,
+      "loss": 0.2243,
+      "step": 7019
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.427553444180523e-05,
+      "loss": 0.3161,
+      "step": 7020
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.426959619952493e-05,
+      "loss": 0.2256,
+      "step": 7021
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.426365795724465e-05,
+      "loss": 0.1877,
+      "step": 7022
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.425771971496437e-05,
+      "loss": 0.2366,
+      "step": 7023
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.42517814726841e-05,
+      "loss": 0.2227,
+      "step": 7024
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.42458432304038e-05,
+      "loss": 0.2453,
+      "step": 7025
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.423990498812352e-05,
+      "loss": 0.1995,
+      "step": 7026
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.423396674584323e-05,
+      "loss": 0.2265,
+      "step": 7027
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.422802850356295e-05,
+      "loss": 0.1889,
+      "step": 7028
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.422209026128267e-05,
+      "loss": 0.2183,
+      "step": 7029
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.421615201900238e-05,
+      "loss": 0.2216,
+      "step": 7030
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.42102137767221e-05,
+      "loss": 0.1927,
+      "step": 7031
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.42042755344418e-05,
+      "loss": 0.2402,
+      "step": 7032
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.419833729216152e-05,
+      "loss": 0.1976,
+      "step": 7033
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.419239904988123e-05,
+      "loss": 0.3533,
+      "step": 7034
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.418646080760095e-05,
+      "loss": 0.3133,
+      "step": 7035
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.418052256532067e-05,
+      "loss": 0.1844,
+      "step": 7036
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.417458432304039e-05,
+      "loss": 0.1625,
+      "step": 7037
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.41686460807601e-05,
+      "loss": 0.1512,
+      "step": 7038
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.416270783847982e-05,
+      "loss": 0.2833,
+      "step": 7039
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.415676959619953e-05,
+      "loss": 0.1797,
+      "step": 7040
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.415083135391923e-05,
+      "loss": 0.1463,
+      "step": 7041
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.414489311163895e-05,
+      "loss": 0.3094,
+      "step": 7042
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.413895486935867e-05,
+      "loss": 0.1363,
+      "step": 7043
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.41330166270784e-05,
+      "loss": 0.2501,
+      "step": 7044
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.41270783847981e-05,
+      "loss": 0.6248,
+      "step": 7045
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.412114014251782e-05,
+      "loss": 0.5364,
+      "step": 7046
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.411520190023753e-05,
+      "loss": 0.3552,
+      "step": 7047
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.410926365795725e-05,
+      "loss": 0.4607,
+      "step": 7048
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.410332541567697e-05,
+      "loss": 0.2749,
+      "step": 7049
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.409738717339668e-05,
+      "loss": 0.3093,
+      "step": 7050
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.40914489311164e-05,
+      "loss": 0.2621,
+      "step": 7051
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.40855106888361e-05,
+      "loss": 0.3556,
+      "step": 7052
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.407957244655582e-05,
+      "loss": 0.3708,
+      "step": 7053
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.407363420427553e-05,
+      "loss": 0.2379,
+      "step": 7054
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.406769596199525e-05,
+      "loss": 0.2931,
+      "step": 7055
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.406175771971497e-05,
+      "loss": 0.2621,
+      "step": 7056
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.405581947743469e-05,
+      "loss": 0.242,
+      "step": 7057
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.40498812351544e-05,
+      "loss": 0.2744,
+      "step": 7058
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.404394299287412e-05,
+      "loss": 0.2512,
+      "step": 7059
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.403800475059382e-05,
+      "loss": 0.2453,
+      "step": 7060
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.403206650831353e-05,
+      "loss": 0.306,
+      "step": 7061
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.402612826603325e-05,
+      "loss": 0.2256,
+      "step": 7062
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.402019002375297e-05,
+      "loss": 0.2967,
+      "step": 7063
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.401425178147269e-05,
+      "loss": 0.2102,
+      "step": 7064
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.40083135391924e-05,
+      "loss": 0.2621,
+      "step": 7065
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.400237529691212e-05,
+      "loss": 0.2943,
+      "step": 7066
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.399643705463183e-05,
+      "loss": 0.2593,
+      "step": 7067
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.399049881235155e-05,
+      "loss": 0.5508,
+      "step": 7068
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.398456057007127e-05,
+      "loss": 0.2642,
+      "step": 7069
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.397862232779099e-05,
+      "loss": 0.3245,
+      "step": 7070
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.39726840855107e-05,
+      "loss": 0.1861,
+      "step": 7071
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.39667458432304e-05,
+      "loss": 0.2352,
+      "step": 7072
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.396080760095012e-05,
+      "loss": 0.1987,
+      "step": 7073
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.395486935866983e-05,
+      "loss": 0.2391,
+      "step": 7074
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.394893111638955e-05,
+      "loss": 0.2516,
+      "step": 7075
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.394299287410927e-05,
+      "loss": 0.2025,
+      "step": 7076
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.393705463182899e-05,
+      "loss": 0.2674,
+      "step": 7077
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.39311163895487e-05,
+      "loss": 0.2578,
+      "step": 7078
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.392517814726842e-05,
+      "loss": 0.3205,
+      "step": 7079
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.391923990498812e-05,
+      "loss": 0.2006,
+      "step": 7080
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.391330166270783e-05,
+      "loss": 0.2224,
+      "step": 7081
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.390736342042755e-05,
+      "loss": 0.1279,
+      "step": 7082
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.390142517814727e-05,
+      "loss": 0.1689,
+      "step": 7083
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.389548693586699e-05,
+      "loss": 0.2118,
+      "step": 7084
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.38895486935867e-05,
+      "loss": 0.1554,
+      "step": 7085
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.388361045130642e-05,
+      "loss": 0.1748,
+      "step": 7086
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.387767220902613e-05,
+      "loss": 0.1633,
+      "step": 7087
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.387173396674585e-05,
+      "loss": 0.1496,
+      "step": 7088
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.386579572446557e-05,
+      "loss": 0.1579,
+      "step": 7089
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.385985748218529e-05,
+      "loss": 0.1794,
+      "step": 7090
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.385391923990499e-05,
+      "loss": 0.3402,
+      "step": 7091
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.38479809976247e-05,
+      "loss": 0.1263,
+      "step": 7092
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.384204275534442e-05,
+      "loss": 0.1405,
+      "step": 7093
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.383610451306413e-05,
+      "loss": 0.404,
+      "step": 7094
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.383016627078385e-05,
+      "loss": 0.4422,
+      "step": 7095
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.382422802850357e-05,
+      "loss": 0.5217,
+      "step": 7096
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.381828978622329e-05,
+      "loss": 0.3597,
+      "step": 7097
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.3812351543943e-05,
+      "loss": 0.387,
+      "step": 7098
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.380641330166272e-05,
+      "loss": 0.3613,
+      "step": 7099
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.380047505938242e-05,
+      "loss": 0.3064,
+      "step": 7100
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.379453681710214e-05,
+      "loss": 0.2585,
+      "step": 7101
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.378859857482186e-05,
+      "loss": 0.2989,
+      "step": 7102
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.378266033254157e-05,
+      "loss": 0.2953,
+      "step": 7103
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.377672209026129e-05,
+      "loss": 0.2406,
+      "step": 7104
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.3770783847981e-05,
+      "loss": 0.331,
+      "step": 7105
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.376484560570072e-05,
+      "loss": 0.2572,
+      "step": 7106
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.375890736342042e-05,
+      "loss": 0.2232,
+      "step": 7107
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.375296912114014e-05,
+      "loss": 0.3123,
+      "step": 7108
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.374703087885986e-05,
+      "loss": 0.3088,
+      "step": 7109
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.374109263657959e-05,
+      "loss": 0.2256,
+      "step": 7110
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.373515439429929e-05,
+      "loss": 0.1695,
+      "step": 7111
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.3729216152019e-05,
+      "loss": 0.2472,
+      "step": 7112
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.372327790973872e-05,
+      "loss": 0.335,
+      "step": 7113
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.371733966745843e-05,
+      "loss": 0.1975,
+      "step": 7114
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.371140142517815e-05,
+      "loss": 0.2788,
+      "step": 7115
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.370546318289787e-05,
+      "loss": 0.2427,
+      "step": 7116
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.369952494061759e-05,
+      "loss": 0.2662,
+      "step": 7117
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.36935866983373e-05,
+      "loss": 0.1824,
+      "step": 7118
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.368764845605701e-05,
+      "loss": 0.2175,
+      "step": 7119
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.368171021377672e-05,
+      "loss": 0.2295,
+      "step": 7120
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.367577197149644e-05,
+      "loss": 0.264,
+      "step": 7121
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.366983372921616e-05,
+      "loss": 0.2156,
+      "step": 7122
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.366389548693587e-05,
+      "loss": 0.283,
+      "step": 7123
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.365795724465559e-05,
+      "loss": 0.1597,
+      "step": 7124
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.36520190023753e-05,
+      "loss": 0.181,
+      "step": 7125
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.364608076009502e-05,
+      "loss": 0.3287,
+      "step": 7126
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.364014251781472e-05,
+      "loss": 0.1633,
+      "step": 7127
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.363420427553444e-05,
+      "loss": 0.2525,
+      "step": 7128
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.362826603325416e-05,
+      "loss": 0.2542,
+      "step": 7129
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.362232779097388e-05,
+      "loss": 0.2163,
+      "step": 7130
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.361638954869359e-05,
+      "loss": 0.1465,
+      "step": 7131
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.361045130641331e-05,
+      "loss": 0.1996,
+      "step": 7132
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.360451306413302e-05,
+      "loss": 0.2151,
+      "step": 7133
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.359857482185272e-05,
+      "loss": 0.156,
+      "step": 7134
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.359263657957244e-05,
+      "loss": 0.4016,
+      "step": 7135
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.358669833729217e-05,
+      "loss": 0.2108,
+      "step": 7136
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.358076009501189e-05,
+      "loss": 0.3666,
+      "step": 7137
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.357482185273159e-05,
+      "loss": 0.4247,
+      "step": 7138
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.356888361045131e-05,
+      "loss": 0.3356,
+      "step": 7139
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.356294536817102e-05,
+      "loss": 0.234,
+      "step": 7140
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.355700712589074e-05,
+      "loss": 0.2075,
+      "step": 7141
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.355106888361046e-05,
+      "loss": 0.2539,
+      "step": 7142
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.354513064133017e-05,
+      "loss": 0.184,
+      "step": 7143
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.353919239904989e-05,
+      "loss": 0.2024,
+      "step": 7144
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.35332541567696e-05,
+      "loss": 0.1496,
+      "step": 7145
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.352731591448931e-05,
+      "loss": 0.2612,
+      "step": 7146
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.352137767220902e-05,
+      "loss": 0.1448,
+      "step": 7147
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.351543942992874e-05,
+      "loss": 0.2057,
+      "step": 7148
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.350950118764846e-05,
+      "loss": 0.1829,
+      "step": 7149
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.350356294536818e-05,
+      "loss": 0.2397,
+      "step": 7150
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.349762470308789e-05,
+      "loss": 0.1954,
+      "step": 7151
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.349168646080761e-05,
+      "loss": 0.1557,
+      "step": 7152
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.348574821852732e-05,
+      "loss": 0.2548,
+      "step": 7153
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.347980997624702e-05,
+      "loss": 0.1691,
+      "step": 7154
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.347387173396674e-05,
+      "loss": 0.1773,
+      "step": 7155
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.346793349168646e-05,
+      "loss": 0.1071,
+      "step": 7156
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.346199524940618e-05,
+      "loss": 0.1417,
+      "step": 7157
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.345605700712589e-05,
+      "loss": 0.1559,
+      "step": 7158
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.345011876484561e-05,
+      "loss": 0.1358,
+      "step": 7159
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.344418052256532e-05,
+      "loss": 0.2185,
+      "step": 7160
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.343824228028504e-05,
+      "loss": 0.2042,
+      "step": 7161
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.343230403800476e-05,
+      "loss": 0.1939,
+      "step": 7162
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.342636579572448e-05,
+      "loss": 0.2069,
+      "step": 7163
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.342042755344419e-05,
+      "loss": 0.1555,
+      "step": 7164
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.341448931116389e-05,
+      "loss": 0.2344,
+      "step": 7165
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.340855106888361e-05,
+      "loss": 0.1478,
+      "step": 7166
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.340261282660332e-05,
+      "loss": 0.1459,
+      "step": 7167
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.339667458432304e-05,
+      "loss": 0.1196,
+      "step": 7168
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.339073634204276e-05,
+      "loss": 0.1733,
+      "step": 7169
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.338479809976248e-05,
+      "loss": 0.2476,
+      "step": 7170
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.337885985748219e-05,
+      "loss": 0.1702,
+      "step": 7171
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.337292161520191e-05,
+      "loss": 0.1401,
+      "step": 7172
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.336698337292162e-05,
+      "loss": 0.1281,
+      "step": 7173
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.336104513064132e-05,
+      "loss": 0.1456,
+      "step": 7174
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.335510688836104e-05,
+      "loss": 0.1413,
+      "step": 7175
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.334916864608076e-05,
+      "loss": 0.1173,
+      "step": 7176
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.334323040380048e-05,
+      "loss": 0.2181,
+      "step": 7177
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.333729216152019e-05,
+      "loss": 0.1759,
+      "step": 7178
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.333135391923991e-05,
+      "loss": 0.1434,
+      "step": 7179
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.332541567695962e-05,
+      "loss": 0.0908,
+      "step": 7180
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.331947743467934e-05,
+      "loss": 0.1323,
+      "step": 7181
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.331353919239906e-05,
+      "loss": 0.159,
+      "step": 7182
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.330760095011878e-05,
+      "loss": 0.1319,
+      "step": 7183
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.330166270783848e-05,
+      "loss": 0.2102,
+      "step": 7184
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.329572446555819e-05,
+      "loss": 0.1435,
+      "step": 7185
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.328978622327791e-05,
+      "loss": 0.1047,
+      "step": 7186
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.328384798099762e-05,
+      "loss": 0.4981,
+      "step": 7187
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.327790973871734e-05,
+      "loss": 0.4777,
+      "step": 7188
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.327197149643706e-05,
+      "loss": 0.3577,
+      "step": 7189
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.326603325415678e-05,
+      "loss": 0.2349,
+      "step": 7190
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.326009501187649e-05,
+      "loss": 0.2644,
+      "step": 7191
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.325415676959621e-05,
+      "loss": 0.2404,
+      "step": 7192
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.324821852731591e-05,
+      "loss": 0.222,
+      "step": 7193
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.324228028503563e-05,
+      "loss": 0.223,
+      "step": 7194
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.323634204275534e-05,
+      "loss": 0.2336,
+      "step": 7195
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.323040380047506e-05,
+      "loss": 0.2937,
+      "step": 7196
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.322446555819478e-05,
+      "loss": 0.2014,
+      "step": 7197
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.321852731591449e-05,
+      "loss": 0.2478,
+      "step": 7198
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.321258907363421e-05,
+      "loss": 0.3155,
+      "step": 7199
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.320665083135392e-05,
+      "loss": 0.2111,
+      "step": 7200
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.320071258907364e-05,
+      "loss": 0.2206,
+      "step": 7201
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.319477434679336e-05,
+      "loss": 0.2444,
+      "step": 7202
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.318883610451308e-05,
+      "loss": 0.1841,
+      "step": 7203
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.318289786223278e-05,
+      "loss": 0.1712,
+      "step": 7204
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.317695961995249e-05,
+      "loss": 0.1506,
+      "step": 7205
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.317102137767221e-05,
+      "loss": 0.2364,
+      "step": 7206
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.316508313539192e-05,
+      "loss": 0.13,
+      "step": 7207
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.315914489311164e-05,
+      "loss": 0.1711,
+      "step": 7208
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.315320665083136e-05,
+      "loss": 0.1915,
+      "step": 7209
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.314726840855108e-05,
+      "loss": 0.2251,
+      "step": 7210
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.314133016627079e-05,
+      "loss": 0.1622,
+      "step": 7211
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.31353919239905e-05,
+      "loss": 0.1482,
+      "step": 7212
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.312945368171021e-05,
+      "loss": 0.0981,
+      "step": 7213
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.312351543942993e-05,
+      "loss": 0.1871,
+      "step": 7214
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.311757719714964e-05,
+      "loss": 0.2198,
+      "step": 7215
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.311163895486936e-05,
+      "loss": 0.1933,
+      "step": 7216
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.310570071258908e-05,
+      "loss": 0.1506,
+      "step": 7217
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.309976247030879e-05,
+      "loss": 0.1334,
+      "step": 7218
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.309382422802851e-05,
+      "loss": 0.1338,
+      "step": 7219
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.308788598574821e-05,
+      "loss": 0.1696,
+      "step": 7220
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.308194774346793e-05,
+      "loss": 0.1576,
+      "step": 7221
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.307600950118766e-05,
+      "loss": 0.142,
+      "step": 7222
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.307007125890738e-05,
+      "loss": 0.1715,
+      "step": 7223
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.306413301662708e-05,
+      "loss": 0.1244,
+      "step": 7224
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.30581947743468e-05,
+      "loss": 0.1497,
+      "step": 7225
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.305225653206651e-05,
+      "loss": 0.1679,
+      "step": 7226
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.304631828978622e-05,
+      "loss": 0.1804,
+      "step": 7227
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.304038004750594e-05,
+      "loss": 0.0965,
+      "step": 7228
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.303444180522566e-05,
+      "loss": 0.2552,
+      "step": 7229
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.302850356294538e-05,
+      "loss": 0.1756,
+      "step": 7230
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.302256532066508e-05,
+      "loss": 0.2651,
+      "step": 7231
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.30166270783848e-05,
+      "loss": 0.1422,
+      "step": 7232
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.301068883610451e-05,
+      "loss": 0.0675,
+      "step": 7233
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.300475059382423e-05,
+      "loss": 0.0767,
+      "step": 7234
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.299881235154395e-05,
+      "loss": 0.1236,
+      "step": 7235
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.299287410926367e-05,
+      "loss": 0.1124,
+      "step": 7236
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.298693586698338e-05,
+      "loss": 0.4993,
+      "step": 7237
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.298099762470309e-05,
+      "loss": 0.3795,
+      "step": 7238
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.29750593824228e-05,
+      "loss": 0.3288,
+      "step": 7239
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.296912114014251e-05,
+      "loss": 0.274,
+      "step": 7240
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.296318289786223e-05,
+      "loss": 0.2877,
+      "step": 7241
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.295724465558195e-05,
+      "loss": 0.2838,
+      "step": 7242
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.295130641330167e-05,
+      "loss": 0.2147,
+      "step": 7243
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.294536817102138e-05,
+      "loss": 0.1752,
+      "step": 7244
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.29394299287411e-05,
+      "loss": 0.2324,
+      "step": 7245
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.293349168646081e-05,
+      "loss": 0.162,
+      "step": 7246
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.292755344418052e-05,
+      "loss": 0.1642,
+      "step": 7247
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.292161520190024e-05,
+      "loss": 0.1824,
+      "step": 7248
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.291567695961996e-05,
+      "loss": 0.2508,
+      "step": 7249
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.290973871733968e-05,
+      "loss": 0.2232,
+      "step": 7250
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.290380047505938e-05,
+      "loss": 0.2026,
+      "step": 7251
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.28978622327791e-05,
+      "loss": 0.223,
+      "step": 7252
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.289192399049881e-05,
+      "loss": 0.2183,
+      "step": 7253
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.288598574821853e-05,
+      "loss": 0.1712,
+      "step": 7254
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.288004750593825e-05,
+      "loss": 0.2037,
+      "step": 7255
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.287410926365797e-05,
+      "loss": 0.1005,
+      "step": 7256
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.286817102137768e-05,
+      "loss": 0.2,
+      "step": 7257
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.286223277909738e-05,
+      "loss": 0.2358,
+      "step": 7258
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.28562945368171e-05,
+      "loss": 0.1772,
+      "step": 7259
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.285035629453681e-05,
+      "loss": 0.1912,
+      "step": 7260
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.284441805225653e-05,
+      "loss": 0.2726,
+      "step": 7261
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.283847980997625e-05,
+      "loss": 0.1501,
+      "step": 7262
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.283254156769597e-05,
+      "loss": 0.15,
+      "step": 7263
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.282660332541568e-05,
+      "loss": 0.1398,
+      "step": 7264
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.28206650831354e-05,
+      "loss": 0.1357,
+      "step": 7265
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.281472684085511e-05,
+      "loss": 0.1229,
+      "step": 7266
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.280878859857483e-05,
+      "loss": 0.139,
+      "step": 7267
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.280285035629453e-05,
+      "loss": 0.1989,
+      "step": 7268
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.279691211401425e-05,
+      "loss": 0.1772,
+      "step": 7269
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.279097387173397e-05,
+      "loss": 0.1677,
+      "step": 7270
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.278503562945368e-05,
+      "loss": 0.1298,
+      "step": 7271
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.27790973871734e-05,
+      "loss": 0.1433,
+      "step": 7272
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.277315914489311e-05,
+      "loss": 0.1709,
+      "step": 7273
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.276722090261283e-05,
+      "loss": 0.4118,
+      "step": 7274
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.276128266033255e-05,
+      "loss": 0.2104,
+      "step": 7275
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.275534441805227e-05,
+      "loss": 0.142,
+      "step": 7276
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.274940617577198e-05,
+      "loss": 0.1387,
+      "step": 7277
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.274346793349168e-05,
+      "loss": 0.1218,
+      "step": 7278
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.27375296912114e-05,
+      "loss": 0.1517,
+      "step": 7279
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.273159144893111e-05,
+      "loss": 0.1505,
+      "step": 7280
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.272565320665083e-05,
+      "loss": 0.0874,
+      "step": 7281
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.271971496437055e-05,
+      "loss": 0.0868,
+      "step": 7282
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.271377672209027e-05,
+      "loss": 0.1176,
+      "step": 7283
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 6.270783847980998e-05,
+      "loss": 0.0834,
+      "step": 7284
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 6.27019002375297e-05,
+      "loss": 0.1099,
+      "step": 7285
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 6.26959619952494e-05,
+      "loss": 0.1801,
+      "step": 7286
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 6.269002375296913e-05,
+      "loss": 0.4313,
+      "step": 7287
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 6.268408551068883e-05,
+      "loss": 0.4951,
+      "step": 7288
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 6.267814726840855e-05,
+      "loss": 0.3831,
+      "step": 7289
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 6.267220902612827e-05,
+      "loss": 0.3072,
+      "step": 7290
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 6.266627078384798e-05,
+      "loss": 0.2419,
+      "step": 7291
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 6.26603325415677e-05,
+      "loss": 0.1661,
+      "step": 7292
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 6.265439429928741e-05,
+      "loss": 0.2701,
+      "step": 7293
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 6.264845605700713e-05,
+      "loss": 0.1804,
+      "step": 7294
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 6.264251781472685e-05,
+      "loss": 0.296,
+      "step": 7295
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 6.263657957244657e-05,
+      "loss": 0.2115,
+      "step": 7296
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 6.263064133016628e-05,
+      "loss": 0.2365,
+      "step": 7297
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 6.2624703087886e-05,
+      "loss": 0.231,
+      "step": 7298
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 6.26187648456057e-05,
+      "loss": 0.2627,
+      "step": 7299
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 6.261282660332541e-05,
+      "loss": 0.2193,
+      "step": 7300
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 6.260688836104513e-05,
+      "loss": 0.1937,
+      "step": 7301
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 6.260095011876485e-05,
+      "loss": 0.2279,
+      "step": 7302
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 6.259501187648457e-05,
+      "loss": 0.2251,
+      "step": 7303
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 6.258907363420428e-05,
+      "loss": 0.3033,
+      "step": 7304
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 6.2583135391924e-05,
+      "loss": 0.2062,
+      "step": 7305
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 6.25771971496437e-05,
+      "loss": 0.1799,
+      "step": 7306
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 6.257125890736342e-05,
+      "loss": 0.1864,
+      "step": 7307
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 6.256532066508313e-05,
+      "loss": 0.1317,
+      "step": 7308
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 6.255938242280285e-05,
+      "loss": 0.1494,
+      "step": 7309
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 6.255344418052257e-05,
+      "loss": 0.123,
+      "step": 7310
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 6.254750593824228e-05,
+      "loss": 0.17,
+      "step": 7311
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 6.2541567695962e-05,
+      "loss": 0.144,
+      "step": 7312
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 6.25356294536817e-05,
+      "loss": 0.1572,
+      "step": 7313
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 6.252969121140143e-05,
+      "loss": 0.1547,
+      "step": 7314
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 6.252375296912115e-05,
+      "loss": 0.1267,
+      "step": 7315
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 6.251781472684087e-05,
+      "loss": 0.2869,
+      "step": 7316
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 6.251187648456057e-05,
+      "loss": 0.1481,
+      "step": 7317
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 6.25059382422803e-05,
+      "loss": 0.2108,
+      "step": 7318
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 6.25e-05,
+      "loss": 0.155,
+      "step": 7319
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 6.249406175771971e-05,
+      "loss": 0.1987,
+      "step": 7320
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 6.248812351543943e-05,
+      "loss": 0.1466,
+      "step": 7321
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 6.248218527315915e-05,
+      "loss": 0.1442,
+      "step": 7322
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 6.247624703087887e-05,
+      "loss": 0.151,
+      "step": 7323
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 6.247030878859858e-05,
+      "loss": 0.1579,
+      "step": 7324
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 6.24643705463183e-05,
+      "loss": 0.2656,
+      "step": 7325
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 6.2458432304038e-05,
+      "loss": 0.1983,
+      "step": 7326
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 6.245249406175772e-05,
+      "loss": 0.1653,
+      "step": 7327
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 6.244655581947743e-05,
+      "loss": 0.1984,
+      "step": 7328
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 6.244061757719716e-05,
+      "loss": 0.0809,
+      "step": 7329
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 6.243467933491687e-05,
+      "loss": 0.1452,
+      "step": 7330
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 6.242874109263658e-05,
+      "loss": 0.1051,
+      "step": 7331
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 6.24228028503563e-05,
+      "loss": 0.2955,
+      "step": 7332
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 6.2416864608076e-05,
+      "loss": 0.1276,
+      "step": 7333
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 6.241092636579573e-05,
+      "loss": 0.2477,
+      "step": 7334
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 6.240498812351545e-05,
+      "loss": 0.2311,
+      "step": 7335
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 6.239904988123517e-05,
+      "loss": 0.2598,
+      "step": 7336
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 6.239311163895487e-05,
+      "loss": 0.3529,
+      "step": 7337
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 6.238717339667459e-05,
+      "loss": 0.28,
+      "step": 7338
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 6.23812351543943e-05,
+      "loss": 0.4445,
+      "step": 7339
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 6.237529691211401e-05,
+      "loss": 0.3764,
+      "step": 7340
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 6.236935866983373e-05,
+      "loss": 0.3076,
+      "step": 7341
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 6.236342042755345e-05,
+      "loss": 0.2431,
+      "step": 7342
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 6.235748218527317e-05,
+      "loss": 0.4611,
+      "step": 7343
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 6.235154394299287e-05,
+      "loss": 0.3228,
+      "step": 7344
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 6.23456057007126e-05,
+      "loss": 0.2839,
+      "step": 7345
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 6.23396674584323e-05,
+      "loss": 0.2243,
+      "step": 7346
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 6.233372921615202e-05,
+      "loss": 0.2597,
+      "step": 7347
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 6.232779097387174e-05,
+      "loss": 0.2144,
+      "step": 7348
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 6.232185273159146e-05,
+      "loss": 0.2796,
+      "step": 7349
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 6.231591448931117e-05,
+      "loss": 0.209,
+      "step": 7350
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 6.230997624703088e-05,
+      "loss": 0.1766,
+      "step": 7351
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 6.23040380047506e-05,
+      "loss": 0.2022,
+      "step": 7352
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 6.22980997624703e-05,
+      "loss": 0.2168,
+      "step": 7353
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 6.229216152019002e-05,
+      "loss": 0.2002,
+      "step": 7354
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 6.228622327790974e-05,
+      "loss": 0.1723,
+      "step": 7355
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 6.228028503562946e-05,
+      "loss": 0.1858,
+      "step": 7356
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 6.227434679334917e-05,
+      "loss": 0.1425,
+      "step": 7357
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 6.226840855106889e-05,
+      "loss": 0.2225,
+      "step": 7358
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 6.22624703087886e-05,
+      "loss": 0.1446,
+      "step": 7359
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 6.225653206650832e-05,
+      "loss": 0.1366,
+      "step": 7360
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 6.225059382422803e-05,
+      "loss": 0.1784,
+      "step": 7361
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 6.224465558194775e-05,
+      "loss": 0.1759,
+      "step": 7362
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 6.223871733966747e-05,
+      "loss": 0.142,
+      "step": 7363
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 6.223277909738717e-05,
+      "loss": 0.2029,
+      "step": 7364
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 6.22268408551069e-05,
+      "loss": 0.1439,
+      "step": 7365
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 6.22209026128266e-05,
+      "loss": 0.1669,
+      "step": 7366
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 6.221496437054632e-05,
+      "loss": 0.1764,
+      "step": 7367
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 6.220902612826604e-05,
+      "loss": 0.1523,
+      "step": 7368
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 6.220308788598576e-05,
+      "loss": 0.2531,
+      "step": 7369
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 6.219714964370547e-05,
+      "loss": 0.2189,
+      "step": 7370
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 6.219121140142518e-05,
+      "loss": 0.1601,
+      "step": 7371
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 6.21852731591449e-05,
+      "loss": 0.1718,
+      "step": 7372
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 6.21793349168646e-05,
+      "loss": 0.2271,
+      "step": 7373
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 6.217339667458432e-05,
+      "loss": 0.1114,
+      "step": 7374
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 6.216745843230404e-05,
+      "loss": 0.1244,
+      "step": 7375
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 6.216152019002376e-05,
+      "loss": 0.1473,
+      "step": 7376
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 6.215558194774347e-05,
+      "loss": 0.0937,
+      "step": 7377
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 6.214964370546319e-05,
+      "loss": 0.1825,
+      "step": 7378
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 6.21437054631829e-05,
+      "loss": 0.1811,
+      "step": 7379
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 6.213776722090262e-05,
+      "loss": 0.0844,
+      "step": 7380
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 6.213182897862232e-05,
+      "loss": 0.1179,
+      "step": 7381
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 6.212589073634204e-05,
+      "loss": 0.2269,
+      "step": 7382
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 6.211995249406177e-05,
+      "loss": 0.1637,
+      "step": 7383
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 6.211401425178147e-05,
+      "loss": 0.0905,
+      "step": 7384
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 6.210807600950119e-05,
+      "loss": 0.4134,
+      "step": 7385
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 6.21021377672209e-05,
+      "loss": 0.1219,
+      "step": 7386
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 6.209619952494062e-05,
+      "loss": 0.4265,
+      "step": 7387
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 6.209026128266034e-05,
+      "loss": 0.4617,
+      "step": 7388
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 6.208432304038006e-05,
+      "loss": 0.3723,
+      "step": 7389
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 6.207838479809977e-05,
+      "loss": 0.2747,
+      "step": 7390
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 6.207244655581949e-05,
+      "loss": 0.3141,
+      "step": 7391
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 6.20665083135392e-05,
+      "loss": 0.1932,
+      "step": 7392
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 6.20605700712589e-05,
+      "loss": 0.1444,
+      "step": 7393
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 6.205463182897862e-05,
+      "loss": 0.2867,
+      "step": 7394
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 6.204869358669834e-05,
+      "loss": 0.1962,
+      "step": 7395
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 6.204275534441806e-05,
+      "loss": 0.2059,
+      "step": 7396
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 6.203681710213777e-05,
+      "loss": 0.3036,
+      "step": 7397
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 6.203087885985749e-05,
+      "loss": 0.2438,
+      "step": 7398
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 6.20249406175772e-05,
+      "loss": 0.2184,
+      "step": 7399
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 6.201900237529692e-05,
+      "loss": 0.179,
+      "step": 7400
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 6.201306413301662e-05,
+      "loss": 0.1877,
+      "step": 7401
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 6.200712589073634e-05,
+      "loss": 0.2011,
+      "step": 7402
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 6.200118764845606e-05,
+      "loss": 0.1686,
+      "step": 7403
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 6.199524940617577e-05,
+      "loss": 0.171,
+      "step": 7404
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 6.198931116389549e-05,
+      "loss": 0.1934,
+      "step": 7405
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 6.19833729216152e-05,
+      "loss": 0.1762,
+      "step": 7406
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 6.197743467933492e-05,
+      "loss": 0.1282,
+      "step": 7407
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 6.197149643705464e-05,
+      "loss": 0.1656,
+      "step": 7408
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 6.196555819477436e-05,
+      "loss": 0.1964,
+      "step": 7409
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 6.195961995249407e-05,
+      "loss": 0.1362,
+      "step": 7410
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 6.195368171021379e-05,
+      "loss": 0.165,
+      "step": 7411
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 6.194774346793349e-05,
+      "loss": 0.2119,
+      "step": 7412
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 6.19418052256532e-05,
+      "loss": 0.2104,
+      "step": 7413
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 6.193586698337292e-05,
+      "loss": 0.1867,
+      "step": 7414
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 6.192992874109264e-05,
+      "loss": 0.1267,
+      "step": 7415
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 6.192399049881236e-05,
+      "loss": 0.1499,
+      "step": 7416
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 6.191805225653207e-05,
+      "loss": 0.1827,
+      "step": 7417
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 6.191211401425179e-05,
+      "loss": 0.2087,
+      "step": 7418
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 6.19061757719715e-05,
+      "loss": 0.1708,
+      "step": 7419
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 6.190023752969122e-05,
+      "loss": 0.1948,
+      "step": 7420
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 6.189429928741092e-05,
+      "loss": 0.1274,
+      "step": 7421
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 6.188836104513066e-05,
+      "loss": 0.128,
+      "step": 7422
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 6.188242280285036e-05,
+      "loss": 0.1148,
+      "step": 7423
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 6.187648456057007e-05,
+      "loss": 0.1368,
+      "step": 7424
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 6.187054631828979e-05,
+      "loss": 0.1067,
+      "step": 7425
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 6.18646080760095e-05,
+      "loss": 0.1955,
+      "step": 7426
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 6.185866983372922e-05,
+      "loss": 0.1425,
+      "step": 7427
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 6.185273159144894e-05,
+      "loss": 0.1763,
+      "step": 7428
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 6.184679334916866e-05,
+      "loss": 0.1358,
+      "step": 7429
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 6.184085510688836e-05,
+      "loss": 0.1204,
+      "step": 7430
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 6.183491686460808e-05,
+      "loss": 0.1562,
+      "step": 7431
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 6.182897862232779e-05,
+      "loss": 0.1195,
+      "step": 7432
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 6.182304038004751e-05,
+      "loss": 0.106,
+      "step": 7433
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 6.181710213776722e-05,
+      "loss": 0.1032,
+      "step": 7434
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 6.181116389548694e-05,
+      "loss": 0.1245,
+      "step": 7435
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 6.180522565320666e-05,
+      "loss": 0.1575,
+      "step": 7436
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 6.179928741092637e-05,
+      "loss": 0.3817,
+      "step": 7437
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 6.179334916864609e-05,
+      "loss": 0.3643,
+      "step": 7438
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 6.17874109263658e-05,
+      "loss": 0.2933,
+      "step": 7439
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 6.178147268408551e-05,
+      "loss": 0.2764,
+      "step": 7440
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 6.177553444180522e-05,
+      "loss": 0.2372,
+      "step": 7441
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 6.176959619952495e-05,
+      "loss": 0.195,
+      "step": 7442
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 6.176365795724466e-05,
+      "loss": 0.217,
+      "step": 7443
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 6.175771971496437e-05,
+      "loss": 0.1484,
+      "step": 7444
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 6.175178147268409e-05,
+      "loss": 0.2741,
+      "step": 7445
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 6.17458432304038e-05,
+      "loss": 0.1595,
+      "step": 7446
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 6.173990498812352e-05,
+      "loss": 0.2058,
+      "step": 7447
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 6.173396674584324e-05,
+      "loss": 0.1783,
+      "step": 7448
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 6.172802850356296e-05,
+      "loss": 0.1752,
+      "step": 7449
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 6.172209026128266e-05,
+      "loss": 0.2209,
+      "step": 7450
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 6.171615201900238e-05,
+      "loss": 0.1698,
+      "step": 7451
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 6.171021377672209e-05,
+      "loss": 0.1416,
+      "step": 7452
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 6.170427553444181e-05,
+      "loss": 0.2377,
+      "step": 7453
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 6.169833729216152e-05,
+      "loss": 0.1267,
+      "step": 7454
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 6.169239904988124e-05,
+      "loss": 0.1492,
+      "step": 7455
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 6.168646080760096e-05,
+      "loss": 0.1932,
+      "step": 7456
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 6.168052256532067e-05,
+      "loss": 0.1222,
+      "step": 7457
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 6.167458432304039e-05,
+      "loss": 0.2799,
+      "step": 7458
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 6.166864608076009e-05,
+      "loss": 0.1454,
+      "step": 7459
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 6.166270783847981e-05,
+      "loss": 0.106,
+      "step": 7460
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 6.165676959619952e-05,
+      "loss": 0.1462,
+      "step": 7461
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 6.165083135391925e-05,
+      "loss": 0.1354,
+      "step": 7462
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 6.164489311163896e-05,
+      "loss": 0.1762,
+      "step": 7463
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 6.163895486935868e-05,
+      "loss": 0.1653,
+      "step": 7464
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 6.163301662707839e-05,
+      "loss": 0.1831,
+      "step": 7465
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 6.16270783847981e-05,
+      "loss": 0.1706,
+      "step": 7466
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 6.162114014251781e-05,
+      "loss": 0.1432,
+      "step": 7467
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 6.161520190023753e-05,
+      "loss": 0.1747,
+      "step": 7468
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 6.160926365795726e-05,
+      "loss": 0.1818,
+      "step": 7469
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 6.160332541567696e-05,
+      "loss": 0.3195,
+      "step": 7470
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 6.159738717339668e-05,
+      "loss": 0.0991,
+      "step": 7471
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 6.159144893111639e-05,
+      "loss": 0.1988,
+      "step": 7472
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 6.158551068883611e-05,
+      "loss": 0.1365,
+      "step": 7473
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 6.157957244655582e-05,
+      "loss": 0.1134,
+      "step": 7474
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 6.157363420427554e-05,
+      "loss": 0.1588,
+      "step": 7475
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 6.156769596199526e-05,
+      "loss": 0.1297,
+      "step": 7476
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 6.156175771971496e-05,
+      "loss": 0.1842,
+      "step": 7477
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 6.155581947743468e-05,
+      "loss": 0.2159,
+      "step": 7478
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 6.154988123515439e-05,
+      "loss": 0.1389,
+      "step": 7479
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 6.154394299287411e-05,
+      "loss": 0.1118,
+      "step": 7480
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 6.153800475059383e-05,
+      "loss": 0.1429,
+      "step": 7481
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 6.153206650831355e-05,
+      "loss": 0.1086,
+      "step": 7482
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 6.152612826603326e-05,
+      "loss": 0.1203,
+      "step": 7483
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 6.152019002375298e-05,
+      "loss": 0.1671,
+      "step": 7484
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 6.151425178147269e-05,
+      "loss": 0.1943,
+      "step": 7485
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 6.150831353919239e-05,
+      "loss": 0.1813,
+      "step": 7486
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 6.150237529691211e-05,
+      "loss": 0.3742,
+      "step": 7487
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 6.149643705463183e-05,
+      "loss": 0.2956,
+      "step": 7488
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 6.149049881235155e-05,
+      "loss": 0.3198,
+      "step": 7489
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 6.148456057007126e-05,
+      "loss": 0.2542,
+      "step": 7490
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 6.147862232779098e-05,
+      "loss": 0.2702,
+      "step": 7491
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 6.147268408551069e-05,
+      "loss": 0.182,
+      "step": 7492
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 6.146674584323041e-05,
+      "loss": 0.1841,
+      "step": 7493
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 6.146080760095012e-05,
+      "loss": 0.2708,
+      "step": 7494
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 6.145486935866984e-05,
+      "loss": 0.2712,
+      "step": 7495
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 6.144893111638956e-05,
+      "loss": 0.2309,
+      "step": 7496
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 6.144299287410926e-05,
+      "loss": 0.2522,
+      "step": 7497
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 6.143705463182898e-05,
+      "loss": 0.2198,
+      "step": 7498
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 6.143111638954869e-05,
+      "loss": 0.2133,
+      "step": 7499
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 6.142517814726841e-05,
+      "loss": 0.2736,
+      "step": 7500
+    },
+    {
+      "epoch": 8.41,
+      "eval_loss": 1.6549711227416992,
+      "eval_runtime": 899.3827,
+      "eval_samples_per_second": 2.938,
+      "eval_steps_per_second": 0.368,
+      "eval_wer": 1.0372272907576359,
+      "step": 7500
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 6.141923990498813e-05,
+      "loss": 0.1623,
+      "step": 7501
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 6.141330166270785e-05,
+      "loss": 0.2359,
+      "step": 7502
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 6.140736342042756e-05,
+      "loss": 0.1576,
+      "step": 7503
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 6.140142517814728e-05,
+      "loss": 0.1907,
+      "step": 7504
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 6.139548693586698e-05,
+      "loss": 0.2804,
+      "step": 7505
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 6.138954869358669e-05,
+      "loss": 0.1539,
+      "step": 7506
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 6.138361045130641e-05,
+      "loss": 0.2507,
+      "step": 7507
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 6.137767220902613e-05,
+      "loss": 0.1904,
+      "step": 7508
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 6.137173396674585e-05,
+      "loss": 0.1938,
+      "step": 7509
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 6.136579572446556e-05,
+      "loss": 0.1878,
+      "step": 7510
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 6.135985748218528e-05,
+      "loss": 0.1869,
+      "step": 7511
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 6.135391923990499e-05,
+      "loss": 0.1692,
+      "step": 7512
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 6.134798099762471e-05,
+      "loss": 0.2059,
+      "step": 7513
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 6.134204275534441e-05,
+      "loss": 0.1311,
+      "step": 7514
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 6.133610451306415e-05,
+      "loss": 0.1857,
+      "step": 7515
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 6.133016627078385e-05,
+      "loss": 0.1639,
+      "step": 7516
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 6.132422802850356e-05,
+      "loss": 0.1262,
+      "step": 7517
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 6.131828978622328e-05,
+      "loss": 0.1654,
+      "step": 7518
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 6.131235154394299e-05,
+      "loss": 0.1514,
+      "step": 7519
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 6.130641330166271e-05,
+      "loss": 0.1139,
+      "step": 7520
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 6.130047505938243e-05,
+      "loss": 0.1775,
+      "step": 7521
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 6.129453681710215e-05,
+      "loss": 0.1783,
+      "step": 7522
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 6.128859857482186e-05,
+      "loss": 0.0895,
+      "step": 7523
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 6.128266033254158e-05,
+      "loss": 0.151,
+      "step": 7524
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 6.127672209026128e-05,
+      "loss": 0.1903,
+      "step": 7525
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 6.1270783847981e-05,
+      "loss": 0.1305,
+      "step": 7526
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 6.126484560570071e-05,
+      "loss": 0.141,
+      "step": 7527
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 6.125890736342043e-05,
+      "loss": 0.1973,
+      "step": 7528
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 6.125296912114015e-05,
+      "loss": 0.2326,
+      "step": 7529
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 6.124703087885986e-05,
+      "loss": 0.3077,
+      "step": 7530
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 6.124109263657958e-05,
+      "loss": 0.0692,
+      "step": 7531
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 6.123515439429929e-05,
+      "loss": 0.0719,
+      "step": 7532
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 6.1229216152019e-05,
+      "loss": 0.1461,
+      "step": 7533
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 6.122327790973871e-05,
+      "loss": 0.0923,
+      "step": 7534
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 6.121733966745845e-05,
+      "loss": 0.13,
+      "step": 7535
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 6.121140142517815e-05,
+      "loss": 0.1073,
+      "step": 7536
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 6.120546318289786e-05,
+      "loss": 0.4943,
+      "step": 7537
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 6.119952494061758e-05,
+      "loss": 0.3512,
+      "step": 7538
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 6.119358669833729e-05,
+      "loss": 0.2342,
+      "step": 7539
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 6.118764845605701e-05,
+      "loss": 0.2855,
+      "step": 7540
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 6.118171021377673e-05,
+      "loss": 0.2786,
+      "step": 7541
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 6.117577197149645e-05,
+      "loss": 0.1572,
+      "step": 7542
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 6.116983372921616e-05,
+      "loss": 0.2182,
+      "step": 7543
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 6.116389548693588e-05,
+      "loss": 0.1985,
+      "step": 7544
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 6.115795724465558e-05,
+      "loss": 0.2209,
+      "step": 7545
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 6.11520190023753e-05,
+      "loss": 0.3368,
+      "step": 7546
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 6.114608076009501e-05,
+      "loss": 0.1933,
+      "step": 7547
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 6.114014251781473e-05,
+      "loss": 0.1706,
+      "step": 7548
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 6.113420427553445e-05,
+      "loss": 0.1975,
+      "step": 7549
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 6.112826603325416e-05,
+      "loss": 0.202,
+      "step": 7550
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 6.112232779097388e-05,
+      "loss": 0.1683,
+      "step": 7551
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 6.111638954869358e-05,
+      "loss": 0.1777,
+      "step": 7552
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 6.11104513064133e-05,
+      "loss": 0.1631,
+      "step": 7553
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 6.110451306413301e-05,
+      "loss": 0.2708,
+      "step": 7554
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 6.109857482185275e-05,
+      "loss": 0.1991,
+      "step": 7555
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 6.109263657957245e-05,
+      "loss": 0.1344,
+      "step": 7556
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 6.108669833729217e-05,
+      "loss": 0.1825,
+      "step": 7557
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 6.108076009501188e-05,
+      "loss": 0.1898,
+      "step": 7558
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 6.107482185273159e-05,
+      "loss": 0.2022,
+      "step": 7559
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 6.10688836104513e-05,
+      "loss": 0.2048,
+      "step": 7560
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 6.106294536817103e-05,
+      "loss": 0.1533,
+      "step": 7561
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 6.105700712589075e-05,
+      "loss": 0.1624,
+      "step": 7562
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 6.105106888361045e-05,
+      "loss": 0.1297,
+      "step": 7563
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 6.104513064133017e-05,
+      "loss": 0.2176,
+      "step": 7564
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 6.103919239904988e-05,
+      "loss": 0.1369,
+      "step": 7565
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 6.10332541567696e-05,
+      "loss": 0.1401,
+      "step": 7566
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 6.1027315914489315e-05,
+      "loss": 0.2481,
+      "step": 7567
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 6.102137767220902e-05,
+      "loss": 0.1771,
+      "step": 7568
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 6.101543942992874e-05,
+      "loss": 0.1389,
+      "step": 7569
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 6.1009501187648456e-05,
+      "loss": 0.1194,
+      "step": 7570
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 6.1003562945368176e-05,
+      "loss": 0.1639,
+      "step": 7571
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 6.099762470308788e-05,
+      "loss": 0.1231,
+      "step": 7572
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 6.099168646080761e-05,
+      "loss": 0.1472,
+      "step": 7573
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 6.098574821852732e-05,
+      "loss": 0.144,
+      "step": 7574
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 6.097980997624704e-05,
+      "loss": 0.0834,
+      "step": 7575
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 6.097387173396675e-05,
+      "loss": 0.2922,
+      "step": 7576
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 6.096793349168647e-05,
+      "loss": 0.3302,
+      "step": 7577
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 6.096199524940618e-05,
+      "loss": 0.1193,
+      "step": 7578
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 6.095605700712589e-05,
+      "loss": 0.3172,
+      "step": 7579
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 6.095011876484561e-05,
+      "loss": 0.0581,
+      "step": 7580
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 6.094418052256532e-05,
+      "loss": 0.1549,
+      "step": 7581
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 6.093824228028504e-05,
+      "loss": 0.1078,
+      "step": 7582
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 6.093230403800475e-05,
+      "loss": 0.1293,
+      "step": 7583
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 6.092636579572447e-05,
+      "loss": 0.0835,
+      "step": 7584
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 6.092042755344418e-05,
+      "loss": 0.0709,
+      "step": 7585
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 6.09144893111639e-05,
+      "loss": 0.152,
+      "step": 7586
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 6.0908551068883614e-05,
+      "loss": 0.2989,
+      "step": 7587
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 6.0902612826603334e-05,
+      "loss": 0.2957,
+      "step": 7588
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 6.089667458432304e-05,
+      "loss": 0.3443,
+      "step": 7589
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 6.0890736342042754e-05,
+      "loss": 0.2291,
+      "step": 7590
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 6.0884798099762475e-05,
+      "loss": 0.1509,
+      "step": 7591
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 6.087885985748218e-05,
+      "loss": 0.2414,
+      "step": 7592
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 6.087292161520191e-05,
+      "loss": 0.2438,
+      "step": 7593
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 6.0866983372921615e-05,
+      "loss": 0.2049,
+      "step": 7594
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 6.0861045130641336e-05,
+      "loss": 0.212,
+      "step": 7595
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 6.085510688836105e-05,
+      "loss": 0.2016,
+      "step": 7596
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 6.084916864608077e-05,
+      "loss": 0.2793,
+      "step": 7597
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 6.0843230403800477e-05,
+      "loss": 0.1975,
+      "step": 7598
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 6.083729216152019e-05,
+      "loss": 0.2483,
+      "step": 7599
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 6.083135391923991e-05,
+      "loss": 0.2457,
+      "step": 7600
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 6.082541567695962e-05,
+      "loss": 0.1837,
+      "step": 7601
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 6.081947743467934e-05,
+      "loss": 0.1377,
+      "step": 7602
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 6.081353919239905e-05,
+      "loss": 0.2095,
+      "step": 7603
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 6.080760095011877e-05,
+      "loss": 0.1845,
+      "step": 7604
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 6.080166270783848e-05,
+      "loss": 0.1559,
+      "step": 7605
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 6.07957244655582e-05,
+      "loss": 0.4095,
+      "step": 7606
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 6.078978622327791e-05,
+      "loss": 0.2031,
+      "step": 7607
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 6.078384798099763e-05,
+      "loss": 0.1954,
+      "step": 7608
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 6.077790973871734e-05,
+      "loss": 0.1567,
+      "step": 7609
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 6.077197149643705e-05,
+      "loss": 0.1302,
+      "step": 7610
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 6.0766033254156773e-05,
+      "loss": 0.1418,
+      "step": 7611
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 6.076009501187648e-05,
+      "loss": 0.1804,
+      "step": 7612
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 6.075415676959621e-05,
+      "loss": 0.1901,
+      "step": 7613
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 6.0748218527315914e-05,
+      "loss": 0.251,
+      "step": 7614
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 6.0742280285035635e-05,
+      "loss": 0.1608,
+      "step": 7615
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 6.073634204275535e-05,
+      "loss": 0.1746,
+      "step": 7616
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 6.073040380047507e-05,
+      "loss": 0.1768,
+      "step": 7617
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 6.0724465558194775e-05,
+      "loss": 0.171,
+      "step": 7618
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 6.0718527315914496e-05,
+      "loss": 0.1553,
+      "step": 7619
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 6.071258907363421e-05,
+      "loss": 0.1056,
+      "step": 7620
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 6.0706650831353916e-05,
+      "loss": 0.2088,
+      "step": 7621
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 6.0700712589073636e-05,
+      "loss": 0.1251,
+      "step": 7622
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 6.069477434679335e-05,
+      "loss": 0.1544,
+      "step": 7623
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 6.068883610451307e-05,
+      "loss": 0.1504,
+      "step": 7624
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 6.068289786223278e-05,
+      "loss": 0.1967,
+      "step": 7625
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 6.06769596199525e-05,
+      "loss": 0.1395,
+      "step": 7626
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 6.067102137767221e-05,
+      "loss": 0.1046,
+      "step": 7627
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 6.066508313539193e-05,
+      "loss": 0.108,
+      "step": 7628
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 6.065914489311164e-05,
+      "loss": 0.1743,
+      "step": 7629
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 6.0653206650831365e-05,
+      "loss": 0.0861,
+      "step": 7630
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 6.064726840855107e-05,
+      "loss": 0.135,
+      "step": 7631
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 6.064133016627078e-05,
+      "loss": 0.1094,
+      "step": 7632
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 6.0635391923990506e-05,
+      "loss": 0.1185,
+      "step": 7633
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 6.062945368171021e-05,
+      "loss": 0.1747,
+      "step": 7634
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 6.062351543942993e-05,
+      "loss": 0.1202,
+      "step": 7635
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 6.061757719714965e-05,
+      "loss": 0.3279,
+      "step": 7636
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 6.061163895486937e-05,
+      "loss": 0.3494,
+      "step": 7637
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 6.0605700712589074e-05,
+      "loss": 0.3948,
+      "step": 7638
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 6.0599762470308794e-05,
+      "loss": 0.3136,
+      "step": 7639
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 6.059382422802851e-05,
+      "loss": 0.3094,
+      "step": 7640
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 6.0587885985748215e-05,
+      "loss": 0.1613,
+      "step": 7641
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 6.0581947743467935e-05,
+      "loss": 0.259,
+      "step": 7642
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 6.057600950118765e-05,
+      "loss": 0.1633,
+      "step": 7643
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 6.057007125890737e-05,
+      "loss": 0.1706,
+      "step": 7644
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 6.0564133016627076e-05,
+      "loss": 0.1647,
+      "step": 7645
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 6.0558194774346796e-05,
+      "loss": 0.165,
+      "step": 7646
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 6.055225653206651e-05,
+      "loss": 0.1889,
+      "step": 7647
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 6.054631828978623e-05,
+      "loss": 0.3312,
+      "step": 7648
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 6.054038004750594e-05,
+      "loss": 0.2758,
+      "step": 7649
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 6.0534441805225664e-05,
+      "loss": 0.1751,
+      "step": 7650
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 6.052850356294537e-05,
+      "loss": 0.1568,
+      "step": 7651
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 6.0522565320665084e-05,
+      "loss": 0.1797,
+      "step": 7652
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 6.0516627078384805e-05,
+      "loss": 0.2005,
+      "step": 7653
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 6.051068883610451e-05,
+      "loss": 0.194,
+      "step": 7654
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 6.050475059382423e-05,
+      "loss": 0.1609,
+      "step": 7655
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 6.0498812351543946e-05,
+      "loss": 0.1327,
+      "step": 7656
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 6.0492874109263666e-05,
+      "loss": 0.1613,
+      "step": 7657
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 6.048693586698337e-05,
+      "loss": 0.1837,
+      "step": 7658
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 6.048099762470309e-05,
+      "loss": 0.2028,
+      "step": 7659
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 6.047505938242281e-05,
+      "loss": 0.2492,
+      "step": 7660
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 6.046912114014253e-05,
+      "loss": 0.1771,
+      "step": 7661
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 6.0463182897862234e-05,
+      "loss": 0.1973,
+      "step": 7662
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 6.045724465558195e-05,
+      "loss": 0.1046,
+      "step": 7663
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 6.045130641330167e-05,
+      "loss": 0.1542,
+      "step": 7664
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 6.0445368171021375e-05,
+      "loss": 0.1831,
+      "step": 7665
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 6.0439429928741095e-05,
+      "loss": 0.179,
+      "step": 7666
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 6.043349168646081e-05,
+      "loss": 0.2934,
+      "step": 7667
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 6.042755344418053e-05,
+      "loss": 0.1729,
+      "step": 7668
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 6.042161520190024e-05,
+      "loss": 0.1254,
+      "step": 7669
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 6.041567695961996e-05,
+      "loss": 0.0974,
+      "step": 7670
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 6.040973871733967e-05,
+      "loss": 0.1308,
+      "step": 7671
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 6.040380047505938e-05,
+      "loss": 0.1732,
+      "step": 7672
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 6.0397862232779104e-05,
+      "loss": 0.1309,
+      "step": 7673
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 6.039192399049881e-05,
+      "loss": 0.1376,
+      "step": 7674
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 6.038598574821853e-05,
+      "loss": 0.0998,
+      "step": 7675
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 6.0380047505938244e-05,
+      "loss": 0.1374,
+      "step": 7676
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 6.0374109263657965e-05,
+      "loss": 0.1447,
+      "step": 7677
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 6.036817102137767e-05,
+      "loss": 0.1337,
+      "step": 7678
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 6.036223277909739e-05,
+      "loss": 0.1278,
+      "step": 7679
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 6.0356294536817105e-05,
+      "loss": 0.1514,
+      "step": 7680
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 6.0350356294536826e-05,
+      "loss": 0.1836,
+      "step": 7681
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 6.034441805225653e-05,
+      "loss": 0.2126,
+      "step": 7682
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 6.0338479809976246e-05,
+      "loss": 0.1273,
+      "step": 7683
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 6.0332541567695966e-05,
+      "loss": 0.1318,
+      "step": 7684
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 6.032660332541567e-05,
+      "loss": 0.1305,
+      "step": 7685
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 6.0320665083135394e-05,
+      "loss": 0.0488,
+      "step": 7686
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 6.031472684085511e-05,
+      "loss": 0.3922,
+      "step": 7687
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 6.030878859857483e-05,
+      "loss": 0.3806,
+      "step": 7688
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 6.030285035629454e-05,
+      "loss": 0.3493,
+      "step": 7689
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 6.029691211401426e-05,
+      "loss": 0.3133,
+      "step": 7690
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 6.029097387173397e-05,
+      "loss": 0.2015,
+      "step": 7691
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 6.028503562945369e-05,
+      "loss": 0.1817,
+      "step": 7692
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 6.02790973871734e-05,
+      "loss": 0.1481,
+      "step": 7693
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.027315914489311e-05,
+      "loss": 0.2069,
+      "step": 7694
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.026722090261283e-05,
+      "loss": 0.2033,
+      "step": 7695
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.026128266033254e-05,
+      "loss": 0.2166,
+      "step": 7696
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.025534441805226e-05,
+      "loss": 0.1526,
+      "step": 7697
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.024940617577197e-05,
+      "loss": 0.3172,
+      "step": 7698
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.024346793349169e-05,
+      "loss": 0.2447,
+      "step": 7699
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.0237529691211404e-05,
+      "loss": 0.1756,
+      "step": 7700
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.0231591448931124e-05,
+      "loss": 0.2533,
+      "step": 7701
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.022565320665083e-05,
+      "loss": 0.1427,
+      "step": 7702
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.0219714964370545e-05,
+      "loss": 0.3381,
+      "step": 7703
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.0213776722090265e-05,
+      "loss": 0.1757,
+      "step": 7704
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.020783847980997e-05,
+      "loss": 0.1842,
+      "step": 7705
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.02019002375297e-05,
+      "loss": 0.3879,
+      "step": 7706
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.0195961995249406e-05,
+      "loss": 0.1705,
+      "step": 7707
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.0190023752969126e-05,
+      "loss": 0.1216,
+      "step": 7708
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.018408551068884e-05,
+      "loss": 0.1638,
+      "step": 7709
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.017814726840856e-05,
+      "loss": 0.2165,
+      "step": 7710
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.017220902612827e-05,
+      "loss": 0.1614,
+      "step": 7711
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.016627078384799e-05,
+      "loss": 0.1647,
+      "step": 7712
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.01603325415677e-05,
+      "loss": 0.1935,
+      "step": 7713
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.015439429928741e-05,
+      "loss": 0.1923,
+      "step": 7714
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.014845605700713e-05,
+      "loss": 0.1581,
+      "step": 7715
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.014251781472684e-05,
+      "loss": 0.1445,
+      "step": 7716
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.013657957244656e-05,
+      "loss": 0.2184,
+      "step": 7717
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.013064133016627e-05,
+      "loss": 0.1448,
+      "step": 7718
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.012470308788599e-05,
+      "loss": 0.1965,
+      "step": 7719
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.01187648456057e-05,
+      "loss": 0.1097,
+      "step": 7720
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.011282660332542e-05,
+      "loss": 0.1666,
+      "step": 7721
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.010688836104513e-05,
+      "loss": 0.204,
+      "step": 7722
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.010095011876486e-05,
+      "loss": 0.1031,
+      "step": 7723
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.0095011876484564e-05,
+      "loss": 0.1272,
+      "step": 7724
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.008907363420427e-05,
+      "loss": 0.1626,
+      "step": 7725
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.0083135391924e-05,
+      "loss": 0.1584,
+      "step": 7726
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.0077197149643705e-05,
+      "loss": 0.1195,
+      "step": 7727
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.0071258907363425e-05,
+      "loss": 0.1127,
+      "step": 7728
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.006532066508314e-05,
+      "loss": 0.2314,
+      "step": 7729
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 6.005938242280286e-05,
+      "loss": 0.1447,
+      "step": 7730
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 6.0053444180522566e-05,
+      "loss": 0.107,
+      "step": 7731
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 6.0047505938242286e-05,
+      "loss": 0.1718,
+      "step": 7732
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 6.0041567695962e-05,
+      "loss": 0.1425,
+      "step": 7733
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 6.0035629453681706e-05,
+      "loss": 0.2156,
+      "step": 7734
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 6.002969121140143e-05,
+      "loss": 0.1734,
+      "step": 7735
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 6.002375296912114e-05,
+      "loss": 0.0964,
+      "step": 7736
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 6.001781472684086e-05,
+      "loss": 0.4946,
+      "step": 7737
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 6.001187648456057e-05,
+      "loss": 0.404,
+      "step": 7738
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 6.000593824228029e-05,
+      "loss": 0.4148,
+      "step": 7739
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 6e-05,
+      "loss": 0.2311,
+      "step": 7740
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.999406175771972e-05,
+      "loss": 0.2577,
+      "step": 7741
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.998812351543943e-05,
+      "loss": 0.2639,
+      "step": 7742
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.9982185273159156e-05,
+      "loss": 0.1436,
+      "step": 7743
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.997624703087886e-05,
+      "loss": 0.231,
+      "step": 7744
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.997030878859857e-05,
+      "loss": 0.2569,
+      "step": 7745
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.9964370546318297e-05,
+      "loss": 0.2337,
+      "step": 7746
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.9958432304038e-05,
+      "loss": 0.1689,
+      "step": 7747
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.9952494061757724e-05,
+      "loss": 0.1864,
+      "step": 7748
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.994655581947744e-05,
+      "loss": 0.1952,
+      "step": 7749
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.994061757719716e-05,
+      "loss": 0.1755,
+      "step": 7750
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.9934679334916864e-05,
+      "loss": 0.192,
+      "step": 7751
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.9928741092636585e-05,
+      "loss": 0.2527,
+      "step": 7752
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.99228028503563e-05,
+      "loss": 0.1692,
+      "step": 7753
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.991686460807602e-05,
+      "loss": 0.2011,
+      "step": 7754
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.9910926365795726e-05,
+      "loss": 0.2313,
+      "step": 7755
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.990498812351544e-05,
+      "loss": 0.1693,
+      "step": 7756
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.989904988123516e-05,
+      "loss": 0.2311,
+      "step": 7757
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.9893111638954866e-05,
+      "loss": 0.222,
+      "step": 7758
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.988717339667459e-05,
+      "loss": 0.1498,
+      "step": 7759
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.98812351543943e-05,
+      "loss": 0.1708,
+      "step": 7760
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.987529691211402e-05,
+      "loss": 0.1925,
+      "step": 7761
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.986935866983373e-05,
+      "loss": 0.2746,
+      "step": 7762
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.9863420427553455e-05,
+      "loss": 0.2199,
+      "step": 7763
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.985748218527316e-05,
+      "loss": 0.2612,
+      "step": 7764
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.985154394299287e-05,
+      "loss": 0.1476,
+      "step": 7765
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.9845605700712595e-05,
+      "loss": 1.0475,
+      "step": 7766
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.98396674584323e-05,
+      "loss": 0.2054,
+      "step": 7767
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.983372921615202e-05,
+      "loss": 0.1962,
+      "step": 7768
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.9827790973871736e-05,
+      "loss": 0.1536,
+      "step": 7769
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.9821852731591456e-05,
+      "loss": 0.2195,
+      "step": 7770
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.981591448931116e-05,
+      "loss": 0.1736,
+      "step": 7771
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.9809976247030884e-05,
+      "loss": 0.1566,
+      "step": 7772
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.98040380047506e-05,
+      "loss": 0.1925,
+      "step": 7773
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.979809976247032e-05,
+      "loss": 0.1774,
+      "step": 7774
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.9792161520190024e-05,
+      "loss": 0.0879,
+      "step": 7775
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.978622327790974e-05,
+      "loss": 0.1413,
+      "step": 7776
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.978028503562946e-05,
+      "loss": 0.1559,
+      "step": 7777
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.9774346793349165e-05,
+      "loss": 0.1403,
+      "step": 7778
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.9768408551068885e-05,
+      "loss": 0.1362,
+      "step": 7779
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.97624703087886e-05,
+      "loss": 0.1308,
+      "step": 7780
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.975653206650832e-05,
+      "loss": 0.1065,
+      "step": 7781
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.9750593824228026e-05,
+      "loss": 0.1629,
+      "step": 7782
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.974465558194775e-05,
+      "loss": 0.1222,
+      "step": 7783
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.973871733966746e-05,
+      "loss": 0.1119,
+      "step": 7784
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.973277909738718e-05,
+      "loss": 0.0921,
+      "step": 7785
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.9726840855106894e-05,
+      "loss": 0.2135,
+      "step": 7786
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.97209026128266e-05,
+      "loss": 0.439,
+      "step": 7787
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.971496437054632e-05,
+      "loss": 0.2869,
+      "step": 7788
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.9709026128266035e-05,
+      "loss": 0.236,
+      "step": 7789
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.9703087885985755e-05,
+      "loss": 0.2706,
+      "step": 7790
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.969714964370546e-05,
+      "loss": 0.2166,
+      "step": 7791
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.969121140142518e-05,
+      "loss": 0.2518,
+      "step": 7792
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.9685273159144896e-05,
+      "loss": 0.1972,
+      "step": 7793
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.9679334916864616e-05,
+      "loss": 0.2103,
+      "step": 7794
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.967339667458432e-05,
+      "loss": 0.2308,
+      "step": 7795
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.9667458432304037e-05,
+      "loss": 0.1921,
+      "step": 7796
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.966152019002376e-05,
+      "loss": 0.1688,
+      "step": 7797
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.9655581947743464e-05,
+      "loss": 0.1581,
+      "step": 7798
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.9649643705463184e-05,
+      "loss": 0.1456,
+      "step": 7799
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.96437054631829e-05,
+      "loss": 0.2748,
+      "step": 7800
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.963776722090262e-05,
+      "loss": 0.217,
+      "step": 7801
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.963182897862233e-05,
+      "loss": 0.217,
+      "step": 7802
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.962589073634205e-05,
+      "loss": 0.1968,
+      "step": 7803
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.961995249406176e-05,
+      "loss": 0.175,
+      "step": 7804
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.961401425178148e-05,
+      "loss": 0.1317,
+      "step": 7805
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.960807600950119e-05,
+      "loss": 0.1605,
+      "step": 7806
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.96021377672209e-05,
+      "loss": 0.2373,
+      "step": 7807
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.959619952494062e-05,
+      "loss": 0.1974,
+      "step": 7808
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.9590261282660333e-05,
+      "loss": 0.2157,
+      "step": 7809
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.9584323040380054e-05,
+      "loss": 0.1509,
+      "step": 7810
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.957838479809976e-05,
+      "loss": 0.305,
+      "step": 7811
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.957244655581948e-05,
+      "loss": 0.176,
+      "step": 7812
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.9566508313539195e-05,
+      "loss": 0.1806,
+      "step": 7813
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.9560570071258915e-05,
+      "loss": 0.1548,
+      "step": 7814
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.955463182897862e-05,
+      "loss": 0.1601,
+      "step": 7815
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.954869358669834e-05,
+      "loss": 0.1505,
+      "step": 7816
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.9542755344418056e-05,
+      "loss": 0.2195,
+      "step": 7817
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.953681710213776e-05,
+      "loss": 0.1499,
+      "step": 7818
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.953087885985749e-05,
+      "loss": 0.1754,
+      "step": 7819
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.9524940617577196e-05,
+      "loss": 0.3051,
+      "step": 7820
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.951900237529692e-05,
+      "loss": 0.2138,
+      "step": 7821
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.951306413301663e-05,
+      "loss": 0.1164,
+      "step": 7822
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.950712589073635e-05,
+      "loss": 0.2751,
+      "step": 7823
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.950118764845606e-05,
+      "loss": 0.1145,
+      "step": 7824
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.949524940617578e-05,
+      "loss": 0.3054,
+      "step": 7825
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.948931116389549e-05,
+      "loss": 0.1407,
+      "step": 7826
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.948337292161521e-05,
+      "loss": 0.1266,
+      "step": 7827
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.947743467933492e-05,
+      "loss": 0.1402,
+      "step": 7828
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.947149643705463e-05,
+      "loss": 0.0864,
+      "step": 7829
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.946555819477435e-05,
+      "loss": 0.1662,
+      "step": 7830
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.945961995249406e-05,
+      "loss": 0.1202,
+      "step": 7831
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.945368171021378e-05,
+      "loss": 0.2031,
+      "step": 7832
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.944774346793349e-05,
+      "loss": 0.2485,
+      "step": 7833
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.9441805225653214e-05,
+      "loss": 0.1536,
+      "step": 7834
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.943586698337292e-05,
+      "loss": 0.1516,
+      "step": 7835
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.942992874109265e-05,
+      "loss": 0.1853,
+      "step": 7836
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.9423990498812354e-05,
+      "loss": 0.5384,
+      "step": 7837
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.941805225653206e-05,
+      "loss": 0.2622,
+      "step": 7838
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.941211401425179e-05,
+      "loss": 0.3845,
+      "step": 7839
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.9406175771971495e-05,
+      "loss": 0.3504,
+      "step": 7840
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.9400237529691215e-05,
+      "loss": 0.2217,
+      "step": 7841
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.939429928741093e-05,
+      "loss": 0.2518,
+      "step": 7842
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.938836104513065e-05,
+      "loss": 0.176,
+      "step": 7843
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.9382422802850356e-05,
+      "loss": 0.1588,
+      "step": 7844
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.9376484560570077e-05,
+      "loss": 0.1671,
+      "step": 7845
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.937054631828979e-05,
+      "loss": 0.1976,
+      "step": 7846
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.936460807600951e-05,
+      "loss": 0.2261,
+      "step": 7847
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.935866983372922e-05,
+      "loss": 0.2609,
+      "step": 7848
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.935273159144893e-05,
+      "loss": 0.1464,
+      "step": 7849
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.934679334916865e-05,
+      "loss": 0.2066,
+      "step": 7850
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.934085510688836e-05,
+      "loss": 0.1927,
+      "step": 7851
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.933491686460808e-05,
+      "loss": 0.2323,
+      "step": 7852
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.932897862232779e-05,
+      "loss": 0.1688,
+      "step": 7853
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.932304038004751e-05,
+      "loss": 0.2974,
+      "step": 7854
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.931710213776722e-05,
+      "loss": 0.1498,
+      "step": 7855
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.9311163895486946e-05,
+      "loss": 0.227,
+      "step": 7856
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.930522565320665e-05,
+      "loss": 0.2255,
+      "step": 7857
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.9299287410926373e-05,
+      "loss": 0.1993,
+      "step": 7858
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.929334916864609e-05,
+      "loss": 0.2165,
+      "step": 7859
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.9287410926365794e-05,
+      "loss": 0.213,
+      "step": 7860
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.9281472684085514e-05,
+      "loss": 0.1894,
+      "step": 7861
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.927553444180523e-05,
+      "loss": 0.1332,
+      "step": 7862
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.926959619952495e-05,
+      "loss": 0.2251,
+      "step": 7863
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.9263657957244655e-05,
+      "loss": 0.1526,
+      "step": 7864
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.9257719714964375e-05,
+      "loss": 0.1777,
+      "step": 7865
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.925178147268409e-05,
+      "loss": 0.1657,
+      "step": 7866
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.924584323040381e-05,
+      "loss": 0.1473,
+      "step": 7867
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.9239904988123516e-05,
+      "loss": 0.1433,
+      "step": 7868
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.923396674584323e-05,
+      "loss": 0.2662,
+      "step": 7869
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.922802850356295e-05,
+      "loss": 0.1764,
+      "step": 7870
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.922209026128266e-05,
+      "loss": 0.1308,
+      "step": 7871
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.921615201900238e-05,
+      "loss": 0.1588,
+      "step": 7872
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.921021377672209e-05,
+      "loss": 0.1695,
+      "step": 7873
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.920427553444181e-05,
+      "loss": 0.231,
+      "step": 7874
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.919833729216152e-05,
+      "loss": 0.2086,
+      "step": 7875
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.9192399049881245e-05,
+      "loss": 0.2479,
+      "step": 7876
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.918646080760095e-05,
+      "loss": 0.1874,
+      "step": 7877
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.918052256532067e-05,
+      "loss": 0.1725,
+      "step": 7878
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.9174584323040386e-05,
+      "loss": 0.1869,
+      "step": 7879
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.916864608076009e-05,
+      "loss": 0.1501,
+      "step": 7880
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.916270783847981e-05,
+      "loss": 0.1411,
+      "step": 7881
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.9156769596199526e-05,
+      "loss": 0.1237,
+      "step": 7882
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.915083135391925e-05,
+      "loss": 0.2105,
+      "step": 7883
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.9144893111638954e-05,
+      "loss": 0.0943,
+      "step": 7884
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.9138954869358674e-05,
+      "loss": 0.2026,
+      "step": 7885
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.913301662707839e-05,
+      "loss": 0.1834,
+      "step": 7886
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.912707838479811e-05,
+      "loss": 0.4387,
+      "step": 7887
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.9121140142517815e-05,
+      "loss": 0.3213,
+      "step": 7888
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.9115201900237535e-05,
+      "loss": 0.3065,
+      "step": 7889
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.910926365795725e-05,
+      "loss": 0.2113,
+      "step": 7890
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.9103325415676955e-05,
+      "loss": 0.2677,
+      "step": 7891
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.9097387173396676e-05,
+      "loss": 0.1855,
+      "step": 7892
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.909144893111639e-05,
+      "loss": 0.1945,
+      "step": 7893
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.908551068883611e-05,
+      "loss": 0.1737,
+      "step": 7894
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.9079572446555817e-05,
+      "loss": 0.1883,
+      "step": 7895
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.9073634204275544e-05,
+      "loss": 0.2,
+      "step": 7896
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.906769596199525e-05,
+      "loss": 0.1858,
+      "step": 7897
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.906175771971497e-05,
+      "loss": 0.2383,
+      "step": 7898
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.9055819477434684e-05,
+      "loss": 0.2219,
+      "step": 7899
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.904988123515439e-05,
+      "loss": 0.2925,
+      "step": 7900
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.904394299287411e-05,
+      "loss": 0.1696,
+      "step": 7901
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.9038004750593825e-05,
+      "loss": 0.1809,
+      "step": 7902
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.9032066508313546e-05,
+      "loss": 0.1633,
+      "step": 7903
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.902612826603325e-05,
+      "loss": 0.26,
+      "step": 7904
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.902019002375297e-05,
+      "loss": 0.142,
+      "step": 7905
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.9014251781472686e-05,
+      "loss": 0.2266,
+      "step": 7906
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.900831353919241e-05,
+      "loss": 0.2323,
+      "step": 7907
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.9002375296912113e-05,
+      "loss": 0.1742,
+      "step": 7908
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.8996437054631834e-05,
+      "loss": 0.1638,
+      "step": 7909
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.899049881235155e-05,
+      "loss": 0.5801,
+      "step": 7910
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.8984560570071254e-05,
+      "loss": 0.1534,
+      "step": 7911
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.8978622327790975e-05,
+      "loss": 0.2005,
+      "step": 7912
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.897268408551069e-05,
+      "loss": 0.1423,
+      "step": 7913
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.896674584323041e-05,
+      "loss": 0.2454,
+      "step": 7914
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.8960807600950115e-05,
+      "loss": 0.1303,
+      "step": 7915
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.895486935866984e-05,
+      "loss": 0.186,
+      "step": 7916
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.894893111638955e-05,
+      "loss": 0.1661,
+      "step": 7917
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.894299287410927e-05,
+      "loss": 0.1556,
+      "step": 7918
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.893705463182898e-05,
+      "loss": 0.1929,
+      "step": 7919
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.8931116389548704e-05,
+      "loss": 0.1863,
+      "step": 7920
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.892517814726841e-05,
+      "loss": 0.1604,
+      "step": 7921
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.8919239904988124e-05,
+      "loss": 0.1188,
+      "step": 7922
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.8913301662707844e-05,
+      "loss": 0.1117,
+      "step": 7923
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.890736342042755e-05,
+      "loss": 0.1084,
+      "step": 7924
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.890142517814727e-05,
+      "loss": 0.5011,
+      "step": 7925
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.8895486935866985e-05,
+      "loss": 0.2965,
+      "step": 7926
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.8889548693586705e-05,
+      "loss": 0.1342,
+      "step": 7927
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.888361045130641e-05,
+      "loss": 0.0727,
+      "step": 7928
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.887767220902613e-05,
+      "loss": 0.093,
+      "step": 7929
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.8871733966745846e-05,
+      "loss": 0.1088,
+      "step": 7930
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.886579572446555e-05,
+      "loss": 0.1713,
+      "step": 7931
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.885985748218527e-05,
+      "loss": 0.1048,
+      "step": 7932
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.885391923990499e-05,
+      "loss": 0.0621,
+      "step": 7933
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.884798099762471e-05,
+      "loss": 0.0898,
+      "step": 7934
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.884204275534442e-05,
+      "loss": 0.2272,
+      "step": 7935
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.883610451306414e-05,
+      "loss": 0.2703,
+      "step": 7936
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.883016627078385e-05,
+      "loss": 0.3958,
+      "step": 7937
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.882422802850357e-05,
+      "loss": 0.3661,
+      "step": 7938
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.881828978622328e-05,
+      "loss": 0.2701,
+      "step": 7939
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.8812351543943e-05,
+      "loss": 0.2965,
+      "step": 7940
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.880641330166271e-05,
+      "loss": 0.2472,
+      "step": 7941
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.880047505938242e-05,
+      "loss": 0.3266,
+      "step": 7942
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.879453681710214e-05,
+      "loss": 0.258,
+      "step": 7943
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.878859857482185e-05,
+      "loss": 0.1731,
+      "step": 7944
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.878266033254157e-05,
+      "loss": 0.186,
+      "step": 7945
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.8776722090261284e-05,
+      "loss": 0.1781,
+      "step": 7946
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.8770783847981004e-05,
+      "loss": 0.2456,
+      "step": 7947
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.876484560570071e-05,
+      "loss": 0.2285,
+      "step": 7948
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.875890736342043e-05,
+      "loss": 0.1795,
+      "step": 7949
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.8752969121140145e-05,
+      "loss": 0.1494,
+      "step": 7950
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.8747030878859865e-05,
+      "loss": 0.2202,
+      "step": 7951
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.874109263657958e-05,
+      "loss": 0.1962,
+      "step": 7952
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.8735154394299286e-05,
+      "loss": 0.1162,
+      "step": 7953
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.8729216152019006e-05,
+      "loss": 0.2057,
+      "step": 7954
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.872327790973872e-05,
+      "loss": 0.243,
+      "step": 7955
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.871733966745844e-05,
+      "loss": 0.2179,
+      "step": 7956
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.871140142517815e-05,
+      "loss": 0.1156,
+      "step": 7957
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.870546318289787e-05,
+      "loss": 0.2295,
+      "step": 7958
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.869952494061758e-05,
+      "loss": 0.1745,
+      "step": 7959
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.86935866983373e-05,
+      "loss": 0.1359,
+      "step": 7960
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.868764845605701e-05,
+      "loss": 0.1501,
+      "step": 7961
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.868171021377672e-05,
+      "loss": 0.1166,
+      "step": 7962
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.867577197149644e-05,
+      "loss": 0.1618,
+      "step": 7963
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.866983372921615e-05,
+      "loss": 0.1594,
+      "step": 7964
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.866389548693587e-05,
+      "loss": 0.2469,
+      "step": 7965
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.865795724465558e-05,
+      "loss": 0.2099,
+      "step": 7966
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.86520190023753e-05,
+      "loss": 0.1594,
+      "step": 7967
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.864608076009501e-05,
+      "loss": 0.3337,
+      "step": 7968
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.864014251781474e-05,
+      "loss": 0.1396,
+      "step": 7969
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.8634204275534444e-05,
+      "loss": 0.1383,
+      "step": 7970
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.8628266033254164e-05,
+      "loss": 0.1707,
+      "step": 7971
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.862232779097388e-05,
+      "loss": 0.173,
+      "step": 7972
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.8616389548693584e-05,
+      "loss": 0.1304,
+      "step": 7973
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.8610451306413305e-05,
+      "loss": 0.1278,
+      "step": 7974
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.860451306413302e-05,
+      "loss": 0.1307,
+      "step": 7975
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.859857482185274e-05,
+      "loss": 0.147,
+      "step": 7976
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.8592636579572445e-05,
+      "loss": 0.1484,
+      "step": 7977
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.8586698337292166e-05,
+      "loss": 0.1247,
+      "step": 7978
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.858076009501188e-05,
+      "loss": 0.2624,
+      "step": 7979
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.85748218527316e-05,
+      "loss": 0.1318,
+      "step": 7980
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.8568883610451306e-05,
+      "loss": 0.1295,
+      "step": 7981
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.856294536817103e-05,
+      "loss": 0.0678,
+      "step": 7982
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.855700712589074e-05,
+      "loss": 0.1053,
+      "step": 7983
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.855106888361045e-05,
+      "loss": 0.2422,
+      "step": 7984
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.854513064133017e-05,
+      "loss": 0.1615,
+      "step": 7985
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.853919239904988e-05,
+      "loss": 0.1529,
+      "step": 7986
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.85332541567696e-05,
+      "loss": 0.3152,
+      "step": 7987
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.852731591448931e-05,
+      "loss": 0.3183,
+      "step": 7988
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.8521377672209035e-05,
+      "loss": 0.2519,
+      "step": 7989
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.851543942992874e-05,
+      "loss": 0.2475,
+      "step": 7990
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.850950118764846e-05,
+      "loss": 0.2471,
+      "step": 7991
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.8503562945368176e-05,
+      "loss": 0.2806,
+      "step": 7992
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.8497624703087897e-05,
+      "loss": 0.2123,
+      "step": 7993
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.84916864608076e-05,
+      "loss": 0.1957,
+      "step": 7994
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.848574821852732e-05,
+      "loss": 0.1764,
+      "step": 7995
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.847980997624704e-05,
+      "loss": 0.2352,
+      "step": 7996
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.8473871733966744e-05,
+      "loss": 0.2224,
+      "step": 7997
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.8467933491686464e-05,
+      "loss": 0.1479,
+      "step": 7998
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.846199524940618e-05,
+      "loss": 0.1739,
+      "step": 7999
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.84560570071259e-05,
+      "loss": 0.1392,
+      "step": 8000
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.8450118764845605e-05,
+      "loss": 0.1753,
+      "step": 8001
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.8444180522565326e-05,
+      "loss": 0.1219,
+      "step": 8002
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.843824228028504e-05,
+      "loss": 0.35,
+      "step": 8003
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.8432304038004746e-05,
+      "loss": 0.1755,
+      "step": 8004
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.8426365795724466e-05,
+      "loss": 0.37,
+      "step": 8005
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.842042755344418e-05,
+      "loss": 0.1786,
+      "step": 8006
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.84144893111639e-05,
+      "loss": 0.1192,
+      "step": 8007
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.840855106888361e-05,
+      "loss": 0.2998,
+      "step": 8008
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.8402612826603334e-05,
+      "loss": 0.1272,
+      "step": 8009
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.839667458432304e-05,
+      "loss": 0.4402,
+      "step": 8010
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.839073634204276e-05,
+      "loss": 0.2508,
+      "step": 8011
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.8384798099762475e-05,
+      "loss": 0.1738,
+      "step": 8012
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.8378859857482195e-05,
+      "loss": 0.1323,
+      "step": 8013
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.83729216152019e-05,
+      "loss": 0.3252,
+      "step": 8014
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.8366983372921616e-05,
+      "loss": 0.1032,
+      "step": 8015
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.8361045130641336e-05,
+      "loss": 0.1343,
+      "step": 8016
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.835510688836104e-05,
+      "loss": 0.1455,
+      "step": 8017
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.834916864608076e-05,
+      "loss": 0.1447,
+      "step": 8018
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.834323040380048e-05,
+      "loss": 0.2201,
+      "step": 8019
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.83372921615202e-05,
+      "loss": 0.1748,
+      "step": 8020
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.8331353919239904e-05,
+      "loss": 0.2039,
+      "step": 8021
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.8325415676959624e-05,
+      "loss": 0.1175,
+      "step": 8022
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.831947743467934e-05,
+      "loss": 0.1684,
+      "step": 8023
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.831353919239906e-05,
+      "loss": 0.1597,
+      "step": 8024
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.8307600950118765e-05,
+      "loss": 0.2579,
+      "step": 8025
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.830166270783848e-05,
+      "loss": 0.1848,
+      "step": 8026
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.82957244655582e-05,
+      "loss": 0.0789,
+      "step": 8027
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.82957244655582e-05,
+      "loss": 0.157,
+      "step": 8028
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.8289786223277906e-05,
+      "loss": 0.2756,
+      "step": 8029
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.828384798099763e-05,
+      "loss": 0.2853,
+      "step": 8030
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.827790973871734e-05,
+      "loss": 0.387,
+      "step": 8031
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.827197149643706e-05,
+      "loss": 0.2414,
+      "step": 8032
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.8266033254156774e-05,
+      "loss": 0.2021,
+      "step": 8033
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.8260095011876494e-05,
+      "loss": 0.1604,
+      "step": 8034
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.82541567695962e-05,
+      "loss": 0.1791,
+      "step": 8035
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.8248218527315914e-05,
+      "loss": 0.1196,
+      "step": 8036
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.8242280285035635e-05,
+      "loss": 0.1944,
+      "step": 8037
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.823634204275534e-05,
+      "loss": 0.157,
+      "step": 8038
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.823040380047506e-05,
+      "loss": 0.1497,
+      "step": 8039
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.8224465558194775e-05,
+      "loss": 0.1705,
+      "step": 8040
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.8218527315914496e-05,
+      "loss": 0.1718,
+      "step": 8041
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.82125890736342e-05,
+      "loss": 0.1135,
+      "step": 8042
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.820665083135392e-05,
+      "loss": 0.1814,
+      "step": 8043
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.8200712589073637e-05,
+      "loss": 0.0758,
+      "step": 8044
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.819477434679336e-05,
+      "loss": 0.2757,
+      "step": 8045
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.8188836104513064e-05,
+      "loss": 0.1383,
+      "step": 8046
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.818289786223278e-05,
+      "loss": 0.2861,
+      "step": 8047
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.81769596199525e-05,
+      "loss": 0.1511,
+      "step": 8048
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.817102137767221e-05,
+      "loss": 0.1025,
+      "step": 8049
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.816508313539193e-05,
+      "loss": 0.1842,
+      "step": 8050
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.815914489311164e-05,
+      "loss": 0.1354,
+      "step": 8051
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.815320665083136e-05,
+      "loss": 0.1202,
+      "step": 8052
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.814726840855107e-05,
+      "loss": 0.1266,
+      "step": 8053
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.814133016627079e-05,
+      "loss": 0.2567,
+      "step": 8054
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.81353919239905e-05,
+      "loss": 0.1122,
+      "step": 8055
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.812945368171022e-05,
+      "loss": 0.0964,
+      "step": 8056
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.8123515439429933e-05,
+      "loss": 0.1235,
+      "step": 8057
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.811757719714964e-05,
+      "loss": 0.1243,
+      "step": 8058
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.811163895486936e-05,
+      "loss": 0.164,
+      "step": 8059
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.8105700712589074e-05,
+      "loss": 0.128,
+      "step": 8060
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.8099762470308795e-05,
+      "loss": 0.1766,
+      "step": 8061
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.80938242280285e-05,
+      "loss": 0.1033,
+      "step": 8062
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.808788598574822e-05,
+      "loss": 0.0837,
+      "step": 8063
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.8081947743467935e-05,
+      "loss": 0.1265,
+      "step": 8064
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.8076009501187656e-05,
+      "loss": 0.107,
+      "step": 8065
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.807007125890737e-05,
+      "loss": 0.1052,
+      "step": 8066
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.8064133016627076e-05,
+      "loss": 0.1116,
+      "step": 8067
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.8058194774346796e-05,
+      "loss": 0.1469,
+      "step": 8068
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.805225653206651e-05,
+      "loss": 0.0763,
+      "step": 8069
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.804631828978623e-05,
+      "loss": 0.1329,
+      "step": 8070
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.804038004750594e-05,
+      "loss": 0.0627,
+      "step": 8071
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.803444180522566e-05,
+      "loss": 0.0493,
+      "step": 8072
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.802850356294537e-05,
+      "loss": 0.1467,
+      "step": 8073
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.802256532066509e-05,
+      "loss": 0.0911,
+      "step": 8074
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.80166270783848e-05,
+      "loss": 0.0912,
+      "step": 8075
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.801068883610452e-05,
+      "loss": 0.0597,
+      "step": 8076
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.800475059382423e-05,
+      "loss": 0.1087,
+      "step": 8077
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.799881235154394e-05,
+      "loss": 0.1341,
+      "step": 8078
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.799287410926366e-05,
+      "loss": 0.2756,
+      "step": 8079
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.798693586698337e-05,
+      "loss": 0.2273,
+      "step": 8080
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.798099762470309e-05,
+      "loss": 0.2531,
+      "step": 8081
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.79750593824228e-05,
+      "loss": 0.1827,
+      "step": 8082
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.796912114014252e-05,
+      "loss": 0.142,
+      "step": 8083
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.7963182897862234e-05,
+      "loss": 0.1988,
+      "step": 8084
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.7957244655581954e-05,
+      "loss": 0.1475,
+      "step": 8085
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.795130641330167e-05,
+      "loss": 0.1907,
+      "step": 8086
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.794536817102139e-05,
+      "loss": 0.1252,
+      "step": 8087
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.7939429928741095e-05,
+      "loss": 0.1542,
+      "step": 8088
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.793349168646081e-05,
+      "loss": 0.1521,
+      "step": 8089
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.792755344418053e-05,
+      "loss": 0.1172,
+      "step": 8090
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.7921615201900236e-05,
+      "loss": 0.1886,
+      "step": 8091
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.7915676959619956e-05,
+      "loss": 0.1279,
+      "step": 8092
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.790973871733967e-05,
+      "loss": 0.0636,
+      "step": 8093
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.790380047505939e-05,
+      "loss": 0.1049,
+      "step": 8094
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.78978622327791e-05,
+      "loss": 0.247,
+      "step": 8095
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.789192399049882e-05,
+      "loss": 0.1378,
+      "step": 8096
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.788598574821853e-05,
+      "loss": 0.1672,
+      "step": 8097
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.788004750593824e-05,
+      "loss": 0.108,
+      "step": 8098
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.787410926365796e-05,
+      "loss": 0.2014,
+      "step": 8099
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.786817102137767e-05,
+      "loss": 0.084,
+      "step": 8100
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.786223277909739e-05,
+      "loss": 0.1084,
+      "step": 8101
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.78562945368171e-05,
+      "loss": 0.1299,
+      "step": 8102
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.7850356294536826e-05,
+      "loss": 0.127,
+      "step": 8103
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.784441805225653e-05,
+      "loss": 0.1112,
+      "step": 8104
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.783847980997625e-05,
+      "loss": 0.1175,
+      "step": 8105
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.783254156769597e-05,
+      "loss": 0.1253,
+      "step": 8106
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.782660332541569e-05,
+      "loss": 0.1764,
+      "step": 8107
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.7820665083135394e-05,
+      "loss": 0.1339,
+      "step": 8108
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.781472684085511e-05,
+      "loss": 0.1678,
+      "step": 8109
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.780878859857483e-05,
+      "loss": 0.1389,
+      "step": 8110
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.7802850356294535e-05,
+      "loss": 0.1377,
+      "step": 8111
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.7796912114014255e-05,
+      "loss": 0.0841,
+      "step": 8112
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.779097387173397e-05,
+      "loss": 0.1026,
+      "step": 8113
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.778503562945369e-05,
+      "loss": 0.0922,
+      "step": 8114
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.7779097387173396e-05,
+      "loss": 0.1086,
+      "step": 8115
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.7773159144893116e-05,
+      "loss": 0.1109,
+      "step": 8116
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.776722090261283e-05,
+      "loss": 0.0522,
+      "step": 8117
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.776128266033255e-05,
+      "loss": 0.1339,
+      "step": 8118
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.775534441805226e-05,
+      "loss": 0.0675,
+      "step": 8119
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.774940617577197e-05,
+      "loss": 0.1531,
+      "step": 8120
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.774346793349169e-05,
+      "loss": 0.0785,
+      "step": 8121
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.77375296912114e-05,
+      "loss": 0.1076,
+      "step": 8122
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.7731591448931125e-05,
+      "loss": 0.0599,
+      "step": 8123
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.772565320665083e-05,
+      "loss": 0.3783,
+      "step": 8124
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.771971496437055e-05,
+      "loss": 0.1129,
+      "step": 8125
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.7713776722090265e-05,
+      "loss": 0.1534,
+      "step": 8126
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.7707838479809986e-05,
+      "loss": 0.1041,
+      "step": 8127
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.770190023752969e-05,
+      "loss": 0.3156,
+      "step": 8128
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.7695961995249406e-05,
+      "loss": 0.3442,
+      "step": 8129
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.7690023752969127e-05,
+      "loss": 0.2792,
+      "step": 8130
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.768408551068883e-05,
+      "loss": 0.1786,
+      "step": 8131
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.7678147268408554e-05,
+      "loss": 0.2508,
+      "step": 8132
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.767220902612827e-05,
+      "loss": 0.2282,
+      "step": 8133
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.766627078384799e-05,
+      "loss": 0.1352,
+      "step": 8134
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.7660332541567694e-05,
+      "loss": 0.1406,
+      "step": 8135
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.7654394299287415e-05,
+      "loss": 0.123,
+      "step": 8136
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.764845605700713e-05,
+      "loss": 0.1539,
+      "step": 8137
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.764251781472685e-05,
+      "loss": 0.1346,
+      "step": 8138
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.7636579572446556e-05,
+      "loss": 0.1845,
+      "step": 8139
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.763064133016627e-05,
+      "loss": 0.1205,
+      "step": 8140
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.762470308788599e-05,
+      "loss": 0.1257,
+      "step": 8141
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.7618764845605696e-05,
+      "loss": 0.1239,
+      "step": 8142
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.761282660332542e-05,
+      "loss": 0.1418,
+      "step": 8143
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.760688836104513e-05,
+      "loss": 0.1129,
+      "step": 8144
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.760095011876485e-05,
+      "loss": 0.1484,
+      "step": 8145
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.7595011876484564e-05,
+      "loss": 0.1244,
+      "step": 8146
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.7589073634204284e-05,
+      "loss": 0.1309,
+      "step": 8147
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.758313539192399e-05,
+      "loss": 0.1298,
+      "step": 8148
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.757719714964371e-05,
+      "loss": 0.1404,
+      "step": 8149
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.7571258907363425e-05,
+      "loss": 0.1388,
+      "step": 8150
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.756532066508313e-05,
+      "loss": 0.1787,
+      "step": 8151
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.755938242280285e-05,
+      "loss": 0.1269,
+      "step": 8152
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.7553444180522566e-05,
+      "loss": 0.1277,
+      "step": 8153
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.7547505938242286e-05,
+      "loss": 0.1222,
+      "step": 8154
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.754156769596199e-05,
+      "loss": 0.1089,
+      "step": 8155
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.7535629453681713e-05,
+      "loss": 0.0831,
+      "step": 8156
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.752969121140143e-05,
+      "loss": 0.1097,
+      "step": 8157
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.752375296912115e-05,
+      "loss": 0.135,
+      "step": 8158
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.7517814726840854e-05,
+      "loss": 0.13,
+      "step": 8159
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.751187648456057e-05,
+      "loss": 0.0906,
+      "step": 8160
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.750593824228029e-05,
+      "loss": 0.0887,
+      "step": 8161
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.7499999999999995e-05,
+      "loss": 0.1292,
+      "step": 8162
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.749406175771972e-05,
+      "loss": 0.0789,
+      "step": 8163
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.748812351543943e-05,
+      "loss": 0.1437,
+      "step": 8164
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.748218527315915e-05,
+      "loss": 0.0781,
+      "step": 8165
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.747624703087886e-05,
+      "loss": 0.0927,
+      "step": 8166
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.747030878859858e-05,
+      "loss": 0.1945,
+      "step": 8167
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.746437054631829e-05,
+      "loss": 0.1216,
+      "step": 8168
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.745843230403801e-05,
+      "loss": 0.1173,
+      "step": 8169
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.7452494061757724e-05,
+      "loss": 0.2474,
+      "step": 8170
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.744655581947743e-05,
+      "loss": 0.0924,
+      "step": 8171
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.744061757719715e-05,
+      "loss": 0.0755,
+      "step": 8172
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.7434679334916865e-05,
+      "loss": 0.0799,
+      "step": 8173
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.7428741092636585e-05,
+      "loss": 0.073,
+      "step": 8174
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.742280285035629e-05,
+      "loss": 0.0926,
+      "step": 8175
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.741686460807601e-05,
+      "loss": 0.1342,
+      "step": 8176
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.7410926365795726e-05,
+      "loss": 0.2883,
+      "step": 8177
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.7404988123515446e-05,
+      "loss": 0.1112,
+      "step": 8178
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.739904988123515e-05,
+      "loss": 0.446,
+      "step": 8179
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.739311163895488e-05,
+      "loss": 0.2034,
+      "step": 8180
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.738717339667459e-05,
+      "loss": 0.2277,
+      "step": 8181
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.73812351543943e-05,
+      "loss": 0.2146,
+      "step": 8182
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.737529691211402e-05,
+      "loss": 0.2375,
+      "step": 8183
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.736935866983373e-05,
+      "loss": 0.1469,
+      "step": 8184
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.736342042755345e-05,
+      "loss": 0.1431,
+      "step": 8185
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.735748218527316e-05,
+      "loss": 0.1431,
+      "step": 8186
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.735154394299288e-05,
+      "loss": 0.1921,
+      "step": 8187
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.734560570071259e-05,
+      "loss": 0.1795,
+      "step": 8188
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.733966745843231e-05,
+      "loss": 0.2416,
+      "step": 8189
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.733372921615202e-05,
+      "loss": 0.1793,
+      "step": 8190
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.732779097387174e-05,
+      "loss": 0.1882,
+      "step": 8191
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.732185273159145e-05,
+      "loss": 0.1248,
+      "step": 8192
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.7315914489311163e-05,
+      "loss": 0.1074,
+      "step": 8193
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.7309976247030884e-05,
+      "loss": 0.1287,
+      "step": 8194
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.730403800475059e-05,
+      "loss": 0.2591,
+      "step": 8195
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.729809976247031e-05,
+      "loss": 0.1324,
+      "step": 8196
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.7292161520190024e-05,
+      "loss": 0.1214,
+      "step": 8197
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.7286223277909745e-05,
+      "loss": 0.1255,
+      "step": 8198
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.728028503562946e-05,
+      "loss": 0.0871,
+      "step": 8199
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.727434679334918e-05,
+      "loss": 0.1621,
+      "step": 8200
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.7268408551068886e-05,
+      "loss": 0.0925,
+      "step": 8201
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.72624703087886e-05,
+      "loss": 0.1377,
+      "step": 8202
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.725653206650832e-05,
+      "loss": 0.066,
+      "step": 8203
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.7250593824228026e-05,
+      "loss": 0.0949,
+      "step": 8204
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.724465558194775e-05,
+      "loss": 0.1314,
+      "step": 8205
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.723871733966746e-05,
+      "loss": 0.104,
+      "step": 8206
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.723277909738718e-05,
+      "loss": 0.0923,
+      "step": 8207
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.722684085510689e-05,
+      "loss": 0.089,
+      "step": 8208
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.722090261282661e-05,
+      "loss": 0.1229,
+      "step": 8209
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.721496437054632e-05,
+      "loss": 0.0845,
+      "step": 8210
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.720902612826604e-05,
+      "loss": 0.1396,
+      "step": 8211
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.720308788598575e-05,
+      "loss": 0.1427,
+      "step": 8212
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.719714964370546e-05,
+      "loss": 0.0828,
+      "step": 8213
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.719121140142518e-05,
+      "loss": 0.1679,
+      "step": 8214
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.718527315914489e-05,
+      "loss": 0.1194,
+      "step": 8215
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.7179334916864616e-05,
+      "loss": 0.1,
+      "step": 8216
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.717339667458432e-05,
+      "loss": 0.1019,
+      "step": 8217
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.7167458432304044e-05,
+      "loss": 0.1892,
+      "step": 8218
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.716152019002376e-05,
+      "loss": 0.0973,
+      "step": 8219
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.715558194774348e-05,
+      "loss": 0.1717,
+      "step": 8220
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.7149643705463184e-05,
+      "loss": 0.0962,
+      "step": 8221
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.7143705463182905e-05,
+      "loss": 0.1844,
+      "step": 8222
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.713776722090262e-05,
+      "loss": 0.0701,
+      "step": 8223
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.7131828978622325e-05,
+      "loss": 0.0895,
+      "step": 8224
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.7125890736342045e-05,
+      "loss": 0.2121,
+      "step": 8225
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.711995249406176e-05,
+      "loss": 0.0858,
+      "step": 8226
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.711401425178148e-05,
+      "loss": 0.0368,
+      "step": 8227
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.7108076009501186e-05,
+      "loss": 0.1304,
+      "step": 8228
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.7102137767220907e-05,
+      "loss": 0.2324,
+      "step": 8229
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.709619952494062e-05,
+      "loss": 0.253,
+      "step": 8230
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.709026128266034e-05,
+      "loss": 0.2142,
+      "step": 8231
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.708432304038005e-05,
+      "loss": 0.2424,
+      "step": 8232
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.707838479809976e-05,
+      "loss": 0.1544,
+      "step": 8233
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.707244655581948e-05,
+      "loss": 0.1937,
+      "step": 8234
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.706650831353919e-05,
+      "loss": 0.1678,
+      "step": 8235
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.7060570071258915e-05,
+      "loss": 0.1312,
+      "step": 8236
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.705463182897862e-05,
+      "loss": 0.2397,
+      "step": 8237
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.704869358669834e-05,
+      "loss": 0.1325,
+      "step": 8238
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.7042755344418056e-05,
+      "loss": 0.2018,
+      "step": 8239
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.7036817102137776e-05,
+      "loss": 0.0967,
+      "step": 8240
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.703087885985748e-05,
+      "loss": 0.1211,
+      "step": 8241
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.7024940617577203e-05,
+      "loss": 0.1319,
+      "step": 8242
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.701900237529692e-05,
+      "loss": 0.1393,
+      "step": 8243
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.7013064133016624e-05,
+      "loss": 0.1889,
+      "step": 8244
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.7007125890736344e-05,
+      "loss": 0.1959,
+      "step": 8245
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.700118764845606e-05,
+      "loss": 0.13,
+      "step": 8246
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.699524940617578e-05,
+      "loss": 0.1319,
+      "step": 8247
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.6989311163895485e-05,
+      "loss": 0.1277,
+      "step": 8248
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.6983372921615205e-05,
+      "loss": 0.1554,
+      "step": 8249
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.697743467933492e-05,
+      "loss": 0.1051,
+      "step": 8250
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.697149643705464e-05,
+      "loss": 0.1454,
+      "step": 8251
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.6965558194774346e-05,
+      "loss": 0.1696,
+      "step": 8252
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.695961995249407e-05,
+      "loss": 0.1541,
+      "step": 8253
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.695368171021378e-05,
+      "loss": 0.2078,
+      "step": 8254
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.694774346793349e-05,
+      "loss": 0.2205,
+      "step": 8255
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.6941805225653214e-05,
+      "loss": 0.1525,
+      "step": 8256
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.693586698337292e-05,
+      "loss": 0.1327,
+      "step": 8257
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.692992874109264e-05,
+      "loss": 0.1421,
+      "step": 8258
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.6923990498812355e-05,
+      "loss": 0.1943,
+      "step": 8259
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.6918052256532075e-05,
+      "loss": 0.1296,
+      "step": 8260
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.691211401425178e-05,
+      "loss": 0.1246,
+      "step": 8261
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.69061757719715e-05,
+      "loss": 0.0949,
+      "step": 8262
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.6900237529691216e-05,
+      "loss": 0.1076,
+      "step": 8263
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.689429928741092e-05,
+      "loss": 0.1037,
+      "step": 8264
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.688836104513064e-05,
+      "loss": 0.1014,
+      "step": 8265
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.6882422802850356e-05,
+      "loss": 0.157,
+      "step": 8266
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.687648456057008e-05,
+      "loss": 0.0586,
+      "step": 8267
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.6870546318289784e-05,
+      "loss": 0.0876,
+      "step": 8268
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.6864608076009504e-05,
+      "loss": 0.1783,
+      "step": 8269
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.685866983372922e-05,
+      "loss": 0.1093,
+      "step": 8270
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.685273159144894e-05,
+      "loss": 0.1141,
+      "step": 8271
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.6846793349168645e-05,
+      "loss": 0.2118,
+      "step": 8272
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.684085510688837e-05,
+      "loss": 0.1224,
+      "step": 8273
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.683491686460808e-05,
+      "loss": 0.0761,
+      "step": 8274
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.6828978622327785e-05,
+      "loss": 0.0983,
+      "step": 8275
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.682304038004751e-05,
+      "loss": 0.1223,
+      "step": 8276
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.681710213776722e-05,
+      "loss": 0.132,
+      "step": 8277
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.681116389548694e-05,
+      "loss": 0.111,
+      "step": 8278
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.680522565320665e-05,
+      "loss": 0.3211,
+      "step": 8279
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.6799287410926374e-05,
+      "loss": 0.2223,
+      "step": 8280
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.679334916864608e-05,
+      "loss": 0.191,
+      "step": 8281
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.67874109263658e-05,
+      "loss": 0.2084,
+      "step": 8282
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.6781472684085514e-05,
+      "loss": 0.1784,
+      "step": 8283
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.6775534441805235e-05,
+      "loss": 0.1739,
+      "step": 8284
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.676959619952494e-05,
+      "loss": 0.1867,
+      "step": 8285
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.6763657957244655e-05,
+      "loss": 0.0992,
+      "step": 8286
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.6757719714964376e-05,
+      "loss": 0.282,
+      "step": 8287
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.675178147268408e-05,
+      "loss": 0.1098,
+      "step": 8288
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.67458432304038e-05,
+      "loss": 0.2197,
+      "step": 8289
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.6739904988123516e-05,
+      "loss": 0.2178,
+      "step": 8290
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.6733966745843237e-05,
+      "loss": 0.1343,
+      "step": 8291
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.6728028503562943e-05,
+      "loss": 0.1245,
+      "step": 8292
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.672209026128267e-05,
+      "loss": 0.1984,
+      "step": 8293
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.671615201900238e-05,
+      "loss": 0.1477,
+      "step": 8294
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.671021377672209e-05,
+      "loss": 0.1014,
+      "step": 8295
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.670427553444181e-05,
+      "loss": 0.1416,
+      "step": 8296
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.669833729216152e-05,
+      "loss": 0.1926,
+      "step": 8297
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.669239904988124e-05,
+      "loss": 0.1158,
+      "step": 8298
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.668646080760095e-05,
+      "loss": 0.1199,
+      "step": 8299
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.668052256532067e-05,
+      "loss": 0.1386,
+      "step": 8300
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.667458432304038e-05,
+      "loss": 0.2204,
+      "step": 8301
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.66686460807601e-05,
+      "loss": 0.2927,
+      "step": 8302
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.666270783847981e-05,
+      "loss": 0.1379,
+      "step": 8303
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.6656769596199533e-05,
+      "loss": 0.1322,
+      "step": 8304
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.665083135391924e-05,
+      "loss": 0.1497,
+      "step": 8305
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.6644893111638954e-05,
+      "loss": 0.1245,
+      "step": 8306
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.6638954869358674e-05,
+      "loss": 0.121,
+      "step": 8307
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.663301662707838e-05,
+      "loss": 0.1552,
+      "step": 8308
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.66270783847981e-05,
+      "loss": 0.1514,
+      "step": 8309
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.6621140142517815e-05,
+      "loss": 0.1052,
+      "step": 8310
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.6615201900237535e-05,
+      "loss": 0.0609,
+      "step": 8311
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.660926365795724e-05,
+      "loss": 0.1101,
+      "step": 8312
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.660332541567697e-05,
+      "loss": 0.3846,
+      "step": 8313
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.6597387173396676e-05,
+      "loss": 0.576,
+      "step": 8314
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.6591448931116396e-05,
+      "loss": 0.0892,
+      "step": 8315
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.658551068883611e-05,
+      "loss": 0.1331,
+      "step": 8316
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.657957244655582e-05,
+      "loss": 0.0616,
+      "step": 8317
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.657363420427554e-05,
+      "loss": 0.0827,
+      "step": 8318
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.656769596199525e-05,
+      "loss": 0.0921,
+      "step": 8319
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.656175771971497e-05,
+      "loss": 0.1718,
+      "step": 8320
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.655581947743468e-05,
+      "loss": 0.1038,
+      "step": 8321
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.65498812351544e-05,
+      "loss": 0.1028,
+      "step": 8322
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.654394299287411e-05,
+      "loss": 0.0777,
+      "step": 8323
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.653800475059383e-05,
+      "loss": 0.0979,
+      "step": 8324
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.653206650831354e-05,
+      "loss": 0.1038,
+      "step": 8325
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.652612826603325e-05,
+      "loss": 0.0825,
+      "step": 8326
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.652019002375297e-05,
+      "loss": 0.0862,
+      "step": 8327
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.651425178147268e-05,
+      "loss": 0.1995,
+      "step": 8328
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.65083135391924e-05,
+      "loss": 0.242,
+      "step": 8329
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.6502375296912114e-05,
+      "loss": 0.2159,
+      "step": 8330
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.6496437054631834e-05,
+      "loss": 0.2344,
+      "step": 8331
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.649049881235155e-05,
+      "loss": 0.2385,
+      "step": 8332
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.648456057007127e-05,
+      "loss": 0.369,
+      "step": 8333
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.6478622327790975e-05,
+      "loss": 0.1369,
+      "step": 8334
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.6472684085510695e-05,
+      "loss": 0.2088,
+      "step": 8335
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.646674584323041e-05,
+      "loss": 0.1297,
+      "step": 8336
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.6460807600950116e-05,
+      "loss": 0.2051,
+      "step": 8337
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.6454869358669836e-05,
+      "loss": 0.229,
+      "step": 8338
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.644893111638955e-05,
+      "loss": 0.1634,
+      "step": 8339
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.644299287410927e-05,
+      "loss": 0.1402,
+      "step": 8340
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.643705463182898e-05,
+      "loss": 0.1787,
+      "step": 8341
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.64311163895487e-05,
+      "loss": 0.1906,
+      "step": 8342
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.642517814726841e-05,
+      "loss": 0.1148,
+      "step": 8343
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.641923990498813e-05,
+      "loss": 0.1969,
+      "step": 8344
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.641330166270784e-05,
+      "loss": 0.1566,
+      "step": 8345
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.640736342042756e-05,
+      "loss": 0.131,
+      "step": 8346
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.640142517814727e-05,
+      "loss": 0.1753,
+      "step": 8347
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.639548693586698e-05,
+      "loss": 0.1287,
+      "step": 8348
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.6389548693586706e-05,
+      "loss": 0.0785,
+      "step": 8349
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.638361045130641e-05,
+      "loss": 0.1428,
+      "step": 8350
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.637767220902613e-05,
+      "loss": 0.1466,
+      "step": 8351
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.6371733966745846e-05,
+      "loss": 0.3048,
+      "step": 8352
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.636579572446557e-05,
+      "loss": 0.1396,
+      "step": 8353
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.6359857482185274e-05,
+      "loss": 0.1801,
+      "step": 8354
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.6353919239904994e-05,
+      "loss": 0.1317,
+      "step": 8355
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.634798099762471e-05,
+      "loss": 0.1419,
+      "step": 8356
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.6342042755344414e-05,
+      "loss": 0.1353,
+      "step": 8357
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.6336104513064135e-05,
+      "loss": 0.1236,
+      "step": 8358
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.633016627078385e-05,
+      "loss": 0.1733,
+      "step": 8359
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.632422802850357e-05,
+      "loss": 0.1587,
+      "step": 8360
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.6318289786223275e-05,
+      "loss": 0.1593,
+      "step": 8361
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.6312351543942996e-05,
+      "loss": 0.1546,
+      "step": 8362
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.630641330166271e-05,
+      "loss": 0.1414,
+      "step": 8363
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.630047505938243e-05,
+      "loss": 0.1603,
+      "step": 8364
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.6294536817102136e-05,
+      "loss": 0.1055,
+      "step": 8365
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.6288598574821864e-05,
+      "loss": 0.1163,
+      "step": 8366
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.628266033254157e-05,
+      "loss": 0.0867,
+      "step": 8367
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.627672209026128e-05,
+      "loss": 0.1322,
+      "step": 8368
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.6270783847981004e-05,
+      "loss": 0.1186,
+      "step": 8369
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.626484560570071e-05,
+      "loss": 0.1533,
+      "step": 8370
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.625890736342043e-05,
+      "loss": 0.0818,
+      "step": 8371
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.6252969121140145e-05,
+      "loss": 0.0742,
+      "step": 8372
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.6247030878859865e-05,
+      "loss": 0.1405,
+      "step": 8373
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.624109263657957e-05,
+      "loss": 0.1047,
+      "step": 8374
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.623515439429929e-05,
+      "loss": 0.0761,
+      "step": 8375
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.6229216152019006e-05,
+      "loss": 0.0651,
+      "step": 8376
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.6223277909738727e-05,
+      "loss": 0.0707,
+      "step": 8377
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.621733966745843e-05,
+      "loss": 0.2251,
+      "step": 8378
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.621140142517815e-05,
+      "loss": 0.2851,
+      "step": 8379
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.620546318289787e-05,
+      "loss": 0.2561,
+      "step": 8380
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.6199524940617574e-05,
+      "loss": 0.1805,
+      "step": 8381
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.6193586698337294e-05,
+      "loss": 0.1932,
+      "step": 8382
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.618764845605701e-05,
+      "loss": 0.1631,
+      "step": 8383
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.618171021377673e-05,
+      "loss": 0.1369,
+      "step": 8384
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.6175771971496435e-05,
+      "loss": 0.1627,
+      "step": 8385
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.616983372921616e-05,
+      "loss": 0.1392,
+      "step": 8386
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.616389548693587e-05,
+      "loss": 0.1243,
+      "step": 8387
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.615795724465559e-05,
+      "loss": 0.1425,
+      "step": 8388
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.61520190023753e-05,
+      "loss": 0.1597,
+      "step": 8389
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.614608076009501e-05,
+      "loss": 0.1973,
+      "step": 8390
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.614014251781473e-05,
+      "loss": 0.2792,
+      "step": 8391
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.6134204275534444e-05,
+      "loss": 0.2393,
+      "step": 8392
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.6128266033254164e-05,
+      "loss": 0.1502,
+      "step": 8393
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.612232779097387e-05,
+      "loss": 0.1726,
+      "step": 8394
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.611638954869359e-05,
+      "loss": 0.1477,
+      "step": 8395
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.6110451306413305e-05,
+      "loss": 0.1517,
+      "step": 8396
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.6104513064133025e-05,
+      "loss": 0.1315,
+      "step": 8397
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.609857482185273e-05,
+      "loss": 0.1259,
+      "step": 8398
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.6092636579572446e-05,
+      "loss": 0.1406,
+      "step": 8399
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.6086698337292166e-05,
+      "loss": 0.1551,
+      "step": 8400
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.608076009501187e-05,
+      "loss": 0.135,
+      "step": 8401
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.607482185273159e-05,
+      "loss": 0.1245,
+      "step": 8402
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.606888361045131e-05,
+      "loss": 0.1623,
+      "step": 8403
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.606294536817103e-05,
+      "loss": 0.1429,
+      "step": 8404
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.6057007125890734e-05,
+      "loss": 0.1511,
+      "step": 8405
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.605106888361046e-05,
+      "loss": 0.1159,
+      "step": 8406
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.604513064133017e-05,
+      "loss": 0.1505,
+      "step": 8407
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.603919239904989e-05,
+      "loss": 0.1116,
+      "step": 8408
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.60332541567696e-05,
+      "loss": 0.1327,
+      "step": 8409
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.602731591448931e-05,
+      "loss": 0.1121,
+      "step": 8410
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.602137767220903e-05,
+      "loss": 0.1819,
+      "step": 8411
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.601543942992874e-05,
+      "loss": 0.1362,
+      "step": 8412
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.600950118764846e-05,
+      "loss": 0.0862,
+      "step": 8413
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.600356294536817e-05,
+      "loss": 0.0953,
+      "step": 8414
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.599762470308789e-05,
+      "loss": 0.094,
+      "step": 8415
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.5991686460807604e-05,
+      "loss": 0.1523,
+      "step": 8416
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.5985748218527324e-05,
+      "loss": 0.0887,
+      "step": 8417
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.597980997624703e-05,
+      "loss": 0.0854,
+      "step": 8418
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.597387173396675e-05,
+      "loss": 0.1566,
+      "step": 8419
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.5967933491686465e-05,
+      "loss": 0.1787,
+      "step": 8420
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.596199524940617e-05,
+      "loss": 0.1156,
+      "step": 8421
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.595605700712589e-05,
+      "loss": 0.0981,
+      "step": 8422
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.5950118764845605e-05,
+      "loss": 0.1201,
+      "step": 8423
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.5944180522565326e-05,
+      "loss": 0.0832,
+      "step": 8424
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.593824228028503e-05,
+      "loss": 0.0709,
+      "step": 8425
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.593230403800476e-05,
+      "loss": 0.073,
+      "step": 8426
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.5926365795724467e-05,
+      "loss": 0.1494,
+      "step": 8427
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.592042755344419e-05,
+      "loss": 0.1532,
+      "step": 8428
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.59144893111639e-05,
+      "loss": 0.3422,
+      "step": 8429
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.590855106888361e-05,
+      "loss": 0.2497,
+      "step": 8430
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.590261282660333e-05,
+      "loss": 0.1862,
+      "step": 8431
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.589667458432304e-05,
+      "loss": 0.1741,
+      "step": 8432
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.589073634204276e-05,
+      "loss": 0.1708,
+      "step": 8433
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.588479809976247e-05,
+      "loss": 0.1667,
+      "step": 8434
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.587885985748219e-05,
+      "loss": 0.1245,
+      "step": 8435
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.58729216152019e-05,
+      "loss": 0.1926,
+      "step": 8436
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.586698337292162e-05,
+      "loss": 0.1509,
+      "step": 8437
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.586104513064133e-05,
+      "loss": 0.1544,
+      "step": 8438
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.585510688836105e-05,
+      "loss": 0.171,
+      "step": 8439
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.5849168646080763e-05,
+      "loss": 0.1284,
+      "step": 8440
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.584323040380047e-05,
+      "loss": 0.1231,
+      "step": 8441
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.583729216152019e-05,
+      "loss": 0.2123,
+      "step": 8442
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.5831353919239904e-05,
+      "loss": 0.1562,
+      "step": 8443
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.5825415676959625e-05,
+      "loss": 0.1185,
+      "step": 8444
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.581947743467934e-05,
+      "loss": 0.1179,
+      "step": 8445
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.581353919239906e-05,
+      "loss": 0.1727,
+      "step": 8446
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.5807600950118765e-05,
+      "loss": 0.0962,
+      "step": 8447
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.5801662707838486e-05,
+      "loss": 0.2157,
+      "step": 8448
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.57957244655582e-05,
+      "loss": 0.1595,
+      "step": 8449
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.578978622327792e-05,
+      "loss": 0.1397,
+      "step": 8450
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.5783847980997626e-05,
+      "loss": 0.1604,
+      "step": 8451
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.577790973871734e-05,
+      "loss": 0.1314,
+      "step": 8452
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.577197149643706e-05,
+      "loss": 0.1347,
+      "step": 8453
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.576603325415677e-05,
+      "loss": 0.1433,
+      "step": 8454
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.576009501187649e-05,
+      "loss": 0.0932,
+      "step": 8455
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.57541567695962e-05,
+      "loss": 0.0903,
+      "step": 8456
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.574821852731592e-05,
+      "loss": 0.1315,
+      "step": 8457
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.574228028503563e-05,
+      "loss": 0.1435,
+      "step": 8458
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.573634204275535e-05,
+      "loss": 0.1835,
+      "step": 8459
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.573040380047506e-05,
+      "loss": 0.1667,
+      "step": 8460
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.572446555819477e-05,
+      "loss": 0.0959,
+      "step": 8461
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.5718527315914496e-05,
+      "loss": 0.1191,
+      "step": 8462
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.57125890736342e-05,
+      "loss": 0.1182,
+      "step": 8463
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.570665083135392e-05,
+      "loss": 0.1452,
+      "step": 8464
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.570071258907364e-05,
+      "loss": 0.0803,
+      "step": 8465
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.569477434679336e-05,
+      "loss": 0.0771,
+      "step": 8466
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.5688836104513064e-05,
+      "loss": 0.1068,
+      "step": 8467
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.5682897862232784e-05,
+      "loss": 0.1197,
+      "step": 8468
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.56769596199525e-05,
+      "loss": 0.0885,
+      "step": 8469
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.567102137767222e-05,
+      "loss": 0.1146,
+      "step": 8470
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.5665083135391925e-05,
+      "loss": 0.1013,
+      "step": 8471
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.565914489311164e-05,
+      "loss": 0.1158,
+      "step": 8472
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.565320665083136e-05,
+      "loss": 0.0603,
+      "step": 8473
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.5647268408551066e-05,
+      "loss": 0.0895,
+      "step": 8474
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.5641330166270786e-05,
+      "loss": 0.7638,
+      "step": 8475
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.56353919239905e-05,
+      "loss": 0.0785,
+      "step": 8476
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.562945368171022e-05,
+      "loss": 0.1494,
+      "step": 8477
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.562351543942993e-05,
+      "loss": 0.1786,
+      "step": 8478
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.561757719714965e-05,
+      "loss": 0.3318,
+      "step": 8479
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.561163895486936e-05,
+      "loss": 0.2305,
+      "step": 8480
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.560570071258908e-05,
+      "loss": 0.2473,
+      "step": 8481
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.5599762470308795e-05,
+      "loss": 0.2391,
+      "step": 8482
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.55938242280285e-05,
+      "loss": 0.1917,
+      "step": 8483
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.558788598574822e-05,
+      "loss": 0.1799,
+      "step": 8484
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.5581947743467936e-05,
+      "loss": 0.1955,
+      "step": 8485
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.5576009501187656e-05,
+      "loss": 0.2155,
+      "step": 8486
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.557007125890736e-05,
+      "loss": 0.117,
+      "step": 8487
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.556413301662708e-05,
+      "loss": 0.1809,
+      "step": 8488
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.55581947743468e-05,
+      "loss": 0.1732,
+      "step": 8489
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.555225653206652e-05,
+      "loss": 0.1705,
+      "step": 8490
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.5546318289786224e-05,
+      "loss": 0.1511,
+      "step": 8491
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.554038004750594e-05,
+      "loss": 0.1257,
+      "step": 8492
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.553444180522566e-05,
+      "loss": 0.1415,
+      "step": 8493
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.5528503562945365e-05,
+      "loss": 0.1515,
+      "step": 8494
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.5522565320665085e-05,
+      "loss": 0.1099,
+      "step": 8495
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.55166270783848e-05,
+      "loss": 0.1307,
+      "step": 8496
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.551068883610452e-05,
+      "loss": 0.1158,
+      "step": 8497
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.5504750593824226e-05,
+      "loss": 0.1776,
+      "step": 8498
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.549881235154395e-05,
+      "loss": 0.1648,
+      "step": 8499
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.549287410926366e-05,
+      "loss": 0.1018,
+      "step": 8500
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.548693586698338e-05,
+      "loss": 0.1862,
+      "step": 8501
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.5480997624703094e-05,
+      "loss": 0.1372,
+      "step": 8502
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.54750593824228e-05,
+      "loss": 0.1287,
+      "step": 8503
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.546912114014252e-05,
+      "loss": 0.148,
+      "step": 8504
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.5463182897862234e-05,
+      "loss": 0.1012,
+      "step": 8505
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.5457244655581955e-05,
+      "loss": 0.1171,
+      "step": 8506
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.545130641330166e-05,
+      "loss": 0.1417,
+      "step": 8507
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.544536817102138e-05,
+      "loss": 0.1645,
+      "step": 8508
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.5439429928741095e-05,
+      "loss": 0.1007,
+      "step": 8509
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.5433491686460816e-05,
+      "loss": 0.0937,
+      "step": 8510
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.542755344418052e-05,
+      "loss": 0.0821,
+      "step": 8511
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.542161520190024e-05,
+      "loss": 0.1317,
+      "step": 8512
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.5415676959619956e-05,
+      "loss": 0.0895,
+      "step": 8513
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.540973871733966e-05,
+      "loss": 0.0996,
+      "step": 8514
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.5403800475059384e-05,
+      "loss": 0.1182,
+      "step": 8515
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.53978622327791e-05,
+      "loss": 0.1312,
+      "step": 8516
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.539192399049882e-05,
+      "loss": 0.1077,
+      "step": 8517
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.5385985748218524e-05,
+      "loss": 0.106,
+      "step": 8518
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.538004750593825e-05,
+      "loss": 0.0952,
+      "step": 8519
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.537410926365796e-05,
+      "loss": 0.1206,
+      "step": 8520
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.536817102137768e-05,
+      "loss": 0.2178,
+      "step": 8521
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.536223277909739e-05,
+      "loss": 0.0751,
+      "step": 8522
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.53562945368171e-05,
+      "loss": 0.1455,
+      "step": 8523
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.535035629453682e-05,
+      "loss": 0.1399,
+      "step": 8524
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.534441805225653e-05,
+      "loss": 0.0561,
+      "step": 8525
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.533847980997625e-05,
+      "loss": 0.0686,
+      "step": 8526
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.533254156769596e-05,
+      "loss": 0.0716,
+      "step": 8527
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.532660332541568e-05,
+      "loss": 0.1264,
+      "step": 8528
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.5320665083135394e-05,
+      "loss": 0.1842,
+      "step": 8529
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.5314726840855114e-05,
+      "loss": 0.2021,
+      "step": 8530
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.530878859857482e-05,
+      "loss": 0.2216,
+      "step": 8531
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.530285035629454e-05,
+      "loss": 0.2095,
+      "step": 8532
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.5296912114014255e-05,
+      "loss": 0.1527,
+      "step": 8533
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.529097387173396e-05,
+      "loss": 0.1353,
+      "step": 8534
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.528503562945368e-05,
+      "loss": 0.1516,
+      "step": 8535
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.5279097387173396e-05,
+      "loss": 0.1723,
+      "step": 8536
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.5273159144893116e-05,
+      "loss": 0.1775,
+      "step": 8537
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.526722090261282e-05,
+      "loss": 0.1951,
+      "step": 8538
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.526128266033255e-05,
+      "loss": 0.1162,
+      "step": 8539
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.525534441805226e-05,
+      "loss": 0.206,
+      "step": 8540
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.524940617577198e-05,
+      "loss": 0.1807,
+      "step": 8541
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.524346793349169e-05,
+      "loss": 0.125,
+      "step": 8542
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.523752969121141e-05,
+      "loss": 0.1312,
+      "step": 8543
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.523159144893112e-05,
+      "loss": 0.1624,
+      "step": 8544
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.522565320665083e-05,
+      "loss": 0.1709,
+      "step": 8545
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.521971496437055e-05,
+      "loss": 0.1545,
+      "step": 8546
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.521377672209026e-05,
+      "loss": 0.1441,
+      "step": 8547
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.520783847980998e-05,
+      "loss": 0.2142,
+      "step": 8548
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.520190023752969e-05,
+      "loss": 0.1453,
+      "step": 8549
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.519596199524941e-05,
+      "loss": 0.1093,
+      "step": 8550
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.519002375296912e-05,
+      "loss": 0.0957,
+      "step": 8551
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.518408551068884e-05,
+      "loss": 0.1256,
+      "step": 8552
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.5178147268408554e-05,
+      "loss": 0.112,
+      "step": 8553
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.517220902612826e-05,
+      "loss": 0.1144,
+      "step": 8554
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.516627078384798e-05,
+      "loss": 0.1422,
+      "step": 8555
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.5160332541567695e-05,
+      "loss": 0.1569,
+      "step": 8556
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.5154394299287415e-05,
+      "loss": 0.1424,
+      "step": 8557
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.514845605700712e-05,
+      "loss": 0.1428,
+      "step": 8558
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.514251781472685e-05,
+      "loss": 0.1258,
+      "step": 8559
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.5136579572446556e-05,
+      "loss": 0.1634,
+      "step": 8560
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.5130641330166276e-05,
+      "loss": 0.097,
+      "step": 8561
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.512470308788599e-05,
+      "loss": 0.1496,
+      "step": 8562
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.511876484560571e-05,
+      "loss": 0.1291,
+      "step": 8563
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.511282660332542e-05,
+      "loss": 0.1064,
+      "step": 8564
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.510688836104513e-05,
+      "loss": 0.1125,
+      "step": 8565
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.510095011876485e-05,
+      "loss": 0.1227,
+      "step": 8566
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.509501187648456e-05,
+      "loss": 0.1532,
+      "step": 8567
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.508907363420428e-05,
+      "loss": 0.1151,
+      "step": 8568
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.508313539192399e-05,
+      "loss": 0.1445,
+      "step": 8569
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.507719714964371e-05,
+      "loss": 0.1382,
+      "step": 8570
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.507125890736342e-05,
+      "loss": 0.106,
+      "step": 8571
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.506532066508314e-05,
+      "loss": 0.1166,
+      "step": 8572
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.505938242280285e-05,
+      "loss": 0.1346,
+      "step": 8573
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.505344418052257e-05,
+      "loss": 0.1436,
+      "step": 8574
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.504750593824228e-05,
+      "loss": 0.1003,
+      "step": 8575
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.504156769596199e-05,
+      "loss": 0.0399,
+      "step": 8576
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.5035629453681714e-05,
+      "loss": 0.0589,
+      "step": 8577
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.502969121140143e-05,
+      "loss": 0.1286,
+      "step": 8578
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.502375296912115e-05,
+      "loss": 0.2387,
+      "step": 8579
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.5017814726840854e-05,
+      "loss": 0.2361,
+      "step": 8580
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.5011876484560575e-05,
+      "loss": 0.2881,
+      "step": 8581
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.500593824228029e-05,
+      "loss": 0.2258,
+      "step": 8582
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 0.1685,
+      "step": 8583
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.4994061757719716e-05,
+      "loss": 0.2236,
+      "step": 8584
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.4988123515439436e-05,
+      "loss": 0.1355,
+      "step": 8585
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.498218527315915e-05,
+      "loss": 0.1224,
+      "step": 8586
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.4976247030878856e-05,
+      "loss": 0.207,
+      "step": 8587
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.497030878859858e-05,
+      "loss": 0.141,
+      "step": 8588
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.496437054631829e-05,
+      "loss": 0.2,
+      "step": 8589
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.495843230403801e-05,
+      "loss": 0.2231,
+      "step": 8590
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.495249406175772e-05,
+      "loss": 0.1316,
+      "step": 8591
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.494655581947744e-05,
+      "loss": 0.1659,
+      "step": 8592
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.494061757719715e-05,
+      "loss": 0.097,
+      "step": 8593
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.493467933491687e-05,
+      "loss": 0.2331,
+      "step": 8594
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.4928741092636585e-05,
+      "loss": 0.1208,
+      "step": 8595
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.492280285035629e-05,
+      "loss": 0.171,
+      "step": 8596
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.491686460807601e-05,
+      "loss": 0.144,
+      "step": 8597
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.4910926365795726e-05,
+      "loss": 0.0658,
+      "step": 8598
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.4904988123515446e-05,
+      "loss": 0.1411,
+      "step": 8599
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.489904988123515e-05,
+      "loss": 0.1366,
+      "step": 8600
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.4893111638954874e-05,
+      "loss": 0.1116,
+      "step": 8601
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.488717339667459e-05,
+      "loss": 0.1418,
+      "step": 8602
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.488123515439431e-05,
+      "loss": 0.0869,
+      "step": 8603
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.4875296912114014e-05,
+      "loss": 0.165,
+      "step": 8604
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.4869358669833735e-05,
+      "loss": 0.1294,
+      "step": 8605
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.486342042755345e-05,
+      "loss": 0.1157,
+      "step": 8606
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.4857482185273155e-05,
+      "loss": 0.1043,
+      "step": 8607
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.4851543942992875e-05,
+      "loss": 0.1094,
+      "step": 8608
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.484560570071259e-05,
+      "loss": 0.0899,
+      "step": 8609
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.483966745843231e-05,
+      "loss": 0.1088,
+      "step": 8610
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.4833729216152016e-05,
+      "loss": 0.1393,
+      "step": 8611
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.482779097387174e-05,
+      "loss": 0.148,
+      "step": 8612
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.482185273159145e-05,
+      "loss": 0.3382,
+      "step": 8613
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.481591448931117e-05,
+      "loss": 0.0854,
+      "step": 8614
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.4809976247030884e-05,
+      "loss": 0.1363,
+      "step": 8615
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.4804038004750604e-05,
+      "loss": 0.1057,
+      "step": 8616
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.479809976247031e-05,
+      "loss": 0.0898,
+      "step": 8617
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.4792161520190025e-05,
+      "loss": 0.1311,
+      "step": 8618
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.4786223277909745e-05,
+      "loss": 0.108,
+      "step": 8619
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.478028503562945e-05,
+      "loss": 0.1026,
+      "step": 8620
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.477434679334917e-05,
+      "loss": 0.1521,
+      "step": 8621
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.4768408551068886e-05,
+      "loss": 0.0594,
+      "step": 8622
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.4762470308788606e-05,
+      "loss": 0.0696,
+      "step": 8623
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.475653206650831e-05,
+      "loss": 0.0862,
+      "step": 8624
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.475059382422803e-05,
+      "loss": 0.0686,
+      "step": 8625
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.474465558194775e-05,
+      "loss": 0.0615,
+      "step": 8626
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.4738717339667454e-05,
+      "loss": 0.5734,
+      "step": 8627
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.4732779097387174e-05,
+      "loss": 0.1868,
+      "step": 8628
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.472684085510689e-05,
+      "loss": 0.2801,
+      "step": 8629
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.472090261282661e-05,
+      "loss": 0.212,
+      "step": 8630
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.4714964370546315e-05,
+      "loss": 0.2734,
+      "step": 8631
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.470902612826604e-05,
+      "loss": 0.1512,
+      "step": 8632
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.470308788598575e-05,
+      "loss": 0.1314,
+      "step": 8633
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.469714964370547e-05,
+      "loss": 0.1479,
+      "step": 8634
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.469121140142518e-05,
+      "loss": 0.2219,
+      "step": 8635
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.46852731591449e-05,
+      "loss": 0.115,
+      "step": 8636
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.467933491686461e-05,
+      "loss": 0.1171,
+      "step": 8637
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.4673396674584323e-05,
+      "loss": 0.1552,
+      "step": 8638
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.4667458432304044e-05,
+      "loss": 0.1865,
+      "step": 8639
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.466152019002375e-05,
+      "loss": 0.329,
+      "step": 8640
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.465558194774347e-05,
+      "loss": 0.1463,
+      "step": 8641
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.4649643705463185e-05,
+      "loss": 0.1391,
+      "step": 8642
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.4643705463182905e-05,
+      "loss": 0.1475,
+      "step": 8643
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.463776722090261e-05,
+      "loss": 0.133,
+      "step": 8644
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.463182897862233e-05,
+      "loss": 0.1791,
+      "step": 8645
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.4625890736342046e-05,
+      "loss": 0.0836,
+      "step": 8646
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.4619952494061766e-05,
+      "loss": 0.1152,
+      "step": 8647
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.461401425178147e-05,
+      "loss": 0.1347,
+      "step": 8648
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.4608076009501186e-05,
+      "loss": 0.1088,
+      "step": 8649
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.460213776722091e-05,
+      "loss": 0.1775,
+      "step": 8650
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.4596199524940614e-05,
+      "loss": 0.1491,
+      "step": 8651
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.459026128266034e-05,
+      "loss": 0.1132,
+      "step": 8652
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.458432304038005e-05,
+      "loss": 0.1185,
+      "step": 8653
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.457838479809977e-05,
+      "loss": 0.1173,
+      "step": 8654
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.457244655581948e-05,
+      "loss": 0.3059,
+      "step": 8655
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.45665083135392e-05,
+      "loss": 0.1054,
+      "step": 8656
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.456057007125891e-05,
+      "loss": 0.1126,
+      "step": 8657
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.455463182897862e-05,
+      "loss": 0.1336,
+      "step": 8658
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.454869358669834e-05,
+      "loss": 0.1251,
+      "step": 8659
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.454275534441805e-05,
+      "loss": 0.1386,
+      "step": 8660
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.453681710213777e-05,
+      "loss": 0.1274,
+      "step": 8661
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.453087885985748e-05,
+      "loss": 0.0794,
+      "step": 8662
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.4524940617577204e-05,
+      "loss": 0.2592,
+      "step": 8663
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.451900237529691e-05,
+      "loss": 0.0795,
+      "step": 8664
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.451306413301663e-05,
+      "loss": 0.1137,
+      "step": 8665
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.4507125890736344e-05,
+      "loss": 0.0998,
+      "step": 8666
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.4501187648456065e-05,
+      "loss": 0.082,
+      "step": 8667
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.449524940617577e-05,
+      "loss": 0.2769,
+      "step": 8668
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.4489311163895485e-05,
+      "loss": 0.1364,
+      "step": 8669
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.4483372921615205e-05,
+      "loss": 0.1363,
+      "step": 8670
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.447743467933491e-05,
+      "loss": 0.0553,
+      "step": 8671
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.447149643705464e-05,
+      "loss": 0.1122,
+      "step": 8672
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.4465558194774346e-05,
+      "loss": 0.1056,
+      "step": 8673
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.4459619952494067e-05,
+      "loss": 0.1479,
+      "step": 8674
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.445368171021378e-05,
+      "loss": 0.0373,
+      "step": 8675
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.44477434679335e-05,
+      "loss": 0.1482,
+      "step": 8676
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.444180522565321e-05,
+      "loss": 0.1078,
+      "step": 8677
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.443586698337293e-05,
+      "loss": 0.1289,
+      "step": 8678
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.442992874109264e-05,
+      "loss": 0.2839,
+      "step": 8679
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.442399049881235e-05,
+      "loss": 0.2383,
+      "step": 8680
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.441805225653207e-05,
+      "loss": 0.38,
+      "step": 8681
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.441211401425178e-05,
+      "loss": 0.1934,
+      "step": 8682
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.44061757719715e-05,
+      "loss": 0.1633,
+      "step": 8683
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.440023752969121e-05,
+      "loss": 0.1376,
+      "step": 8684
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.439429928741093e-05,
+      "loss": 0.1269,
+      "step": 8685
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.438836104513064e-05,
+      "loss": 0.222,
+      "step": 8686
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.4382422802850363e-05,
+      "loss": 0.1538,
+      "step": 8687
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.437648456057007e-05,
+      "loss": 0.1946,
+      "step": 8688
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.4370546318289784e-05,
+      "loss": 0.2699,
+      "step": 8689
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.4364608076009504e-05,
+      "loss": 0.1789,
+      "step": 8690
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.435866983372922e-05,
+      "loss": 0.1637,
+      "step": 8691
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.435273159144894e-05,
+      "loss": 0.1492,
+      "step": 8692
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.4346793349168645e-05,
+      "loss": 0.1489,
+      "step": 8693
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.4340855106888365e-05,
+      "loss": 0.1189,
+      "step": 8694
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.433491686460808e-05,
+      "loss": 0.1461,
+      "step": 8695
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.43289786223278e-05,
+      "loss": 0.1476,
+      "step": 8696
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.4323040380047506e-05,
+      "loss": 0.2517,
+      "step": 8697
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.4317102137767226e-05,
+      "loss": 0.1323,
+      "step": 8698
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.431116389548694e-05,
+      "loss": 0.1981,
+      "step": 8699
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.430522565320665e-05,
+      "loss": 0.1203,
+      "step": 8700
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.429928741092637e-05,
+      "loss": 0.2429,
+      "step": 8701
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.429334916864608e-05,
+      "loss": 0.1178,
+      "step": 8702
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.42874109263658e-05,
+      "loss": 0.1613,
+      "step": 8703
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.428147268408551e-05,
+      "loss": 0.1298,
+      "step": 8704
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.427553444180523e-05,
+      "loss": 0.1151,
+      "step": 8705
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.426959619952494e-05,
+      "loss": 0.1286,
+      "step": 8706
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.426365795724466e-05,
+      "loss": 0.1214,
+      "step": 8707
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.4257719714964376e-05,
+      "loss": 0.3242,
+      "step": 8708
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.4251781472684096e-05,
+      "loss": 0.1418,
+      "step": 8709
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.42458432304038e-05,
+      "loss": 0.0743,
+      "step": 8710
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.4239904988123516e-05,
+      "loss": 0.0959,
+      "step": 8711
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.423396674584324e-05,
+      "loss": 0.1022,
+      "step": 8712
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.4228028503562944e-05,
+      "loss": 0.1397,
+      "step": 8713
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.4222090261282664e-05,
+      "loss": 0.0823,
+      "step": 8714
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.421615201900238e-05,
+      "loss": 0.0968,
+      "step": 8715
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.42102137767221e-05,
+      "loss": 0.1103,
+      "step": 8716
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.4204275534441805e-05,
+      "loss": 0.1071,
+      "step": 8717
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.4198337292161525e-05,
+      "loss": 0.0993,
+      "step": 8718
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.419239904988124e-05,
+      "loss": 0.0815,
+      "step": 8719
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.4186460807600945e-05,
+      "loss": 0.1469,
+      "step": 8720
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.4180522565320666e-05,
+      "loss": 0.1648,
+      "step": 8721
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.417458432304038e-05,
+      "loss": 0.0888,
+      "step": 8722
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.41686460807601e-05,
+      "loss": 0.0844,
+      "step": 8723
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.4162707838479807e-05,
+      "loss": 0.0899,
+      "step": 8724
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.415676959619953e-05,
+      "loss": 0.1113,
+      "step": 8725
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.415083135391924e-05,
+      "loss": 0.1395,
+      "step": 8726
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.414489311163896e-05,
+      "loss": 0.0965,
+      "step": 8727
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.4138954869358674e-05,
+      "loss": 0.1156,
+      "step": 8728
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.4133016627078395e-05,
+      "loss": 0.2862,
+      "step": 8729
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.41270783847981e-05,
+      "loss": 0.2284,
+      "step": 8730
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.4121140142517815e-05,
+      "loss": 0.2321,
+      "step": 8731
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.4115201900237536e-05,
+      "loss": 0.2199,
+      "step": 8732
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.410926365795724e-05,
+      "loss": 0.1364,
+      "step": 8733
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.410332541567696e-05,
+      "loss": 0.1336,
+      "step": 8734
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.4097387173396676e-05,
+      "loss": 0.1349,
+      "step": 8735
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.40914489311164e-05,
+      "loss": 0.1402,
+      "step": 8736
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.4085510688836103e-05,
+      "loss": 0.1584,
+      "step": 8737
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.4079572446555824e-05,
+      "loss": 0.1175,
+      "step": 8738
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.407363420427554e-05,
+      "loss": 0.1457,
+      "step": 8739
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.406769596199526e-05,
+      "loss": 0.1994,
+      "step": 8740
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.4061757719714965e-05,
+      "loss": 0.1197,
+      "step": 8741
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.405581947743468e-05,
+      "loss": 0.129,
+      "step": 8742
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.40498812351544e-05,
+      "loss": 0.1315,
+      "step": 8743
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.4043942992874105e-05,
+      "loss": 0.1084,
+      "step": 8744
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.403800475059383e-05,
+      "loss": 0.1481,
+      "step": 8745
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.403206650831354e-05,
+      "loss": 0.1084,
+      "step": 8746
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.402612826603326e-05,
+      "loss": 0.1207,
+      "step": 8747
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.402019002375297e-05,
+      "loss": 0.1651,
+      "step": 8748
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.4014251781472694e-05,
+      "loss": 0.0762,
+      "step": 8749
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.40083135391924e-05,
+      "loss": 0.1741,
+      "step": 8750
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.4002375296912114e-05,
+      "loss": 0.1359,
+      "step": 8751
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.3996437054631834e-05,
+      "loss": 0.1567,
+      "step": 8752
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.399049881235154e-05,
+      "loss": 0.1472,
+      "step": 8753
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.398456057007126e-05,
+      "loss": 0.2873,
+      "step": 8754
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.3978622327790975e-05,
+      "loss": 0.1391,
+      "step": 8755
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.3972684085510695e-05,
+      "loss": 0.1812,
+      "step": 8756
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.39667458432304e-05,
+      "loss": 0.146,
+      "step": 8757
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.396080760095012e-05,
+      "loss": 0.1056,
+      "step": 8758
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.3954869358669836e-05,
+      "loss": 0.0936,
+      "step": 8759
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.3948931116389556e-05,
+      "loss": 0.121,
+      "step": 8760
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.394299287410926e-05,
+      "loss": 0.1155,
+      "step": 8761
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.393705463182898e-05,
+      "loss": 0.1034,
+      "step": 8762
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.39311163895487e-05,
+      "loss": 0.1512,
+      "step": 8763
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.3925178147268404e-05,
+      "loss": 0.0964,
+      "step": 8764
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.391923990498813e-05,
+      "loss": 0.1402,
+      "step": 8765
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.391330166270784e-05,
+      "loss": 0.1951,
+      "step": 8766
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.390736342042756e-05,
+      "loss": 0.102,
+      "step": 8767
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.390142517814727e-05,
+      "loss": 0.1203,
+      "step": 8768
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.389548693586699e-05,
+      "loss": 0.085,
+      "step": 8769
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.38895486935867e-05,
+      "loss": 0.1066,
+      "step": 8770
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.388361045130642e-05,
+      "loss": 0.1155,
+      "step": 8771
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.387767220902613e-05,
+      "loss": 0.0875,
+      "step": 8772
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.387173396674584e-05,
+      "loss": 0.0976,
+      "step": 8773
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.386579572446556e-05,
+      "loss": 0.0755,
+      "step": 8774
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.3859857482185274e-05,
+      "loss": 0.0996,
+      "step": 8775
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.3853919239904994e-05,
+      "loss": 0.0837,
+      "step": 8776
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.38479809976247e-05,
+      "loss": 0.111,
+      "step": 8777
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.384204275534442e-05,
+      "loss": 0.067,
+      "step": 8778
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.3836104513064135e-05,
+      "loss": 0.2519,
+      "step": 8779
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.3830166270783855e-05,
+      "loss": 0.2478,
+      "step": 8780
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.382422802850356e-05,
+      "loss": 0.2611,
+      "step": 8781
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.381828978622329e-05,
+      "loss": 0.1779,
+      "step": 8782
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.3812351543942996e-05,
+      "loss": 0.1593,
+      "step": 8783
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.38064133016627e-05,
+      "loss": 0.1744,
+      "step": 8784
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.380047505938243e-05,
+      "loss": 0.1553,
+      "step": 8785
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.379453681710214e-05,
+      "loss": 0.1734,
+      "step": 8786
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.378859857482186e-05,
+      "loss": 0.1502,
+      "step": 8787
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.378266033254157e-05,
+      "loss": 0.2084,
+      "step": 8788
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.377672209026129e-05,
+      "loss": 0.1949,
+      "step": 8789
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.3770783847981e-05,
+      "loss": 0.1902,
+      "step": 8790
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.376484560570072e-05,
+      "loss": 0.0991,
+      "step": 8791
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.375890736342043e-05,
+      "loss": 0.1537,
+      "step": 8792
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.375296912114014e-05,
+      "loss": 0.1843,
+      "step": 8793
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.374703087885986e-05,
+      "loss": 0.1231,
+      "step": 8794
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.374109263657957e-05,
+      "loss": 0.1102,
+      "step": 8795
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.373515439429929e-05,
+      "loss": 0.1843,
+      "step": 8796
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.3729216152019e-05,
+      "loss": 0.1896,
+      "step": 8797
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.372327790973872e-05,
+      "loss": 0.1205,
+      "step": 8798
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.3717339667458434e-05,
+      "loss": 0.1002,
+      "step": 8799
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.3711401425178154e-05,
+      "loss": 0.1326,
+      "step": 8800
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.370546318289786e-05,
+      "loss": 0.1373,
+      "step": 8801
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.369952494061759e-05,
+      "loss": 0.1665,
+      "step": 8802
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.3693586698337295e-05,
+      "loss": 0.1275,
+      "step": 8803
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.3687648456057e-05,
+      "loss": 0.1046,
+      "step": 8804
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.368171021377673e-05,
+      "loss": 0.1283,
+      "step": 8805
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.3675771971496435e-05,
+      "loss": 0.1351,
+      "step": 8806
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.3669833729216156e-05,
+      "loss": 0.0985,
+      "step": 8807
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.366389548693587e-05,
+      "loss": 0.1315,
+      "step": 8808
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.365795724465559e-05,
+      "loss": 0.1947,
+      "step": 8809
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.3652019002375296e-05,
+      "loss": 0.0904,
+      "step": 8810
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.364608076009502e-05,
+      "loss": 0.1302,
+      "step": 8811
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.364014251781473e-05,
+      "loss": 0.1473,
+      "step": 8812
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.363420427553445e-05,
+      "loss": 0.1513,
+      "step": 8813
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.362826603325416e-05,
+      "loss": 0.139,
+      "step": 8814
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.362232779097387e-05,
+      "loss": 0.1405,
+      "step": 8815
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.361638954869359e-05,
+      "loss": 0.0841,
+      "step": 8816
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.36104513064133e-05,
+      "loss": 0.0767,
+      "step": 8817
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.360451306413302e-05,
+      "loss": 0.0649,
+      "step": 8818
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.359857482185273e-05,
+      "loss": 0.1117,
+      "step": 8819
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.359263657957245e-05,
+      "loss": 0.0881,
+      "step": 8820
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.358669833729216e-05,
+      "loss": 0.062,
+      "step": 8821
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.3580760095011887e-05,
+      "loss": 0.1137,
+      "step": 8822
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.357482185273159e-05,
+      "loss": 0.0869,
+      "step": 8823
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.356888361045131e-05,
+      "loss": 0.107,
+      "step": 8824
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.356294536817103e-05,
+      "loss": 0.1163,
+      "step": 8825
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.3557007125890734e-05,
+      "loss": 0.18,
+      "step": 8826
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.3551068883610454e-05,
+      "loss": 0.126,
+      "step": 8827
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.354513064133017e-05,
+      "loss": 0.0435,
+      "step": 8828
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.353919239904989e-05,
+      "loss": 0.2919,
+      "step": 8829
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.3533254156769595e-05,
+      "loss": 0.2487,
+      "step": 8830
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.3527315914489316e-05,
+      "loss": 0.1994,
+      "step": 8831
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.352137767220903e-05,
+      "loss": 0.1854,
+      "step": 8832
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.351543942992875e-05,
+      "loss": 0.1764,
+      "step": 8833
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.3509501187648456e-05,
+      "loss": 0.1272,
+      "step": 8834
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.350356294536817e-05,
+      "loss": 0.1411,
+      "step": 8835
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.349762470308789e-05,
+      "loss": 0.1453,
+      "step": 8836
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.34916864608076e-05,
+      "loss": 0.1606,
+      "step": 8837
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.348574821852732e-05,
+      "loss": 0.2096,
+      "step": 8838
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.347980997624703e-05,
+      "loss": 0.1027,
+      "step": 8839
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.347387173396675e-05,
+      "loss": 0.1099,
+      "step": 8840
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.3467933491686465e-05,
+      "loss": 0.1883,
+      "step": 8841
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.3461995249406185e-05,
+      "loss": 0.1127,
+      "step": 8842
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.345605700712589e-05,
+      "loss": 0.1287,
+      "step": 8843
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.345011876484561e-05,
+      "loss": 0.1329,
+      "step": 8844
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.3444180522565326e-05,
+      "loss": 0.102,
+      "step": 8845
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.343824228028503e-05,
+      "loss": 0.1639,
+      "step": 8846
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.343230403800475e-05,
+      "loss": 0.1021,
+      "step": 8847
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.342636579572447e-05,
+      "loss": 0.1754,
+      "step": 8848
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.342042755344419e-05,
+      "loss": 0.1492,
+      "step": 8849
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.3414489311163894e-05,
+      "loss": 0.1553,
+      "step": 8850
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.3408551068883614e-05,
+      "loss": 0.117,
+      "step": 8851
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.340261282660333e-05,
+      "loss": 0.1174,
+      "step": 8852
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.339667458432305e-05,
+      "loss": 0.1079,
+      "step": 8853
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.3390736342042755e-05,
+      "loss": 0.0944,
+      "step": 8854
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.338479809976247e-05,
+      "loss": 0.0905,
+      "step": 8855
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.337885985748219e-05,
+      "loss": 0.1229,
+      "step": 8856
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.3372921615201896e-05,
+      "loss": 0.1277,
+      "step": 8857
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.336698337292162e-05,
+      "loss": 0.1407,
+      "step": 8858
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.336104513064133e-05,
+      "loss": 0.1162,
+      "step": 8859
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.335510688836105e-05,
+      "loss": 0.1181,
+      "step": 8860
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.3349168646080764e-05,
+      "loss": 0.1503,
+      "step": 8861
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.3343230403800484e-05,
+      "loss": 0.099,
+      "step": 8862
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.333729216152019e-05,
+      "loss": 0.088,
+      "step": 8863
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.333135391923991e-05,
+      "loss": 0.1033,
+      "step": 8864
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.3325415676959625e-05,
+      "loss": 0.1804,
+      "step": 8865
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.331947743467933e-05,
+      "loss": 0.0851,
+      "step": 8866
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.331353919239905e-05,
+      "loss": 0.1095,
+      "step": 8867
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.3307600950118765e-05,
+      "loss": 0.1214,
+      "step": 8868
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.3301662707838486e-05,
+      "loss": 0.0847,
+      "step": 8869
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.329572446555819e-05,
+      "loss": 0.1475,
+      "step": 8870
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.328978622327791e-05,
+      "loss": 0.1055,
+      "step": 8871
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.3283847980997627e-05,
+      "loss": 0.134,
+      "step": 8872
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.327790973871735e-05,
+      "loss": 0.1034,
+      "step": 8873
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.3271971496437054e-05,
+      "loss": 0.1158,
+      "step": 8874
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.3266033254156774e-05,
+      "loss": 0.1189,
+      "step": 8875
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.326009501187649e-05,
+      "loss": 0.0554,
+      "step": 8876
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.3254156769596194e-05,
+      "loss": 0.0774,
+      "step": 8877
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.324821852731592e-05,
+      "loss": 0.1139,
+      "step": 8878
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.324228028503563e-05,
+      "loss": 0.2601,
+      "step": 8879
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.323634204275535e-05,
+      "loss": 0.3102,
+      "step": 8880
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.323040380047506e-05,
+      "loss": 0.1452,
+      "step": 8881
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.322446555819478e-05,
+      "loss": 0.1777,
+      "step": 8882
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.321852731591449e-05,
+      "loss": 0.1663,
+      "step": 8883
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.321258907363421e-05,
+      "loss": 0.1142,
+      "step": 8884
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.3206650831353923e-05,
+      "loss": 0.1909,
+      "step": 8885
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.320071258907363e-05,
+      "loss": 0.1666,
+      "step": 8886
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.319477434679335e-05,
+      "loss": 0.1497,
+      "step": 8887
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.3188836104513064e-05,
+      "loss": 0.1465,
+      "step": 8888
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.3182897862232785e-05,
+      "loss": 0.181,
+      "step": 8889
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.317695961995249e-05,
+      "loss": 0.1704,
+      "step": 8890
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.317102137767221e-05,
+      "loss": 0.1456,
+      "step": 8891
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.3165083135391925e-05,
+      "loss": 0.2393,
+      "step": 8892
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.3159144893111646e-05,
+      "loss": 0.183,
+      "step": 8893
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.315320665083135e-05,
+      "loss": 0.1195,
+      "step": 8894
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.314726840855108e-05,
+      "loss": 0.1116,
+      "step": 8895
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.3141330166270786e-05,
+      "loss": 0.2216,
+      "step": 8896
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.313539192399049e-05,
+      "loss": 0.1343,
+      "step": 8897
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.312945368171022e-05,
+      "loss": 0.1101,
+      "step": 8898
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.312351543942993e-05,
+      "loss": 0.1498,
+      "step": 8899
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.311757719714965e-05,
+      "loss": 0.1156,
+      "step": 8900
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.311163895486936e-05,
+      "loss": 0.1345,
+      "step": 8901
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.310570071258908e-05,
+      "loss": 0.1483,
+      "step": 8902
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.309976247030879e-05,
+      "loss": 0.1807,
+      "step": 8903
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.309382422802851e-05,
+      "loss": 0.1217,
+      "step": 8904
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.308788598574822e-05,
+      "loss": 0.1242,
+      "step": 8905
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.308194774346794e-05,
+      "loss": 0.1389,
+      "step": 8906
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.307600950118765e-05,
+      "loss": 0.1062,
+      "step": 8907
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.307007125890736e-05,
+      "loss": 0.1589,
+      "step": 8908
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.306413301662708e-05,
+      "loss": 0.0998,
+      "step": 8909
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.305819477434679e-05,
+      "loss": 0.1239,
+      "step": 8910
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.305225653206651e-05,
+      "loss": 0.1021,
+      "step": 8911
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.3046318289786224e-05,
+      "loss": 0.0767,
+      "step": 8912
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.3040380047505944e-05,
+      "loss": 0.3175,
+      "step": 8913
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.303444180522565e-05,
+      "loss": 0.13,
+      "step": 8914
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.302850356294538e-05,
+      "loss": 0.1126,
+      "step": 8915
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.3022565320665085e-05,
+      "loss": 0.065,
+      "step": 8916
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.301662707838479e-05,
+      "loss": 0.1487,
+      "step": 8917
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.301068883610452e-05,
+      "loss": 0.0711,
+      "step": 8918
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.3004750593824226e-05,
+      "loss": 0.0681,
+      "step": 8919
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.2998812351543946e-05,
+      "loss": 0.1337,
+      "step": 8920
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.299287410926366e-05,
+      "loss": 0.2035,
+      "step": 8921
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.298693586698338e-05,
+      "loss": 0.1606,
+      "step": 8922
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.298099762470309e-05,
+      "loss": 0.1311,
+      "step": 8923
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.297505938242281e-05,
+      "loss": 0.1513,
+      "step": 8924
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.296912114014252e-05,
+      "loss": 0.1588,
+      "step": 8925
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.296318289786224e-05,
+      "loss": 0.0834,
+      "step": 8926
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.295724465558195e-05,
+      "loss": 0.1432,
+      "step": 8927
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.295130641330166e-05,
+      "loss": 0.0958,
+      "step": 8928
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.294536817102138e-05,
+      "loss": 0.1544,
+      "step": 8929
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.293942992874109e-05,
+      "loss": 0.1075,
+      "step": 8930
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.293349168646081e-05,
+      "loss": 0.114,
+      "step": 8931
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.292755344418052e-05,
+      "loss": 0.1998,
+      "step": 8932
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.292161520190024e-05,
+      "loss": 0.104,
+      "step": 8933
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 5.291567695961995e-05,
+      "loss": 0.074,
+      "step": 8934
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 5.290973871733968e-05,
+      "loss": 0.1402,
+      "step": 8935
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 5.2903800475059384e-05,
+      "loss": 0.0715,
+      "step": 8936
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 5.2897862232779104e-05,
+      "loss": 0.0935,
+      "step": 8937
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 5.289192399049882e-05,
+      "loss": 0.0925,
+      "step": 8938
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 5.2885985748218525e-05,
+      "loss": 0.1112,
+      "step": 8939
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 5.2880047505938245e-05,
+      "loss": 0.0962,
+      "step": 8940
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 5.287410926365796e-05,
+      "loss": 0.0601,
+      "step": 8941
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 5.286817102137768e-05,
+      "loss": 0.1017,
+      "step": 8942
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 5.2862232779097386e-05,
+      "loss": 0.0764,
+      "step": 8943
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 5.2856294536817106e-05,
+      "loss": 0.0974,
+      "step": 8944
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 5.285035629453682e-05,
+      "loss": 0.0839,
+      "step": 8945
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 5.284441805225654e-05,
+      "loss": 0.1294,
+      "step": 8946
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 5.283847980997625e-05,
+      "loss": 0.0718,
+      "step": 8947
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 5.283254156769597e-05,
+      "loss": 0.0789,
+      "step": 8948
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 5.282660332541568e-05,
+      "loss": 0.0786,
+      "step": 8949
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 5.282066508313539e-05,
+      "loss": 0.0942,
+      "step": 8950
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 5.281472684085511e-05,
+      "loss": 0.0878,
+      "step": 8951
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 5.280878859857482e-05,
+      "loss": 0.0881,
+      "step": 8952
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 5.280285035629454e-05,
+      "loss": 0.0748,
+      "step": 8953
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 5.279691211401425e-05,
+      "loss": 0.0927,
+      "step": 8954
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 5.2790973871733976e-05,
+      "loss": 0.1013,
+      "step": 8955
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 5.278503562945368e-05,
+      "loss": 0.0793,
+      "step": 8956
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 5.27790973871734e-05,
+      "loss": 0.0538,
+      "step": 8957
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 5.2773159144893116e-05,
+      "loss": 0.0557,
+      "step": 8958
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 5.276722090261282e-05,
+      "loss": 0.1132,
+      "step": 8959
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 5.2761282660332544e-05,
+      "loss": 0.1067,
+      "step": 8960
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 5.275534441805226e-05,
+      "loss": 0.1034,
+      "step": 8961
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 5.274940617577198e-05,
+      "loss": 0.0957,
+      "step": 8962
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 5.2743467933491684e-05,
+      "loss": 0.0923,
+      "step": 8963
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 5.2737529691211405e-05,
+      "loss": 0.093,
+      "step": 8964
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 5.273159144893112e-05,
+      "loss": 0.0972,
+      "step": 8965
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 5.272565320665084e-05,
+      "loss": 0.0283,
+      "step": 8966
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 5.2719714964370545e-05,
+      "loss": 0.092,
+      "step": 8967
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 5.2713776722090266e-05,
+      "loss": 0.0356,
+      "step": 8968
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 5.270783847980998e-05,
+      "loss": 0.0483,
+      "step": 8969
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 5.2701900237529686e-05,
+      "loss": 0.0755,
+      "step": 8970
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 5.2695961995249407e-05,
+      "loss": 0.2097,
+      "step": 8971
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 5.269002375296912e-05,
+      "loss": 0.1317,
+      "step": 8972
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 5.268408551068884e-05,
+      "loss": 0.2236,
+      "step": 8973
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 5.2678147268408554e-05,
+      "loss": 0.1381,
+      "step": 8974
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 5.2672209026128274e-05,
+      "loss": 0.1495,
+      "step": 8975
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 5.266627078384798e-05,
+      "loss": 0.0994,
+      "step": 8976
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 5.26603325415677e-05,
+      "loss": 0.1191,
+      "step": 8977
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 5.2654394299287415e-05,
+      "loss": 0.1365,
+      "step": 8978
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 5.2648456057007136e-05,
+      "loss": 0.0815,
+      "step": 8979
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 5.264251781472684e-05,
+      "loss": 0.1385,
+      "step": 8980
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 5.2636579572446556e-05,
+      "loss": 0.1125,
+      "step": 8981
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 5.2630641330166276e-05,
+      "loss": 0.1041,
+      "step": 8982
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 5.262470308788598e-05,
+      "loss": 0.1529,
+      "step": 8983
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 5.2618764845605703e-05,
+      "loss": 0.0759,
+      "step": 8984
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 5.261282660332542e-05,
+      "loss": 0.1184,
+      "step": 8985
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 5.260688836104514e-05,
+      "loss": 0.1453,
+      "step": 8986
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 5.2600950118764844e-05,
+      "loss": 0.1078,
+      "step": 8987
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 5.2595011876484565e-05,
+      "loss": 0.0931,
+      "step": 8988
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 5.258907363420428e-05,
+      "loss": 0.1004,
+      "step": 8989
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 5.2583135391923985e-05,
+      "loss": 0.1173,
+      "step": 8990
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 5.257719714964371e-05,
+      "loss": 0.118,
+      "step": 8991
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 5.257125890736342e-05,
+      "loss": 0.0865,
+      "step": 8992
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 5.256532066508314e-05,
+      "loss": 0.1057,
+      "step": 8993
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 5.255938242280285e-05,
+      "loss": 0.1018,
+      "step": 8994
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 5.255344418052257e-05,
+      "loss": 0.1143,
+      "step": 8995
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 5.254750593824228e-05,
+      "loss": 0.1202,
+      "step": 8996
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 5.2541567695962e-05,
+      "loss": 0.0809,
+      "step": 8997
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 5.2535629453681714e-05,
+      "loss": 0.0633,
+      "step": 8998
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 5.2529691211401434e-05,
+      "loss": 0.0516,
+      "step": 8999
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 5.252375296912114e-05,
+      "loss": 0.0608,
+      "step": 9000
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 5.2517814726840855e-05,
+      "loss": 0.0833,
+      "step": 9001
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 5.2511876484560575e-05,
+      "loss": 0.0958,
+      "step": 9002
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 5.250593824228028e-05,
+      "loss": 0.0929,
+      "step": 9003
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 5.25e-05,
+      "loss": 0.0928,
+      "step": 9004
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 5.2494061757719716e-05,
+      "loss": 0.0736,
+      "step": 9005
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 5.2488123515439436e-05,
+      "loss": 0.0654,
+      "step": 9006
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 5.248218527315914e-05,
+      "loss": 0.0631,
+      "step": 9007
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 5.247624703087887e-05,
+      "loss": 0.1078,
+      "step": 9008
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 5.247030878859858e-05,
+      "loss": 0.0689,
+      "step": 9009
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 5.24643705463183e-05,
+      "loss": 0.0695,
+      "step": 9010
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 5.245843230403801e-05,
+      "loss": 0.0467,
+      "step": 9011
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 5.245249406175772e-05,
+      "loss": 0.0669,
+      "step": 9012
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 5.244655581947744e-05,
+      "loss": 0.045,
+      "step": 9013
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 5.244061757719715e-05,
+      "loss": 0.1392,
+      "step": 9014
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 5.243467933491687e-05,
+      "loss": 0.0823,
+      "step": 9015
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 5.242874109263658e-05,
+      "loss": 0.0836,
+      "step": 9016
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 5.24228028503563e-05,
+      "loss": 0.1025,
+      "step": 9017
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 5.241686460807601e-05,
+      "loss": 0.0643,
+      "step": 9018
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 5.241092636579573e-05,
+      "loss": 0.0506,
+      "step": 9019
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 5.240498812351544e-05,
+      "loss": 0.1109,
+      "step": 9020
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 5.239904988123515e-05,
+      "loss": 0.1994,
+      "step": 9021
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 5.2393111638954874e-05,
+      "loss": 0.1458,
+      "step": 9022
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 5.238717339667458e-05,
+      "loss": 0.1859,
+      "step": 9023
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 5.23812351543943e-05,
+      "loss": 0.1532,
+      "step": 9024
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 5.2375296912114014e-05,
+      "loss": 0.1483,
+      "step": 9025
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 5.2369358669833735e-05,
+      "loss": 0.129,
+      "step": 9026
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 5.236342042755344e-05,
+      "loss": 0.0994,
+      "step": 9027
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 5.235748218527317e-05,
+      "loss": 0.0857,
+      "step": 9028
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 5.2351543942992876e-05,
+      "loss": 0.1606,
+      "step": 9029
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 5.2345605700712596e-05,
+      "loss": 0.1349,
+      "step": 9030
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 5.233966745843231e-05,
+      "loss": 0.0981,
+      "step": 9031
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 5.2333729216152016e-05,
+      "loss": 0.105,
+      "step": 9032
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 5.232779097387174e-05,
+      "loss": 0.0918,
+      "step": 9033
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 5.232185273159145e-05,
+      "loss": 0.1189,
+      "step": 9034
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 5.231591448931117e-05,
+      "loss": 0.1115,
+      "step": 9035
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 5.230997624703088e-05,
+      "loss": 0.1081,
+      "step": 9036
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 5.23040380047506e-05,
+      "loss": 0.0816,
+      "step": 9037
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 5.229809976247031e-05,
+      "loss": 0.1104,
+      "step": 9038
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 5.229216152019003e-05,
+      "loss": 0.1186,
+      "step": 9039
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 5.228622327790974e-05,
+      "loss": 0.0949,
+      "step": 9040
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 5.228028503562946e-05,
+      "loss": 0.1078,
+      "step": 9041
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 5.227434679334917e-05,
+      "loss": 0.0633,
+      "step": 9042
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 5.226840855106888e-05,
+      "loss": 0.0756,
+      "step": 9043
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 5.22624703087886e-05,
+      "loss": 0.0632,
+      "step": 9044
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 5.225653206650831e-05,
+      "loss": 0.1063,
+      "step": 9045
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 5.2250593824228034e-05,
+      "loss": 0.1243,
+      "step": 9046
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 5.224465558194774e-05,
+      "loss": 0.108,
+      "step": 9047
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 5.223871733966747e-05,
+      "loss": 0.0598,
+      "step": 9048
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 5.2232779097387174e-05,
+      "loss": 0.098,
+      "step": 9049
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 5.2226840855106895e-05,
+      "loss": 0.4106,
+      "step": 9050
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 5.222090261282661e-05,
+      "loss": 0.0669,
+      "step": 9051
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 5.2214964370546315e-05,
+      "loss": 0.0862,
+      "step": 9052
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 5.2209026128266035e-05,
+      "loss": 0.1234,
+      "step": 9053
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 5.220308788598575e-05,
+      "loss": 0.0909,
+      "step": 9054
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 5.219714964370547e-05,
+      "loss": 0.0918,
+      "step": 9055
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 5.2191211401425176e-05,
+      "loss": 0.0698,
+      "step": 9056
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 5.2185273159144896e-05,
+      "loss": 0.0461,
+      "step": 9057
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 5.217933491686461e-05,
+      "loss": 0.068,
+      "step": 9058
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 5.217339667458433e-05,
+      "loss": 0.0759,
+      "step": 9059
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 5.216745843230404e-05,
+      "loss": 0.1184,
+      "step": 9060
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 5.216152019002376e-05,
+      "loss": 0.0631,
+      "step": 9061
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 5.215558194774347e-05,
+      "loss": 0.0788,
+      "step": 9062
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 5.214964370546318e-05,
+      "loss": 0.1052,
+      "step": 9063
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 5.21437054631829e-05,
+      "loss": 0.0612,
+      "step": 9064
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 5.213776722090261e-05,
+      "loss": 0.0573,
+      "step": 9065
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 5.213182897862233e-05,
+      "loss": 0.0496,
+      "step": 9066
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 5.212589073634204e-05,
+      "loss": 0.094,
+      "step": 9067
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 5.2119952494061766e-05,
+      "loss": 0.1314,
+      "step": 9068
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 5.211401425178147e-05,
+      "loss": 0.0738,
+      "step": 9069
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 5.210807600950119e-05,
+      "loss": 0.0654,
+      "step": 9070
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 5.210213776722091e-05,
+      "loss": 0.1987,
+      "step": 9071
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 5.209619952494063e-05,
+      "loss": 0.178,
+      "step": 9072
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 5.2090261282660334e-05,
+      "loss": 0.2608,
+      "step": 9073
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 5.208432304038005e-05,
+      "loss": 0.1442,
+      "step": 9074
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 5.207838479809977e-05,
+      "loss": 0.1122,
+      "step": 9075
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 5.2072446555819475e-05,
+      "loss": 0.0801,
+      "step": 9076
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 5.2066508313539195e-05,
+      "loss": 0.1166,
+      "step": 9077
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 5.206057007125891e-05,
+      "loss": 0.0744,
+      "step": 9078
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 5.205463182897863e-05,
+      "loss": 0.0977,
+      "step": 9079
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 5.2048693586698336e-05,
+      "loss": 0.0912,
+      "step": 9080
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 5.2042755344418056e-05,
+      "loss": 0.2703,
+      "step": 9081
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 5.203681710213777e-05,
+      "loss": 0.2086,
+      "step": 9082
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 5.203087885985748e-05,
+      "loss": 0.1865,
+      "step": 9083
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 5.20249406175772e-05,
+      "loss": 0.0985,
+      "step": 9084
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 5.201900237529691e-05,
+      "loss": 0.1321,
+      "step": 9085
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 5.201306413301663e-05,
+      "loss": 0.1184,
+      "step": 9086
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 5.2007125890736345e-05,
+      "loss": 0.1071,
+      "step": 9087
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 5.2001187648456065e-05,
+      "loss": 0.1226,
+      "step": 9088
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 5.199524940617577e-05,
+      "loss": 0.0916,
+      "step": 9089
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 5.198931116389549e-05,
+      "loss": 0.1444,
+      "step": 9090
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 5.1983372921615206e-05,
+      "loss": 0.1089,
+      "step": 9091
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 5.1977434679334926e-05,
+      "loss": 0.1178,
+      "step": 9092
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 5.197149643705463e-05,
+      "loss": 0.1188,
+      "step": 9093
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 5.1965558194774346e-05,
+      "loss": 0.0897,
+      "step": 9094
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 5.195961995249407e-05,
+      "loss": 0.1358,
+      "step": 9095
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 5.1953681710213774e-05,
+      "loss": 0.1146,
+      "step": 9096
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 5.1947743467933494e-05,
+      "loss": 0.0942,
+      "step": 9097
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 5.194180522565321e-05,
+      "loss": 0.0751,
+      "step": 9098
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 5.193586698337293e-05,
+      "loss": 0.1031,
+      "step": 9099
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 5.1929928741092635e-05,
+      "loss": 0.1167,
+      "step": 9100
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 5.1923990498812355e-05,
+      "loss": 0.0813,
+      "step": 9101
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 5.191805225653207e-05,
+      "loss": 0.0578,
+      "step": 9102
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 5.191211401425179e-05,
+      "loss": 0.0757,
+      "step": 9103
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 5.19061757719715e-05,
+      "loss": 0.0763,
+      "step": 9104
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 5.190023752969121e-05,
+      "loss": 0.0587,
+      "step": 9105
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 5.189429928741093e-05,
+      "loss": 0.076,
+      "step": 9106
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 5.188836104513064e-05,
+      "loss": 0.0738,
+      "step": 9107
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 5.1882422802850364e-05,
+      "loss": 0.0911,
+      "step": 9108
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 5.187648456057007e-05,
+      "loss": 0.1132,
+      "step": 9109
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 5.187054631828979e-05,
+      "loss": 0.0644,
+      "step": 9110
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 5.1864608076009504e-05,
+      "loss": 0.138,
+      "step": 9111
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 5.1858669833729225e-05,
+      "loss": 0.155,
+      "step": 9112
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 5.185273159144893e-05,
+      "loss": 0.0565,
+      "step": 9113
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 5.1846793349168645e-05,
+      "loss": 0.0634,
+      "step": 9114
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 5.1840855106888365e-05,
+      "loss": 0.0915,
+      "step": 9115
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 5.183491686460807e-05,
+      "loss": 0.08,
+      "step": 9116
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 5.182897862232779e-05,
+      "loss": 0.0842,
+      "step": 9117
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 5.1823040380047506e-05,
+      "loss": 0.1167,
+      "step": 9118
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 5.1817102137767227e-05,
+      "loss": 0.2526,
+      "step": 9119
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 5.1811163895486933e-05,
+      "loss": 0.0729,
+      "step": 9120
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 5.1805225653206654e-05,
+      "loss": 0.1868,
+      "step": 9121
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 5.179928741092637e-05,
+      "loss": 0.19,
+      "step": 9122
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 5.179334916864609e-05,
+      "loss": 0.1604,
+      "step": 9123
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 5.17874109263658e-05,
+      "loss": 0.1411,
+      "step": 9124
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 5.178147268408551e-05,
+      "loss": 0.1192,
+      "step": 9125
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 5.177553444180523e-05,
+      "loss": 0.0824,
+      "step": 9126
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 5.176959619952494e-05,
+      "loss": 0.1099,
+      "step": 9127
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 5.176365795724466e-05,
+      "loss": 0.1114,
+      "step": 9128
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 5.175771971496437e-05,
+      "loss": 0.1211,
+      "step": 9129
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 5.175178147268409e-05,
+      "loss": 0.1844,
+      "step": 9130
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 5.17458432304038e-05,
+      "loss": 0.0909,
+      "step": 9131
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 5.1739904988123523e-05,
+      "loss": 0.0949,
+      "step": 9132
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 5.173396674584323e-05,
+      "loss": 0.113,
+      "step": 9133
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 5.172802850356295e-05,
+      "loss": 0.0751,
+      "step": 9134
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 5.1722090261282664e-05,
+      "loss": 0.0853,
+      "step": 9135
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 5.171615201900237e-05,
+      "loss": 0.1116,
+      "step": 9136
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 5.171021377672209e-05,
+      "loss": 0.0789,
+      "step": 9137
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 5.1704275534441805e-05,
+      "loss": 0.1038,
+      "step": 9138
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 5.1698337292161525e-05,
+      "loss": 0.1103,
+      "step": 9139
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 5.169239904988123e-05,
+      "loss": 0.0965,
+      "step": 9140
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 5.168646080760096e-05,
+      "loss": 0.0954,
+      "step": 9141
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 5.1680522565320666e-05,
+      "loss": 0.0885,
+      "step": 9142
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 5.1674584323040386e-05,
+      "loss": 0.1446,
+      "step": 9143
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 5.16686460807601e-05,
+      "loss": 0.1221,
+      "step": 9144
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 5.166270783847982e-05,
+      "loss": 0.0624,
+      "step": 9145
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 5.165676959619953e-05,
+      "loss": 0.0983,
+      "step": 9146
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 5.165083135391924e-05,
+      "loss": 0.075,
+      "step": 9147
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 5.164489311163896e-05,
+      "loss": 0.1577,
+      "step": 9148
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 5.163895486935867e-05,
+      "loss": 0.1145,
+      "step": 9149
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 5.163301662707839e-05,
+      "loss": 0.1027,
+      "step": 9150
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 5.16270783847981e-05,
+      "loss": 0.105,
+      "step": 9151
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 5.162114014251782e-05,
+      "loss": 0.0699,
+      "step": 9152
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 5.161520190023753e-05,
+      "loss": 0.0784,
+      "step": 9153
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 5.160926365795725e-05,
+      "loss": 0.0771,
+      "step": 9154
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 5.160332541567696e-05,
+      "loss": 0.0811,
+      "step": 9155
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 5.159738717339667e-05,
+      "loss": 0.0768,
+      "step": 9156
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 5.159144893111639e-05,
+      "loss": 0.0812,
+      "step": 9157
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 5.1585510688836104e-05,
+      "loss": 0.1089,
+      "step": 9158
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 5.1579572446555824e-05,
+      "loss": 0.0452,
+      "step": 9159
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 5.157363420427553e-05,
+      "loss": 0.0533,
+      "step": 9160
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 5.156769596199526e-05,
+      "loss": 0.0662,
+      "step": 9161
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 5.1561757719714965e-05,
+      "loss": 0.0399,
+      "step": 9162
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 5.1555819477434685e-05,
+      "loss": 0.0653,
+      "step": 9163
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 5.15498812351544e-05,
+      "loss": 0.0724,
+      "step": 9164
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 5.154394299287412e-05,
+      "loss": 0.0487,
+      "step": 9165
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 5.1538004750593826e-05,
+      "loss": 0.0704,
+      "step": 9166
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 5.153206650831354e-05,
+      "loss": 0.1624,
+      "step": 9167
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 5.152612826603326e-05,
+      "loss": 0.177,
+      "step": 9168
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 5.1520190023752967e-05,
+      "loss": 0.0887,
+      "step": 9169
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 5.151425178147269e-05,
+      "loss": 0.0863,
+      "step": 9170
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 5.15083135391924e-05,
+      "loss": 0.2294,
+      "step": 9171
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 5.150237529691212e-05,
+      "loss": 0.2253,
+      "step": 9172
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 5.149643705463183e-05,
+      "loss": 0.1581,
+      "step": 9173
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 5.149049881235155e-05,
+      "loss": 0.1577,
+      "step": 9174
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 5.148456057007126e-05,
+      "loss": 0.0884,
+      "step": 9175
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 5.147862232779098e-05,
+      "loss": 0.1097,
+      "step": 9176
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 5.147268408551069e-05,
+      "loss": 0.0954,
+      "step": 9177
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 5.14667458432304e-05,
+      "loss": 0.131,
+      "step": 9178
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 5.146080760095012e-05,
+      "loss": 0.137,
+      "step": 9179
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 5.145486935866983e-05,
+      "loss": 0.1172,
+      "step": 9180
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 5.144893111638956e-05,
+      "loss": 0.123,
+      "step": 9181
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 5.1442992874109263e-05,
+      "loss": 0.1136,
+      "step": 9182
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 5.1437054631828984e-05,
+      "loss": 0.1376,
+      "step": 9183
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 5.14311163895487e-05,
+      "loss": 0.1084,
+      "step": 9184
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 5.142517814726842e-05,
+      "loss": 0.0899,
+      "step": 9185
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 5.1419239904988125e-05,
+      "loss": 0.0874,
+      "step": 9186
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 5.141330166270784e-05,
+      "loss": 0.1179,
+      "step": 9187
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 5.140736342042756e-05,
+      "loss": 0.1247,
+      "step": 9188
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 5.1401425178147265e-05,
+      "loss": 0.0639,
+      "step": 9189
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 5.1395486935866986e-05,
+      "loss": 0.0994,
+      "step": 9190
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 5.13895486935867e-05,
+      "loss": 0.0828,
+      "step": 9191
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 5.138361045130642e-05,
+      "loss": 0.0872,
+      "step": 9192
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 5.1377672209026126e-05,
+      "loss": 0.0934,
+      "step": 9193
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 5.137173396674585e-05,
+      "loss": 0.1183,
+      "step": 9194
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 5.136579572446556e-05,
+      "loss": 0.0954,
+      "step": 9195
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 5.135985748218528e-05,
+      "loss": 0.0984,
+      "step": 9196
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 5.135391923990499e-05,
+      "loss": 0.078,
+      "step": 9197
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 5.13479809976247e-05,
+      "loss": 0.1045,
+      "step": 9198
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 5.134204275534442e-05,
+      "loss": 0.0842,
+      "step": 9199
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 5.133610451306413e-05,
+      "loss": 0.079,
+      "step": 9200
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 5.1330166270783855e-05,
+      "loss": 0.0671,
+      "step": 9201
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 5.132422802850356e-05,
+      "loss": 0.069,
+      "step": 9202
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 5.131828978622328e-05,
+      "loss": 0.0685,
+      "step": 9203
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 5.1312351543942996e-05,
+      "loss": 0.0784,
+      "step": 9204
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 5.1306413301662716e-05,
+      "loss": 0.1009,
+      "step": 9205
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 5.130047505938242e-05,
+      "loss": 0.0506,
+      "step": 9206
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 5.1294536817102144e-05,
+      "loss": 0.089,
+      "step": 9207
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 5.128859857482186e-05,
+      "loss": 0.0671,
+      "step": 9208
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 5.1282660332541564e-05,
+      "loss": 0.0976,
+      "step": 9209
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 5.1276722090261284e-05,
+      "loss": 0.071,
+      "step": 9210
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 5.1270783847981e-05,
+      "loss": 0.1402,
+      "step": 9211
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 5.126484560570072e-05,
+      "loss": 0.0761,
+      "step": 9212
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 5.1258907363420425e-05,
+      "loss": 0.0935,
+      "step": 9213
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 5.1252969121140145e-05,
+      "loss": 0.2418,
+      "step": 9214
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 5.124703087885986e-05,
+      "loss": 0.1045,
+      "step": 9215
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 5.124109263657958e-05,
+      "loss": 0.1133,
+      "step": 9216
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 5.1235154394299286e-05,
+      "loss": 0.0449,
+      "step": 9217
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 5.1229216152019e-05,
+      "loss": 0.0492,
+      "step": 9218
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 5.122327790973872e-05,
+      "loss": 0.0249,
+      "step": 9219
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 5.1217339667458434e-05,
+      "loss": 0.0586,
+      "step": 9220
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 5.1211401425178154e-05,
+      "loss": 0.1932,
+      "step": 9221
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 5.120546318289786e-05,
+      "loss": 0.1736,
+      "step": 9222
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 5.119952494061758e-05,
+      "loss": 0.1413,
+      "step": 9223
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 5.1193586698337295e-05,
+      "loss": 0.1628,
+      "step": 9224
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 5.1187648456057015e-05,
+      "loss": 0.1393,
+      "step": 9225
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 5.118171021377672e-05,
+      "loss": 0.154,
+      "step": 9226
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 5.117577197149644e-05,
+      "loss": 0.1201,
+      "step": 9227
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 5.1169833729216156e-05,
+      "loss": 0.129,
+      "step": 9228
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 5.116389548693586e-05,
+      "loss": 0.1346,
+      "step": 9229
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 5.115795724465558e-05,
+      "loss": 0.1068,
+      "step": 9230
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 5.11520190023753e-05,
+      "loss": 0.0867,
+      "step": 9231
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 5.114608076009502e-05,
+      "loss": 0.124,
+      "step": 9232
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 5.1140142517814724e-05,
+      "loss": 0.1433,
+      "step": 9233
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 5.1134204275534444e-05,
+      "loss": 0.0988,
+      "step": 9234
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 5.112826603325416e-05,
+      "loss": 0.1581,
+      "step": 9235
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 5.112232779097388e-05,
+      "loss": 0.1432,
+      "step": 9236
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 5.111638954869359e-05,
+      "loss": 0.0797,
+      "step": 9237
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 5.111045130641331e-05,
+      "loss": 0.1229,
+      "step": 9238
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 5.110451306413302e-05,
+      "loss": 0.085,
+      "step": 9239
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 5.109857482185273e-05,
+      "loss": 0.092,
+      "step": 9240
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 5.109263657957245e-05,
+      "loss": 0.0974,
+      "step": 9241
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 5.108669833729216e-05,
+      "loss": 0.1088,
+      "step": 9242
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 5.108076009501188e-05,
+      "loss": 0.0914,
+      "step": 9243
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 5.1074821852731594e-05,
+      "loss": 0.138,
+      "step": 9244
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 5.1068883610451314e-05,
+      "loss": 0.0695,
+      "step": 9245
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 5.106294536817102e-05,
+      "loss": 0.1037,
+      "step": 9246
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 5.105700712589074e-05,
+      "loss": 0.0626,
+      "step": 9247
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 5.1051068883610455e-05,
+      "loss": 0.1217,
+      "step": 9248
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 5.104513064133016e-05,
+      "loss": 0.0869,
+      "step": 9249
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 5.103919239904988e-05,
+      "loss": 0.0589,
+      "step": 9250
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 5.1033254156769595e-05,
+      "loss": 0.124,
+      "step": 9251
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 5.1027315914489316e-05,
+      "loss": 0.1412,
+      "step": 9252
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 5.102137767220902e-05,
+      "loss": 0.0956,
+      "step": 9253
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 5.101543942992875e-05,
+      "loss": 0.1334,
+      "step": 9254
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 5.1009501187648457e-05,
+      "loss": 0.087,
+      "step": 9255
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 5.100356294536818e-05,
+      "loss": 0.0922,
+      "step": 9256
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 5.099762470308789e-05,
+      "loss": 0.1116,
+      "step": 9257
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 5.099168646080761e-05,
+      "loss": 0.0736,
+      "step": 9258
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 5.098574821852732e-05,
+      "loss": 0.0628,
+      "step": 9259
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 5.097980997624703e-05,
+      "loss": 0.0885,
+      "step": 9260
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 5.097387173396675e-05,
+      "loss": 0.0934,
+      "step": 9261
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 5.096793349168646e-05,
+      "loss": 0.0652,
+      "step": 9262
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 5.096199524940618e-05,
+      "loss": 0.0877,
+      "step": 9263
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 5.095605700712589e-05,
+      "loss": 0.07,
+      "step": 9264
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 5.095011876484561e-05,
+      "loss": 0.4082,
+      "step": 9265
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 5.094418052256532e-05,
+      "loss": 0.0627,
+      "step": 9266
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 5.093824228028504e-05,
+      "loss": 0.0649,
+      "step": 9267
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 5.0932304038004753e-05,
+      "loss": 0.3581,
+      "step": 9268
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 5.0926365795724474e-05,
+      "loss": 0.0325,
+      "step": 9269
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 5.092042755344418e-05,
+      "loss": 0.1413,
+      "step": 9270
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 5.0914489311163894e-05,
+      "loss": 0.2696,
+      "step": 9271
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 5.0908551068883614e-05,
+      "loss": 0.1427,
+      "step": 9272
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 5.090261282660332e-05,
+      "loss": 0.198,
+      "step": 9273
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 5.089667458432305e-05,
+      "loss": 0.1643,
+      "step": 9274
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 5.0890736342042755e-05,
+      "loss": 0.1277,
+      "step": 9275
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 5.0884798099762476e-05,
+      "loss": 0.1182,
+      "step": 9276
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 5.087885985748219e-05,
+      "loss": 0.1129,
+      "step": 9277
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 5.087292161520191e-05,
+      "loss": 0.1072,
+      "step": 9278
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 5.0866983372921616e-05,
+      "loss": 0.1387,
+      "step": 9279
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 5.086104513064133e-05,
+      "loss": 0.0801,
+      "step": 9280
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 5.085510688836105e-05,
+      "loss": 0.1298,
+      "step": 9281
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 5.084916864608076e-05,
+      "loss": 0.1183,
+      "step": 9282
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 5.084323040380048e-05,
+      "loss": 0.1071,
+      "step": 9283
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 5.083729216152019e-05,
+      "loss": 0.1969,
+      "step": 9284
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 5.083135391923991e-05,
+      "loss": 0.1879,
+      "step": 9285
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 5.082541567695962e-05,
+      "loss": 0.1132,
+      "step": 9286
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 5.081947743467934e-05,
+      "loss": 0.1368,
+      "step": 9287
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 5.081353919239905e-05,
+      "loss": 0.0848,
+      "step": 9288
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 5.080760095011877e-05,
+      "loss": 0.1477,
+      "step": 9289
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 5.080166270783848e-05,
+      "loss": 0.1216,
+      "step": 9290
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 5.079572446555819e-05,
+      "loss": 0.0997,
+      "step": 9291
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 5.078978622327791e-05,
+      "loss": 0.1083,
+      "step": 9292
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 5.078384798099762e-05,
+      "loss": 0.1139,
+      "step": 9293
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 5.077790973871735e-05,
+      "loss": 0.1725,
+      "step": 9294
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 5.0771971496437054e-05,
+      "loss": 0.1447,
+      "step": 9295
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 5.0766033254156774e-05,
+      "loss": 0.0842,
+      "step": 9296
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 5.076009501187649e-05,
+      "loss": 0.1134,
+      "step": 9297
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 5.075415676959621e-05,
+      "loss": 0.134,
+      "step": 9298
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 5.0748218527315915e-05,
+      "loss": 0.09,
+      "step": 9299
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 5.0742280285035635e-05,
+      "loss": 0.1013,
+      "step": 9300
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 5.073634204275535e-05,
+      "loss": 0.1081,
+      "step": 9301
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 5.0730403800475056e-05,
+      "loss": 0.0828,
+      "step": 9302
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 5.0724465558194776e-05,
+      "loss": 0.0917,
+      "step": 9303
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 5.071852731591449e-05,
+      "loss": 0.1133,
+      "step": 9304
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 5.071258907363421e-05,
+      "loss": 0.1063,
+      "step": 9305
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 5.070665083135392e-05,
+      "loss": 0.0803,
+      "step": 9306
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 5.070071258907364e-05,
+      "loss": 0.0454,
+      "step": 9307
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 5.069477434679335e-05,
+      "loss": 0.0727,
+      "step": 9308
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 5.068883610451307e-05,
+      "loss": 0.0955,
+      "step": 9309
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 5.068289786223278e-05,
+      "loss": 0.0897,
+      "step": 9310
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 5.067695961995249e-05,
+      "loss": 0.0657,
+      "step": 9311
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 5.067102137767221e-05,
+      "loss": 0.0875,
+      "step": 9312
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 5.066508313539192e-05,
+      "loss": 0.0824,
+      "step": 9313
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 5.0659144893111646e-05,
+      "loss": 0.194,
+      "step": 9314
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 5.065320665083135e-05,
+      "loss": 0.1171,
+      "step": 9315
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 5.064726840855107e-05,
+      "loss": 0.1098,
+      "step": 9316
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 5.0641330166270787e-05,
+      "loss": 0.0723,
+      "step": 9317
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 5.063539192399051e-05,
+      "loss": 0.0306,
+      "step": 9318
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 5.0629453681710214e-05,
+      "loss": 0.0714,
+      "step": 9319
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 5.0623515439429934e-05,
+      "loss": 0.06,
+      "step": 9320
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 5.061757719714965e-05,
+      "loss": 0.2065,
+      "step": 9321
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 5.0611638954869355e-05,
+      "loss": 0.1839,
+      "step": 9322
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 5.0605700712589075e-05,
+      "loss": 0.1978,
+      "step": 9323
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 5.059976247030879e-05,
+      "loss": 0.1417,
+      "step": 9324
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 5.059382422802851e-05,
+      "loss": 0.1264,
+      "step": 9325
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 5.0587885985748216e-05,
+      "loss": 0.1024,
+      "step": 9326
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 5.0581947743467936e-05,
+      "loss": 0.073,
+      "step": 9327
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 5.057600950118765e-05,
+      "loss": 0.0686,
+      "step": 9328
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 5.057007125890737e-05,
+      "loss": 0.1235,
+      "step": 9329
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 5.056413301662708e-05,
+      "loss": 0.1316,
+      "step": 9330
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 5.0558194774346804e-05,
+      "loss": 0.1125,
+      "step": 9331
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 5.055225653206651e-05,
+      "loss": 0.1218,
+      "step": 9332
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 5.0546318289786224e-05,
+      "loss": 0.1054,
+      "step": 9333
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 5.0540380047505945e-05,
+      "loss": 0.0815,
+      "step": 9334
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 5.053444180522565e-05,
+      "loss": 0.07,
+      "step": 9335
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 5.052850356294537e-05,
+      "loss": 0.1332,
+      "step": 9336
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 5.0522565320665085e-05,
+      "loss": 0.1173,
+      "step": 9337
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 5.0516627078384806e-05,
+      "loss": 0.0986,
+      "step": 9338
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 5.051068883610451e-05,
+      "loss": 0.1186,
+      "step": 9339
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 5.050475059382423e-05,
+      "loss": 0.1079,
+      "step": 9340
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 5.0498812351543946e-05,
+      "loss": 0.0993,
+      "step": 9341
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 5.049287410926367e-05,
+      "loss": 0.0889,
+      "step": 9342
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 5.0486935866983374e-05,
+      "loss": 0.0902,
+      "step": 9343
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 5.048099762470309e-05,
+      "loss": 0.1597,
+      "step": 9344
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 5.047505938242281e-05,
+      "loss": 0.1917,
+      "step": 9345
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 5.0469121140142514e-05,
+      "loss": 0.1007,
+      "step": 9346
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 5.0463182897862235e-05,
+      "loss": 0.111,
+      "step": 9347
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 5.045724465558195e-05,
+      "loss": 0.1001,
+      "step": 9348
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 5.045130641330167e-05,
+      "loss": 0.0884,
+      "step": 9349
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 5.0445368171021375e-05,
+      "loss": 0.1038,
+      "step": 9350
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 5.04394299287411e-05,
+      "loss": 0.1018,
+      "step": 9351
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 5.043349168646081e-05,
+      "loss": 0.115,
+      "step": 9352
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 5.042755344418052e-05,
+      "loss": 0.1429,
+      "step": 9353
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 5.042161520190024e-05,
+      "loss": 0.0872,
+      "step": 9354
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 5.041567695961995e-05,
+      "loss": 0.1348,
+      "step": 9355
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 5.040973871733967e-05,
+      "loss": 0.0448,
+      "step": 9356
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 5.0403800475059384e-05,
+      "loss": 0.1794,
+      "step": 9357
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 5.0397862232779104e-05,
+      "loss": 0.0808,
+      "step": 9358
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 5.039192399049881e-05,
+      "loss": 0.1034,
+      "step": 9359
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 5.038598574821853e-05,
+      "loss": 0.1341,
+      "step": 9360
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 5.0380047505938245e-05,
+      "loss": 0.1251,
+      "step": 9361
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 5.0374109263657965e-05,
+      "loss": 0.067,
+      "step": 9362
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 5.036817102137767e-05,
+      "loss": 0.0633,
+      "step": 9363
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 5.0362232779097386e-05,
+      "loss": 0.0765,
+      "step": 9364
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 5.0356294536817106e-05,
+      "loss": 0.0983,
+      "step": 9365
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 5.035035629453681e-05,
+      "loss": 0.0876,
+      "step": 9366
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 5.0344418052256533e-05,
+      "loss": 0.0462,
+      "step": 9367
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 5.033847980997625e-05,
+      "loss": 0.1295,
+      "step": 9368
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 5.033254156769597e-05,
+      "loss": 0.1353,
+      "step": 9369
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 5.032660332541568e-05,
+      "loss": 0.1151,
+      "step": 9370
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 5.03206650831354e-05,
+      "loss": 0.2499,
+      "step": 9371
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 5.031472684085511e-05,
+      "loss": 0.1485,
+      "step": 9372
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 5.030878859857483e-05,
+      "loss": 0.1854,
+      "step": 9373
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 5.030285035629454e-05,
+      "loss": 0.1394,
+      "step": 9374
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 5.029691211401425e-05,
+      "loss": 0.1107,
+      "step": 9375
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 5.029097387173397e-05,
+      "loss": 0.1211,
+      "step": 9376
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 5.028503562945368e-05,
+      "loss": 0.0901,
+      "step": 9377
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 5.02790973871734e-05,
+      "loss": 0.1066,
+      "step": 9378
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 5.027315914489311e-05,
+      "loss": 0.1365,
+      "step": 9379
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 5.026722090261283e-05,
+      "loss": 0.1058,
+      "step": 9380
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 5.0261282660332544e-05,
+      "loss": 0.1038,
+      "step": 9381
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 5.0255344418052264e-05,
+      "loss": 0.1191,
+      "step": 9382
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 5.024940617577197e-05,
+      "loss": 0.0791,
+      "step": 9383
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 5.0243467933491685e-05,
+      "loss": 0.0801,
+      "step": 9384
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 5.0237529691211405e-05,
+      "loss": 0.1973,
+      "step": 9385
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 5.023159144893111e-05,
+      "loss": 0.1103,
+      "step": 9386
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 5.022565320665084e-05,
+      "loss": 0.0967,
+      "step": 9387
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 5.0219714964370546e-05,
+      "loss": 0.1046,
+      "step": 9388
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 5.0213776722090266e-05,
+      "loss": 0.1152,
+      "step": 9389
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 5.020783847980998e-05,
+      "loss": 0.104,
+      "step": 9390
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 5.02019002375297e-05,
+      "loss": 0.1372,
+      "step": 9391
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 5.019596199524941e-05,
+      "loss": 0.0838,
+      "step": 9392
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 5.019002375296913e-05,
+      "loss": 0.0751,
+      "step": 9393
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 5.018408551068884e-05,
+      "loss": 0.0739,
+      "step": 9394
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 5.017814726840855e-05,
+      "loss": 0.0842,
+      "step": 9395
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 5.017220902612827e-05,
+      "loss": 0.0864,
+      "step": 9396
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 5.016627078384798e-05,
+      "loss": 0.065,
+      "step": 9397
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 5.01603325415677e-05,
+      "loss": 0.1212,
+      "step": 9398
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 5.015439429928741e-05,
+      "loss": 0.0902,
+      "step": 9399
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 5.014845605700713e-05,
+      "loss": 0.1009,
+      "step": 9400
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 5.014251781472684e-05,
+      "loss": 0.1039,
+      "step": 9401
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 5.013657957244656e-05,
+      "loss": 0.1284,
+      "step": 9402
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 5.013064133016627e-05,
+      "loss": 0.1074,
+      "step": 9403
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 5.0124703087886e-05,
+      "loss": 0.1583,
+      "step": 9404
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 5.0118764845605704e-05,
+      "loss": 0.1426,
+      "step": 9405
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 5.011282660332541e-05,
+      "loss": 0.0909,
+      "step": 9406
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 5.010688836104514e-05,
+      "loss": 0.079,
+      "step": 9407
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 5.0100950118764844e-05,
+      "loss": 0.0651,
+      "step": 9408
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 5.0095011876484565e-05,
+      "loss": 0.0665,
+      "step": 9409
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 5.008907363420428e-05,
+      "loss": 0.2671,
+      "step": 9410
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 5.0083135391924e-05,
+      "loss": 0.0738,
+      "step": 9411
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 5.0077197149643706e-05,
+      "loss": 0.0619,
+      "step": 9412
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 5.0071258907363426e-05,
+      "loss": 0.1158,
+      "step": 9413
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 5.006532066508314e-05,
+      "loss": 0.0712,
+      "step": 9414
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 5.0059382422802846e-05,
+      "loss": 0.0712,
+      "step": 9415
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 5.005344418052257e-05,
+      "loss": 0.0825,
+      "step": 9416
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 5.004750593824228e-05,
+      "loss": 0.0842,
+      "step": 9417
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 5.0041567695962e-05,
+      "loss": 0.121,
+      "step": 9418
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 5.003562945368171e-05,
+      "loss": 0.1267,
+      "step": 9419
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 5.002969121140143e-05,
+      "loss": 0.0432,
+      "step": 9420
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 5.002375296912114e-05,
+      "loss": 0.2749,
+      "step": 9421
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 5.001781472684086e-05,
+      "loss": 0.1756,
+      "step": 9422
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 5.001187648456057e-05,
+      "loss": 0.1656,
+      "step": 9423
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 5.0005938242280296e-05,
+      "loss": 0.1169,
+      "step": 9424
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 5e-05,
+      "loss": 0.1284,
+      "step": 9425
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.9994061757719716e-05,
+      "loss": 0.0879,
+      "step": 9426
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.9988123515439436e-05,
+      "loss": 0.1757,
+      "step": 9427
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.998218527315915e-05,
+      "loss": 0.1594,
+      "step": 9428
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.9976247030878863e-05,
+      "loss": 0.1129,
+      "step": 9429
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.997030878859858e-05,
+      "loss": 0.1662,
+      "step": 9430
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.996437054631829e-05,
+      "loss": 0.1603,
+      "step": 9431
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.9958432304038004e-05,
+      "loss": 0.1405,
+      "step": 9432
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.995249406175772e-05,
+      "loss": 0.1431,
+      "step": 9433
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.994655581947744e-05,
+      "loss": 0.0816,
+      "step": 9434
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.994061757719715e-05,
+      "loss": 0.1208,
+      "step": 9435
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.9934679334916865e-05,
+      "loss": 0.7819,
+      "step": 9436
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.9928741092636586e-05,
+      "loss": 0.1137,
+      "step": 9437
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.99228028503563e-05,
+      "loss": 0.1524,
+      "step": 9438
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.991686460807601e-05,
+      "loss": 0.1863,
+      "step": 9439
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.9910926365795726e-05,
+      "loss": 0.1084,
+      "step": 9440
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.990498812351544e-05,
+      "loss": 0.0893,
+      "step": 9441
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.9899049881235154e-05,
+      "loss": 0.1101,
+      "step": 9442
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.989311163895487e-05,
+      "loss": 0.1068,
+      "step": 9443
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.988717339667459e-05,
+      "loss": 0.1239,
+      "step": 9444
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.98812351543943e-05,
+      "loss": 0.1032,
+      "step": 9445
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.9875296912114015e-05,
+      "loss": 0.0916,
+      "step": 9446
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.9869358669833735e-05,
+      "loss": 0.0949,
+      "step": 9447
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.986342042755345e-05,
+      "loss": 0.0749,
+      "step": 9448
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.985748218527316e-05,
+      "loss": 0.1412,
+      "step": 9449
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.9851543942992876e-05,
+      "loss": 0.1047,
+      "step": 9450
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.984560570071259e-05,
+      "loss": 0.1203,
+      "step": 9451
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.98396674584323e-05,
+      "loss": 0.1049,
+      "step": 9452
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.9833729216152017e-05,
+      "loss": 0.1194,
+      "step": 9453
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.982779097387174e-05,
+      "loss": 0.0852,
+      "step": 9454
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.982185273159145e-05,
+      "loss": 0.0979,
+      "step": 9455
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.9815914489311164e-05,
+      "loss": 0.1081,
+      "step": 9456
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.9809976247030884e-05,
+      "loss": 0.0807,
+      "step": 9457
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.98040380047506e-05,
+      "loss": 0.0972,
+      "step": 9458
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.979809976247031e-05,
+      "loss": 0.0444,
+      "step": 9459
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.9792161520190025e-05,
+      "loss": 0.1088,
+      "step": 9460
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.9786223277909746e-05,
+      "loss": 0.1098,
+      "step": 9461
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.978028503562945e-05,
+      "loss": 0.0471,
+      "step": 9462
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.9774346793349166e-05,
+      "loss": 0.0857,
+      "step": 9463
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.9768408551068886e-05,
+      "loss": 0.0499,
+      "step": 9464
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.97624703087886e-05,
+      "loss": 0.0811,
+      "step": 9465
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.9756532066508313e-05,
+      "loss": 0.1307,
+      "step": 9466
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.9750593824228034e-05,
+      "loss": 0.0934,
+      "step": 9467
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.974465558194775e-05,
+      "loss": 0.09,
+      "step": 9468
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.973871733966746e-05,
+      "loss": 0.1256,
+      "step": 9469
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.9732779097387175e-05,
+      "loss": 0.1025,
+      "step": 9470
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.9726840855106895e-05,
+      "loss": 0.1698,
+      "step": 9471
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.97209026128266e-05,
+      "loss": 0.1702,
+      "step": 9472
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.9714964370546315e-05,
+      "loss": 0.1029,
+      "step": 9473
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.9709026128266036e-05,
+      "loss": 0.1308,
+      "step": 9474
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.970308788598575e-05,
+      "loss": 0.1595,
+      "step": 9475
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.969714964370546e-05,
+      "loss": 0.0722,
+      "step": 9476
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.969121140142518e-05,
+      "loss": 0.2,
+      "step": 9477
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.96852731591449e-05,
+      "loss": 0.1016,
+      "step": 9478
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.967933491686461e-05,
+      "loss": 0.1492,
+      "step": 9479
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.9673396674584324e-05,
+      "loss": 0.1253,
+      "step": 9480
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.9667458432304044e-05,
+      "loss": 0.103,
+      "step": 9481
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.966152019002376e-05,
+      "loss": 0.1249,
+      "step": 9482
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.965558194774347e-05,
+      "loss": 0.0842,
+      "step": 9483
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.9649643705463185e-05,
+      "loss": 0.0982,
+      "step": 9484
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.96437054631829e-05,
+      "loss": 0.1288,
+      "step": 9485
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.963776722090261e-05,
+      "loss": 0.1219,
+      "step": 9486
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.963182897862233e-05,
+      "loss": 0.142,
+      "step": 9487
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.9625890736342046e-05,
+      "loss": 0.0998,
+      "step": 9488
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.961995249406176e-05,
+      "loss": 0.0759,
+      "step": 9489
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.961401425178147e-05,
+      "loss": 0.08,
+      "step": 9490
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.9608076009501194e-05,
+      "loss": 0.1866,
+      "step": 9491
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.960213776722091e-05,
+      "loss": 0.1153,
+      "step": 9492
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.959619952494062e-05,
+      "loss": 0.1653,
+      "step": 9493
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.9590261282660334e-05,
+      "loss": 0.0695,
+      "step": 9494
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.958432304038005e-05,
+      "loss": 0.0713,
+      "step": 9495
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.957838479809976e-05,
+      "loss": 0.0822,
+      "step": 9496
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.957244655581948e-05,
+      "loss": 0.0748,
+      "step": 9497
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.9566508313539195e-05,
+      "loss": 0.1382,
+      "step": 9498
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.956057007125891e-05,
+      "loss": 0.0893,
+      "step": 9499
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.955463182897863e-05,
+      "loss": 0.0643,
+      "step": 9500
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.954869358669834e-05,
+      "loss": 0.0826,
+      "step": 9501
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.9542755344418057e-05,
+      "loss": 0.0738,
+      "step": 9502
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.953681710213777e-05,
+      "loss": 0.1173,
+      "step": 9503
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.9530878859857484e-05,
+      "loss": 0.1231,
+      "step": 9504
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.95249406175772e-05,
+      "loss": 0.0991,
+      "step": 9505
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.951900237529691e-05,
+      "loss": 0.0866,
+      "step": 9506
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.951306413301663e-05,
+      "loss": 0.1151,
+      "step": 9507
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.9507125890736345e-05,
+      "loss": 0.0929,
+      "step": 9508
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.950118764845606e-05,
+      "loss": 0.086,
+      "step": 9509
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.949524940617578e-05,
+      "loss": 0.084,
+      "step": 9510
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.948931116389549e-05,
+      "loss": 0.0905,
+      "step": 9511
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.9483372921615206e-05,
+      "loss": 0.1122,
+      "step": 9512
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.947743467933492e-05,
+      "loss": 0.0653,
+      "step": 9513
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.947149643705463e-05,
+      "loss": 0.079,
+      "step": 9514
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.946555819477435e-05,
+      "loss": 0.0855,
+      "step": 9515
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.945961995249406e-05,
+      "loss": 0.0461,
+      "step": 9516
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.945368171021378e-05,
+      "loss": 0.0839,
+      "step": 9517
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.9447743467933494e-05,
+      "loss": 0.0387,
+      "step": 9518
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.944180522565321e-05,
+      "loss": 0.1019,
+      "step": 9519
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.943586698337293e-05,
+      "loss": 0.0934,
+      "step": 9520
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.942992874109264e-05,
+      "loss": 0.1904,
+      "step": 9521
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.9423990498812355e-05,
+      "loss": 0.1799,
+      "step": 9522
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.941805225653207e-05,
+      "loss": 0.1623,
+      "step": 9523
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.941211401425178e-05,
+      "loss": 0.1407,
+      "step": 9524
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.9406175771971496e-05,
+      "loss": 0.1185,
+      "step": 9525
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.940023752969121e-05,
+      "loss": 0.1263,
+      "step": 9526
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.939429928741093e-05,
+      "loss": 0.1555,
+      "step": 9527
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.9388361045130644e-05,
+      "loss": 0.1301,
+      "step": 9528
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.938242280285036e-05,
+      "loss": 0.1493,
+      "step": 9529
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.937648456057008e-05,
+      "loss": 0.1341,
+      "step": 9530
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.937054631828979e-05,
+      "loss": 0.0752,
+      "step": 9531
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.9364608076009505e-05,
+      "loss": 0.1366,
+      "step": 9532
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.935866983372922e-05,
+      "loss": 0.1135,
+      "step": 9533
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.935273159144893e-05,
+      "loss": 0.1086,
+      "step": 9534
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.9346793349168645e-05,
+      "loss": 0.064,
+      "step": 9535
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.934085510688836e-05,
+      "loss": 0.0946,
+      "step": 9536
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.933491686460808e-05,
+      "loss": 0.16,
+      "step": 9537
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.932897862232779e-05,
+      "loss": 0.1161,
+      "step": 9538
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.9323040380047506e-05,
+      "loss": 0.0968,
+      "step": 9539
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.931710213776723e-05,
+      "loss": 0.0777,
+      "step": 9540
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.931116389548694e-05,
+      "loss": 0.1191,
+      "step": 9541
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.9305225653206654e-05,
+      "loss": 0.1158,
+      "step": 9542
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.929928741092637e-05,
+      "loss": 0.0627,
+      "step": 9543
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.929334916864609e-05,
+      "loss": 0.0953,
+      "step": 9544
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.9287410926365795e-05,
+      "loss": 0.0869,
+      "step": 9545
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.928147268408551e-05,
+      "loss": 0.1591,
+      "step": 9546
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.927553444180523e-05,
+      "loss": 0.0901,
+      "step": 9547
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.926959619952494e-05,
+      "loss": 0.1093,
+      "step": 9548
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.9263657957244656e-05,
+      "loss": 0.1344,
+      "step": 9549
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.9257719714964376e-05,
+      "loss": 0.1287,
+      "step": 9550
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.925178147268409e-05,
+      "loss": 0.0852,
+      "step": 9551
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.92458432304038e-05,
+      "loss": 0.1075,
+      "step": 9552
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.923990498812352e-05,
+      "loss": 0.0986,
+      "step": 9553
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.923396674584324e-05,
+      "loss": 0.0814,
+      "step": 9554
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.9228028503562944e-05,
+      "loss": 0.074,
+      "step": 9555
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.922209026128266e-05,
+      "loss": 0.1001,
+      "step": 9556
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.921615201900238e-05,
+      "loss": 0.0603,
+      "step": 9557
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.921021377672209e-05,
+      "loss": 0.0917,
+      "step": 9558
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.9204275534441805e-05,
+      "loss": 0.0557,
+      "step": 9559
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.9198337292161526e-05,
+      "loss": 0.067,
+      "step": 9560
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.919239904988124e-05,
+      "loss": 0.0864,
+      "step": 9561
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.918646080760095e-05,
+      "loss": 0.079,
+      "step": 9562
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.9180522565320666e-05,
+      "loss": 0.0742,
+      "step": 9563
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.917458432304039e-05,
+      "loss": 0.0609,
+      "step": 9564
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.9168646080760093e-05,
+      "loss": 0.0559,
+      "step": 9565
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.916270783847981e-05,
+      "loss": 0.0991,
+      "step": 9566
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.915676959619953e-05,
+      "loss": 0.1201,
+      "step": 9567
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.915083135391924e-05,
+      "loss": 0.083,
+      "step": 9568
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.9144893111638955e-05,
+      "loss": 0.0587,
+      "step": 9569
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.9138954869358675e-05,
+      "loss": 0.1376,
+      "step": 9570
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.913301662707839e-05,
+      "loss": 0.4534,
+      "step": 9571
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.91270783847981e-05,
+      "loss": 0.1499,
+      "step": 9572
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.9121140142517816e-05,
+      "loss": 0.1619,
+      "step": 9573
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.9115201900237536e-05,
+      "loss": 0.188,
+      "step": 9574
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.910926365795725e-05,
+      "loss": 0.1772,
+      "step": 9575
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.9103325415676956e-05,
+      "loss": 0.1401,
+      "step": 9576
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.909738717339668e-05,
+      "loss": 0.1232,
+      "step": 9577
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.909144893111639e-05,
+      "loss": 0.0804,
+      "step": 9578
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.9085510688836104e-05,
+      "loss": 0.1233,
+      "step": 9579
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.9079572446555824e-05,
+      "loss": 0.1288,
+      "step": 9580
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.907363420427554e-05,
+      "loss": 0.0703,
+      "step": 9581
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.906769596199525e-05,
+      "loss": 0.1091,
+      "step": 9582
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.9061757719714965e-05,
+      "loss": 0.1271,
+      "step": 9583
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.9055819477434685e-05,
+      "loss": 0.1082,
+      "step": 9584
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.90498812351544e-05,
+      "loss": 0.1117,
+      "step": 9585
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.9043942992874106e-05,
+      "loss": 0.097,
+      "step": 9586
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.9038004750593826e-05,
+      "loss": 0.1069,
+      "step": 9587
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.903206650831354e-05,
+      "loss": 0.1157,
+      "step": 9588
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.902612826603325e-05,
+      "loss": 0.1059,
+      "step": 9589
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.9020190023752974e-05,
+      "loss": 0.0993,
+      "step": 9590
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.901425178147269e-05,
+      "loss": 0.0893,
+      "step": 9591
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.90083135391924e-05,
+      "loss": 0.0742,
+      "step": 9592
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.9002375296912114e-05,
+      "loss": 0.1287,
+      "step": 9593
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.8996437054631835e-05,
+      "loss": 0.1744,
+      "step": 9594
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.899049881235155e-05,
+      "loss": 0.1197,
+      "step": 9595
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.898456057007126e-05,
+      "loss": 0.0689,
+      "step": 9596
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.8978622327790975e-05,
+      "loss": 0.0852,
+      "step": 9597
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.897268408551069e-05,
+      "loss": 0.0834,
+      "step": 9598
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.89667458432304e-05,
+      "loss": 0.1178,
+      "step": 9599
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.896080760095012e-05,
+      "loss": 0.0876,
+      "step": 9600
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.8954869358669837e-05,
+      "loss": 0.0524,
+      "step": 9601
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.894893111638955e-05,
+      "loss": 0.0794,
+      "step": 9602
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.8942992874109264e-05,
+      "loss": 0.0815,
+      "step": 9603
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.8937054631828984e-05,
+      "loss": 0.0577,
+      "step": 9604
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.89311163895487e-05,
+      "loss": 0.098,
+      "step": 9605
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.892517814726841e-05,
+      "loss": 0.0572,
+      "step": 9606
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.8919239904988125e-05,
+      "loss": 0.1062,
+      "step": 9607
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.891330166270784e-05,
+      "loss": 0.1207,
+      "step": 9608
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.890736342042755e-05,
+      "loss": 0.2221,
+      "step": 9609
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.890142517814727e-05,
+      "loss": 0.0699,
+      "step": 9610
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.8895486935866986e-05,
+      "loss": 0.0941,
+      "step": 9611
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.88895486935867e-05,
+      "loss": 0.0821,
+      "step": 9612
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.888361045130641e-05,
+      "loss": 0.0563,
+      "step": 9613
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.8877672209026133e-05,
+      "loss": 0.0552,
+      "step": 9614
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.887173396674585e-05,
+      "loss": 0.103,
+      "step": 9615
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.886579572446556e-05,
+      "loss": 0.0375,
+      "step": 9616
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.8859857482185274e-05,
+      "loss": 0.345,
+      "step": 9617
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.885391923990499e-05,
+      "loss": 0.2031,
+      "step": 9618
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.88479809976247e-05,
+      "loss": 0.0812,
+      "step": 9619
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.884204275534442e-05,
+      "loss": 0.1798,
+      "step": 9620
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.8836104513064135e-05,
+      "loss": 0.1387,
+      "step": 9621
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.883016627078385e-05,
+      "loss": 0.1317,
+      "step": 9622
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.882422802850357e-05,
+      "loss": 0.1228,
+      "step": 9623
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.881828978622328e-05,
+      "loss": 0.0969,
+      "step": 9624
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.8812351543942996e-05,
+      "loss": 0.159,
+      "step": 9625
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.880641330166271e-05,
+      "loss": 0.1329,
+      "step": 9626
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.880047505938243e-05,
+      "loss": 0.1098,
+      "step": 9627
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.879453681710214e-05,
+      "loss": 0.0917,
+      "step": 9628
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.878859857482185e-05,
+      "loss": 0.1014,
+      "step": 9629
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.878266033254157e-05,
+      "loss": 0.1466,
+      "step": 9630
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.8776722090261285e-05,
+      "loss": 0.1514,
+      "step": 9631
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.8770783847981e-05,
+      "loss": 0.1922,
+      "step": 9632
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.876484560570072e-05,
+      "loss": 0.0852,
+      "step": 9633
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.875890736342043e-05,
+      "loss": 0.0679,
+      "step": 9634
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.8752969121140146e-05,
+      "loss": 0.0942,
+      "step": 9635
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.874703087885986e-05,
+      "loss": 0.0914,
+      "step": 9636
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.874109263657958e-05,
+      "loss": 0.1205,
+      "step": 9637
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.8735154394299286e-05,
+      "loss": 0.1315,
+      "step": 9638
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.8729216152019e-05,
+      "loss": 0.0958,
+      "step": 9639
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.872327790973872e-05,
+      "loss": 0.0874,
+      "step": 9640
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.8717339667458434e-05,
+      "loss": 0.112,
+      "step": 9641
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.871140142517815e-05,
+      "loss": 0.0781,
+      "step": 9642
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.870546318289787e-05,
+      "loss": 0.0861,
+      "step": 9643
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.869952494061758e-05,
+      "loss": 0.0802,
+      "step": 9644
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.8693586698337295e-05,
+      "loss": 0.1281,
+      "step": 9645
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.868764845605701e-05,
+      "loss": 0.1211,
+      "step": 9646
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.868171021377673e-05,
+      "loss": 0.0697,
+      "step": 9647
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.8675771971496436e-05,
+      "loss": 0.121,
+      "step": 9648
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.866983372921615e-05,
+      "loss": 0.08,
+      "step": 9649
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.866389548693587e-05,
+      "loss": 0.1099,
+      "step": 9650
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.865795724465558e-05,
+      "loss": 0.099,
+      "step": 9651
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.86520190023753e-05,
+      "loss": 0.1195,
+      "step": 9652
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.864608076009502e-05,
+      "loss": 0.1028,
+      "step": 9653
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.864014251781473e-05,
+      "loss": 0.0918,
+      "step": 9654
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.8634204275534444e-05,
+      "loss": 0.0769,
+      "step": 9655
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.862826603325416e-05,
+      "loss": 0.0945,
+      "step": 9656
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.862232779097388e-05,
+      "loss": 0.1717,
+      "step": 9657
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.861638954869359e-05,
+      "loss": 0.0623,
+      "step": 9658
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.86104513064133e-05,
+      "loss": 0.0837,
+      "step": 9659
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.860451306413302e-05,
+      "loss": 0.0993,
+      "step": 9660
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.859857482185273e-05,
+      "loss": 0.0866,
+      "step": 9661
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.8592636579572446e-05,
+      "loss": 0.1196,
+      "step": 9662
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.858669833729217e-05,
+      "loss": 0.102,
+      "step": 9663
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.858076009501188e-05,
+      "loss": 0.0544,
+      "step": 9664
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.8574821852731594e-05,
+      "loss": 0.0554,
+      "step": 9665
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.856888361045131e-05,
+      "loss": 0.1063,
+      "step": 9666
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.856294536817103e-05,
+      "loss": 0.0384,
+      "step": 9667
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.855700712589074e-05,
+      "loss": 0.0668,
+      "step": 9668
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.855106888361045e-05,
+      "loss": 0.0747,
+      "step": 9669
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.854513064133017e-05,
+      "loss": 0.0775,
+      "step": 9670
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.853919239904988e-05,
+      "loss": 0.1715,
+      "step": 9671
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.8533254156769596e-05,
+      "loss": 0.2449,
+      "step": 9672
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.8527315914489316e-05,
+      "loss": 0.1749,
+      "step": 9673
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.852137767220903e-05,
+      "loss": 0.1887,
+      "step": 9674
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.851543942992874e-05,
+      "loss": 0.0883,
+      "step": 9675
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.850950118764846e-05,
+      "loss": 0.0984,
+      "step": 9676
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.850356294536818e-05,
+      "loss": 0.1269,
+      "step": 9677
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.849762470308789e-05,
+      "loss": 0.1025,
+      "step": 9678
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.8491686460807604e-05,
+      "loss": 0.1352,
+      "step": 9679
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.848574821852732e-05,
+      "loss": 0.1496,
+      "step": 9680
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.847980997624703e-05,
+      "loss": 0.0955,
+      "step": 9681
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.8473871733966745e-05,
+      "loss": 0.134,
+      "step": 9682
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.8467933491686465e-05,
+      "loss": 0.2134,
+      "step": 9683
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.846199524940618e-05,
+      "loss": 0.1206,
+      "step": 9684
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.845605700712589e-05,
+      "loss": 0.0647,
+      "step": 9685
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.8450118764845606e-05,
+      "loss": 0.1123,
+      "step": 9686
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.8444180522565326e-05,
+      "loss": 0.1253,
+      "step": 9687
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.843824228028504e-05,
+      "loss": 0.0897,
+      "step": 9688
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.8432304038004754e-05,
+      "loss": 0.1026,
+      "step": 9689
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.842636579572447e-05,
+      "loss": 0.0738,
+      "step": 9690
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.842042755344418e-05,
+      "loss": 0.1177,
+      "step": 9691
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.8414489311163894e-05,
+      "loss": 0.1239,
+      "step": 9692
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.8408551068883615e-05,
+      "loss": 0.0757,
+      "step": 9693
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.840261282660333e-05,
+      "loss": 0.0439,
+      "step": 9694
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.839667458432304e-05,
+      "loss": 0.0712,
+      "step": 9695
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.8390736342042755e-05,
+      "loss": 0.0688,
+      "step": 9696
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.8384798099762476e-05,
+      "loss": 0.133,
+      "step": 9697
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.837885985748219e-05,
+      "loss": 0.0854,
+      "step": 9698
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.83729216152019e-05,
+      "loss": 0.0496,
+      "step": 9699
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.8366983372921617e-05,
+      "loss": 0.0784,
+      "step": 9700
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.836104513064133e-05,
+      "loss": 0.1307,
+      "step": 9701
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.8355106888361044e-05,
+      "loss": 0.0953,
+      "step": 9702
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.8349168646080764e-05,
+      "loss": 0.0829,
+      "step": 9703
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.834323040380048e-05,
+      "loss": 0.0917,
+      "step": 9704
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.833729216152019e-05,
+      "loss": 0.0909,
+      "step": 9705
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.8331353919239905e-05,
+      "loss": 0.1371,
+      "step": 9706
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.8325415676959625e-05,
+      "loss": 0.0945,
+      "step": 9707
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.831947743467934e-05,
+      "loss": 0.0538,
+      "step": 9708
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.831353919239905e-05,
+      "loss": 0.0477,
+      "step": 9709
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.830760095011877e-05,
+      "loss": 0.1061,
+      "step": 9710
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.830166270783848e-05,
+      "loss": 0.078,
+      "step": 9711
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.829572446555819e-05,
+      "loss": 0.1337,
+      "step": 9712
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.8289786223277913e-05,
+      "loss": 0.095,
+      "step": 9713
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.828384798099763e-05,
+      "loss": 0.0713,
+      "step": 9714
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.827790973871734e-05,
+      "loss": 0.0477,
+      "step": 9715
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.8271971496437054e-05,
+      "loss": 0.1002,
+      "step": 9716
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.8266033254156775e-05,
+      "loss": 0.1203,
+      "step": 9717
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.826009501187649e-05,
+      "loss": 0.1105,
+      "step": 9718
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.82541567695962e-05,
+      "loss": 0.0584,
+      "step": 9719
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.824821852731592e-05,
+      "loss": 0.1513,
+      "step": 9720
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.824228028503563e-05,
+      "loss": 0.3743,
+      "step": 9721
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.823634204275534e-05,
+      "loss": 0.195,
+      "step": 9722
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.823040380047506e-05,
+      "loss": 0.2103,
+      "step": 9723
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.8224465558194776e-05,
+      "loss": 0.1387,
+      "step": 9724
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.821852731591449e-05,
+      "loss": 0.126,
+      "step": 9725
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.8212589073634204e-05,
+      "loss": 0.1261,
+      "step": 9726
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.8206650831353924e-05,
+      "loss": 0.1274,
+      "step": 9727
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.820071258907364e-05,
+      "loss": 0.1754,
+      "step": 9728
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.819477434679335e-05,
+      "loss": 0.1285,
+      "step": 9729
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.818883610451307e-05,
+      "loss": 0.2116,
+      "step": 9730
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.818289786223278e-05,
+      "loss": 0.1057,
+      "step": 9731
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.817695961995249e-05,
+      "loss": 0.1044,
+      "step": 9732
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.817102137767221e-05,
+      "loss": 0.1239,
+      "step": 9733
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.8165083135391926e-05,
+      "loss": 0.0982,
+      "step": 9734
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.815914489311164e-05,
+      "loss": 0.1298,
+      "step": 9735
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.815320665083135e-05,
+      "loss": 0.0861,
+      "step": 9736
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.814726840855107e-05,
+      "loss": 0.0813,
+      "step": 9737
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.814133016627079e-05,
+      "loss": 0.0813,
+      "step": 9738
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.81353919239905e-05,
+      "loss": 0.188,
+      "step": 9739
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.812945368171022e-05,
+      "loss": 0.0828,
+      "step": 9740
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.8123515439429934e-05,
+      "loss": 0.0865,
+      "step": 9741
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.811757719714964e-05,
+      "loss": 0.1671,
+      "step": 9742
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.811163895486936e-05,
+      "loss": 0.1205,
+      "step": 9743
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.8105700712589075e-05,
+      "loss": 0.0835,
+      "step": 9744
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.809976247030879e-05,
+      "loss": 0.0709,
+      "step": 9745
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.80938242280285e-05,
+      "loss": 0.0841,
+      "step": 9746
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.808788598574822e-05,
+      "loss": 0.1126,
+      "step": 9747
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.8081947743467936e-05,
+      "loss": 0.0592,
+      "step": 9748
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.807600950118765e-05,
+      "loss": 0.1065,
+      "step": 9749
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.807007125890737e-05,
+      "loss": 0.1054,
+      "step": 9750
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.8064133016627084e-05,
+      "loss": 0.1112,
+      "step": 9751
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.805819477434679e-05,
+      "loss": 0.0613,
+      "step": 9752
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.805225653206651e-05,
+      "loss": 0.1071,
+      "step": 9753
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.8046318289786224e-05,
+      "loss": 0.1289,
+      "step": 9754
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.804038004750594e-05,
+      "loss": 0.0832,
+      "step": 9755
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.803444180522566e-05,
+      "loss": 0.0423,
+      "step": 9756
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.802850356294537e-05,
+      "loss": 0.0544,
+      "step": 9757
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.8022565320665086e-05,
+      "loss": 0.149,
+      "step": 9758
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.80166270783848e-05,
+      "loss": 0.0651,
+      "step": 9759
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.801068883610452e-05,
+      "loss": 0.0821,
+      "step": 9760
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.800475059382423e-05,
+      "loss": 0.0815,
+      "step": 9761
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.799881235154395e-05,
+      "loss": 0.1044,
+      "step": 9762
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.799287410926366e-05,
+      "loss": 0.1661,
+      "step": 9763
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.7986935866983374e-05,
+      "loss": 0.0663,
+      "step": 9764
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.798099762470309e-05,
+      "loss": 0.0603,
+      "step": 9765
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.797505938242281e-05,
+      "loss": 0.0658,
+      "step": 9766
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.796912114014252e-05,
+      "loss": 0.0998,
+      "step": 9767
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.7963182897862235e-05,
+      "loss": 0.0829,
+      "step": 9768
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.795724465558195e-05,
+      "loss": 0.0385,
+      "step": 9769
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.795130641330167e-05,
+      "loss": 0.0864,
+      "step": 9770
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.794536817102138e-05,
+      "loss": 0.1901,
+      "step": 9771
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.7939429928741096e-05,
+      "loss": 0.1363,
+      "step": 9772
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.793349168646081e-05,
+      "loss": 0.1534,
+      "step": 9773
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.792755344418052e-05,
+      "loss": 0.154,
+      "step": 9774
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.792161520190024e-05,
+      "loss": 0.2441,
+      "step": 9775
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.791567695961996e-05,
+      "loss": 0.1303,
+      "step": 9776
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.790973871733967e-05,
+      "loss": 0.1892,
+      "step": 9777
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.7903800475059384e-05,
+      "loss": 0.1523,
+      "step": 9778
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.78978622327791e-05,
+      "loss": 0.0905,
+      "step": 9779
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.789192399049882e-05,
+      "loss": 0.1162,
+      "step": 9780
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.788598574821853e-05,
+      "loss": 0.0799,
+      "step": 9781
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.7880047505938245e-05,
+      "loss": 0.0896,
+      "step": 9782
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.787410926365796e-05,
+      "loss": 0.0481,
+      "step": 9783
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.786817102137767e-05,
+      "loss": 0.1031,
+      "step": 9784
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.7862232779097386e-05,
+      "loss": 0.1518,
+      "step": 9785
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.7856294536817106e-05,
+      "loss": 0.1171,
+      "step": 9786
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.785035629453682e-05,
+      "loss": 0.0946,
+      "step": 9787
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.7844418052256534e-05,
+      "loss": 0.098,
+      "step": 9788
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.783847980997625e-05,
+      "loss": 0.1774,
+      "step": 9789
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.783254156769597e-05,
+      "loss": 0.1333,
+      "step": 9790
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.782660332541568e-05,
+      "loss": 0.1119,
+      "step": 9791
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.7820665083135395e-05,
+      "loss": 0.1166,
+      "step": 9792
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.7814726840855115e-05,
+      "loss": 0.0829,
+      "step": 9793
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.780878859857482e-05,
+      "loss": 0.1273,
+      "step": 9794
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.7802850356294535e-05,
+      "loss": 0.1096,
+      "step": 9795
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.7796912114014256e-05,
+      "loss": 0.0655,
+      "step": 9796
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.779097387173397e-05,
+      "loss": 0.0638,
+      "step": 9797
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.778503562945368e-05,
+      "loss": 0.1092,
+      "step": 9798
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.7779097387173397e-05,
+      "loss": 0.1093,
+      "step": 9799
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.777315914489312e-05,
+      "loss": 0.0534,
+      "step": 9800
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.776722090261283e-05,
+      "loss": 0.0648,
+      "step": 9801
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.7761282660332544e-05,
+      "loss": 0.0711,
+      "step": 9802
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.7755344418052264e-05,
+      "loss": 0.0853,
+      "step": 9803
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.774940617577197e-05,
+      "loss": 0.0783,
+      "step": 9804
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.7743467933491685e-05,
+      "loss": 0.0951,
+      "step": 9805
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.7737529691211405e-05,
+      "loss": 0.0835,
+      "step": 9806
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.773159144893112e-05,
+      "loss": 0.0436,
+      "step": 9807
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.772565320665083e-05,
+      "loss": 0.094,
+      "step": 9808
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.7719714964370546e-05,
+      "loss": 0.0574,
+      "step": 9809
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.7713776722090266e-05,
+      "loss": 0.0419,
+      "step": 9810
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.770783847980998e-05,
+      "loss": 0.1225,
+      "step": 9811
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.7701900237529693e-05,
+      "loss": 0.0882,
+      "step": 9812
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.7695961995249414e-05,
+      "loss": 0.142,
+      "step": 9813
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.769002375296912e-05,
+      "loss": 0.2316,
+      "step": 9814
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.7684085510688834e-05,
+      "loss": 0.0971,
+      "step": 9815
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.7678147268408555e-05,
+      "loss": 0.0927,
+      "step": 9816
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.767220902612827e-05,
+      "loss": 0.1127,
+      "step": 9817
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.766627078384798e-05,
+      "loss": 0.0725,
+      "step": 9818
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.7660332541567695e-05,
+      "loss": 0.0905,
+      "step": 9819
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.7654394299287416e-05,
+      "loss": 0.1118,
+      "step": 9820
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.764845605700713e-05,
+      "loss": 0.1032,
+      "step": 9821
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.764251781472684e-05,
+      "loss": 0.0751,
+      "step": 9822
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.763657957244656e-05,
+      "loss": 0.1108,
+      "step": 9823
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.763064133016628e-05,
+      "loss": 0.0985,
+      "step": 9824
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.7624703087885984e-05,
+      "loss": 0.0811,
+      "step": 9825
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.7618764845605704e-05,
+      "loss": 0.0764,
+      "step": 9826
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.761282660332542e-05,
+      "loss": 0.0783,
+      "step": 9827
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.760688836104513e-05,
+      "loss": 0.0755,
+      "step": 9828
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.7600950118764845e-05,
+      "loss": 0.1204,
+      "step": 9829
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.7595011876484565e-05,
+      "loss": 0.0866,
+      "step": 9830
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.758907363420428e-05,
+      "loss": 0.0723,
+      "step": 9831
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.758313539192399e-05,
+      "loss": 0.0902,
+      "step": 9832
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.757719714964371e-05,
+      "loss": 0.0517,
+      "step": 9833
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.7571258907363426e-05,
+      "loss": 0.0813,
+      "step": 9834
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.756532066508313e-05,
+      "loss": 0.1822,
+      "step": 9835
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.755938242280285e-05,
+      "loss": 0.0579,
+      "step": 9836
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.755344418052257e-05,
+      "loss": 0.0637,
+      "step": 9837
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.754750593824228e-05,
+      "loss": 0.1419,
+      "step": 9838
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.7541567695961994e-05,
+      "loss": 0.0632,
+      "step": 9839
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.7535629453681714e-05,
+      "loss": 0.0669,
+      "step": 9840
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.752969121140143e-05,
+      "loss": 0.0579,
+      "step": 9841
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.752375296912114e-05,
+      "loss": 0.0949,
+      "step": 9842
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.751781472684086e-05,
+      "loss": 0.0558,
+      "step": 9843
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.7511876484560575e-05,
+      "loss": 0.0794,
+      "step": 9844
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.750593824228028e-05,
+      "loss": 0.1111,
+      "step": 9845
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.75e-05,
+      "loss": 0.0587,
+      "step": 9846
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.7494061757719716e-05,
+      "loss": 0.0595,
+      "step": 9847
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.748812351543943e-05,
+      "loss": 0.0804,
+      "step": 9848
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.748218527315914e-05,
+      "loss": 0.0366,
+      "step": 9849
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.7476247030878864e-05,
+      "loss": 0.0615,
+      "step": 9850
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.747030878859858e-05,
+      "loss": 0.0846,
+      "step": 9851
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.746437054631829e-05,
+      "loss": 0.1098,
+      "step": 9852
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.745843230403801e-05,
+      "loss": 0.0527,
+      "step": 9853
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.7452494061757725e-05,
+      "loss": 0.0819,
+      "step": 9854
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.744655581947744e-05,
+      "loss": 0.2181,
+      "step": 9855
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.744061757719715e-05,
+      "loss": 0.0609,
+      "step": 9856
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.7434679334916866e-05,
+      "loss": 0.0643,
+      "step": 9857
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.742874109263658e-05,
+      "loss": 0.0532,
+      "step": 9858
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.742280285035629e-05,
+      "loss": 0.1006,
+      "step": 9859
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.741686460807601e-05,
+      "loss": 0.0616,
+      "step": 9860
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.741092636579573e-05,
+      "loss": 0.0876,
+      "step": 9861
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.740498812351544e-05,
+      "loss": 0.1574,
+      "step": 9862
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.739904988123516e-05,
+      "loss": 0.1363,
+      "step": 9863
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.7393111638954874e-05,
+      "loss": 0.1164,
+      "step": 9864
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.738717339667459e-05,
+      "loss": 0.1302,
+      "step": 9865
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.73812351543943e-05,
+      "loss": 0.1355,
+      "step": 9866
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.7375296912114015e-05,
+      "loss": 0.107,
+      "step": 9867
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.736935866983373e-05,
+      "loss": 0.0668,
+      "step": 9868
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.736342042755344e-05,
+      "loss": 0.071,
+      "step": 9869
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.735748218527316e-05,
+      "loss": 0.1055,
+      "step": 9870
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.7351543942992876e-05,
+      "loss": 0.0717,
+      "step": 9871
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.734560570071259e-05,
+      "loss": 0.1026,
+      "step": 9872
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.733966745843231e-05,
+      "loss": 0.0784,
+      "step": 9873
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.7333729216152024e-05,
+      "loss": 0.1025,
+      "step": 9874
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.732779097387174e-05,
+      "loss": 0.0918,
+      "step": 9875
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.732185273159145e-05,
+      "loss": 0.0847,
+      "step": 9876
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.7315914489311164e-05,
+      "loss": 0.0701,
+      "step": 9877
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.730997624703088e-05,
+      "loss": 0.0814,
+      "step": 9878
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.73040380047506e-05,
+      "loss": 0.0896,
+      "step": 9879
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.729809976247031e-05,
+      "loss": 0.092,
+      "step": 9880
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.7292161520190025e-05,
+      "loss": 0.1054,
+      "step": 9881
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.728622327790974e-05,
+      "loss": 0.0735,
+      "step": 9882
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.728028503562946e-05,
+      "loss": 0.1081,
+      "step": 9883
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.727434679334917e-05,
+      "loss": 0.0705,
+      "step": 9884
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.7268408551068886e-05,
+      "loss": 0.0823,
+      "step": 9885
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.72624703087886e-05,
+      "loss": 0.1398,
+      "step": 9886
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.7256532066508314e-05,
+      "loss": 0.1501,
+      "step": 9887
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.725059382422803e-05,
+      "loss": 0.0886,
+      "step": 9888
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.724465558194775e-05,
+      "loss": 0.0703,
+      "step": 9889
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.723871733966746e-05,
+      "loss": 0.0644,
+      "step": 9890
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.7232779097387175e-05,
+      "loss": 0.1088,
+      "step": 9891
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.722684085510689e-05,
+      "loss": 0.0753,
+      "step": 9892
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.722090261282661e-05,
+      "loss": 0.0788,
+      "step": 9893
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.721496437054632e-05,
+      "loss": 0.0983,
+      "step": 9894
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.7209026128266036e-05,
+      "loss": 0.065,
+      "step": 9895
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.7203087885985756e-05,
+      "loss": 0.0663,
+      "step": 9896
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.719714964370546e-05,
+      "loss": 0.0802,
+      "step": 9897
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.7191211401425177e-05,
+      "loss": 0.1665,
+      "step": 9898
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.71852731591449e-05,
+      "loss": 0.0553,
+      "step": 9899
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.717933491686461e-05,
+      "loss": 0.0449,
+      "step": 9900
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.7173396674584324e-05,
+      "loss": 0.043,
+      "step": 9901
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.716745843230404e-05,
+      "loss": 0.0407,
+      "step": 9902
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.716152019002376e-05,
+      "loss": 0.0926,
+      "step": 9903
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.715558194774347e-05,
+      "loss": 0.0823,
+      "step": 9904
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.7149643705463185e-05,
+      "loss": 0.0411,
+      "step": 9905
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.7143705463182906e-05,
+      "loss": 0.0933,
+      "step": 9906
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.713776722090262e-05,
+      "loss": 0.0637,
+      "step": 9907
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.7131828978622326e-05,
+      "loss": 0.0378,
+      "step": 9908
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.7125890736342046e-05,
+      "loss": 0.0316,
+      "step": 9909
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.711995249406176e-05,
+      "loss": 0.0476,
+      "step": 9910
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.7114014251781473e-05,
+      "loss": 0.0973,
+      "step": 9911
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.710807600950119e-05,
+      "loss": 0.1534,
+      "step": 9912
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.710213776722091e-05,
+      "loss": 0.2011,
+      "step": 9913
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.709619952494062e-05,
+      "loss": 0.1502,
+      "step": 9914
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.7090261282660335e-05,
+      "loss": 0.1433,
+      "step": 9915
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.7084323040380055e-05,
+      "loss": 0.0806,
+      "step": 9916
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.707838479809977e-05,
+      "loss": 0.0682,
+      "step": 9917
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.7072446555819475e-05,
+      "loss": 0.0621,
+      "step": 9918
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.7066508313539196e-05,
+      "loss": 0.0959,
+      "step": 9919
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.706057007125891e-05,
+      "loss": 0.0615,
+      "step": 9920
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.705463182897862e-05,
+      "loss": 0.0892,
+      "step": 9921
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.7048693586698336e-05,
+      "loss": 0.101,
+      "step": 9922
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.704275534441806e-05,
+      "loss": 0.0794,
+      "step": 9923
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.703681710213777e-05,
+      "loss": 0.1178,
+      "step": 9924
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.7030878859857484e-05,
+      "loss": 0.0877,
+      "step": 9925
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.7024940617577204e-05,
+      "loss": 0.0918,
+      "step": 9926
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.701900237529692e-05,
+      "loss": 0.0998,
+      "step": 9927
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.7013064133016625e-05,
+      "loss": 0.0627,
+      "step": 9928
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.7007125890736345e-05,
+      "loss": 0.0616,
+      "step": 9929
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.700118764845606e-05,
+      "loss": 0.0594,
+      "step": 9930
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.699524940617577e-05,
+      "loss": 0.1014,
+      "step": 9931
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.6989311163895486e-05,
+      "loss": 0.0573,
+      "step": 9932
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.6983372921615206e-05,
+      "loss": 0.0624,
+      "step": 9933
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.697743467933492e-05,
+      "loss": 0.0834,
+      "step": 9934
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.697149643705463e-05,
+      "loss": 0.0679,
+      "step": 9935
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.6965558194774354e-05,
+      "loss": 0.068,
+      "step": 9936
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.695961995249407e-05,
+      "loss": 0.0841,
+      "step": 9937
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.695368171021378e-05,
+      "loss": 0.0841,
+      "step": 9938
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.6947743467933494e-05,
+      "loss": 0.0733,
+      "step": 9939
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.694180522565321e-05,
+      "loss": 0.1786,
+      "step": 9940
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.693586698337292e-05,
+      "loss": 0.0742,
+      "step": 9941
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.6929928741092635e-05,
+      "loss": 0.106,
+      "step": 9942
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.6923990498812355e-05,
+      "loss": 0.0815,
+      "step": 9943
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.691805225653207e-05,
+      "loss": 0.0669,
+      "step": 9944
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.691211401425178e-05,
+      "loss": 0.0555,
+      "step": 9945
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.69061757719715e-05,
+      "loss": 0.0587,
+      "step": 9946
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.6900237529691217e-05,
+      "loss": 0.1484,
+      "step": 9947
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.689429928741093e-05,
+      "loss": 0.0543,
+      "step": 9948
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.6888361045130644e-05,
+      "loss": 0.0779,
+      "step": 9949
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.688242280285036e-05,
+      "loss": 0.0897,
+      "step": 9950
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.687648456057007e-05,
+      "loss": 0.0768,
+      "step": 9951
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.6870546318289784e-05,
+      "loss": 0.0848,
+      "step": 9952
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.6864608076009505e-05,
+      "loss": 0.0532,
+      "step": 9953
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.685866983372922e-05,
+      "loss": 0.0606,
+      "step": 9954
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.685273159144893e-05,
+      "loss": 0.0497,
+      "step": 9955
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.684679334916865e-05,
+      "loss": 0.0676,
+      "step": 9956
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.6840855106888366e-05,
+      "loss": 0.0472,
+      "step": 9957
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.683491686460808e-05,
+      "loss": 0.0816,
+      "step": 9958
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.682897862232779e-05,
+      "loss": 0.0484,
+      "step": 9959
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.682304038004751e-05,
+      "loss": 0.0173,
+      "step": 9960
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.681710213776722e-05,
+      "loss": 0.0295,
+      "step": 9961
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.6811163895486934e-05,
+      "loss": 0.1448,
+      "step": 9962
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.6805225653206654e-05,
+      "loss": 0.1821,
+      "step": 9963
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.679928741092637e-05,
+      "loss": 0.0846,
+      "step": 9964
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.679334916864608e-05,
+      "loss": 0.1334,
+      "step": 9965
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.67874109263658e-05,
+      "loss": 0.0787,
+      "step": 9966
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.6781472684085515e-05,
+      "loss": 0.07,
+      "step": 9967
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.677553444180523e-05,
+      "loss": 0.0978,
+      "step": 9968
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.676959619952494e-05,
+      "loss": 0.1164,
+      "step": 9969
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.6763657957244656e-05,
+      "loss": 0.0852,
+      "step": 9970
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.675771971496437e-05,
+      "loss": 0.0934,
+      "step": 9971
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.675178147268408e-05,
+      "loss": 0.1103,
+      "step": 9972
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.6745843230403804e-05,
+      "loss": 0.0651,
+      "step": 9973
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.673990498812352e-05,
+      "loss": 0.0628,
+      "step": 9974
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.673396674584323e-05,
+      "loss": 0.1437,
+      "step": 9975
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.672802850356295e-05,
+      "loss": 0.0896,
+      "step": 9976
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.6722090261282665e-05,
+      "loss": 0.0617,
+      "step": 9977
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.671615201900238e-05,
+      "loss": 0.059,
+      "step": 9978
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.671021377672209e-05,
+      "loss": 0.1171,
+      "step": 9979
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.6704275534441805e-05,
+      "loss": 0.0853,
+      "step": 9980
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.669833729216152e-05,
+      "loss": 0.086,
+      "step": 9981
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.669239904988123e-05,
+      "loss": 0.0849,
+      "step": 9982
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.668646080760095e-05,
+      "loss": 0.0806,
+      "step": 9983
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.6680522565320666e-05,
+      "loss": 0.1021,
+      "step": 9984
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.667458432304038e-05,
+      "loss": 0.0979,
+      "step": 9985
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.66686460807601e-05,
+      "loss": 0.0902,
+      "step": 9986
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.6662707838479814e-05,
+      "loss": 0.0613,
+      "step": 9987
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.665676959619953e-05,
+      "loss": 0.0647,
+      "step": 9988
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.665083135391924e-05,
+      "loss": 0.051,
+      "step": 9989
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.664489311163896e-05,
+      "loss": 0.0684,
+      "step": 9990
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.663895486935867e-05,
+      "loss": 0.0698,
+      "step": 9991
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.663301662707838e-05,
+      "loss": 0.0768,
+      "step": 9992
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.66270783847981e-05,
+      "loss": 0.0442,
+      "step": 9993
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.6621140142517816e-05,
+      "loss": 0.0612,
+      "step": 9994
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.661520190023753e-05,
+      "loss": 0.0621,
+      "step": 9995
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.660926365795725e-05,
+      "loss": 0.094,
+      "step": 9996
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.660332541567696e-05,
+      "loss": 0.0503,
+      "step": 9997
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.659738717339668e-05,
+      "loss": 0.0648,
+      "step": 9998
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.659144893111639e-05,
+      "loss": 0.1771,
+      "step": 9999
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.658551068883611e-05,
+      "loss": 0.0807,
+      "step": 10000
+    },
+    {
+      "epoch": 11.21,
+      "eval_loss": 1.7600950002670288,
+      "eval_runtime": 898.9372,
+      "eval_samples_per_second": 2.939,
+      "eval_steps_per_second": 0.368,
+      "eval_wer": 1.000416501388338,
+      "step": 10000
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.657957244655582e-05,
+      "loss": 0.0574,
+      "step": 10001
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.657363420427554e-05,
+      "loss": 0.0791,
+      "step": 10002
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.656769596199525e-05,
+      "loss": 0.0866,
+      "step": 10003
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.6561757719714965e-05,
+      "loss": 0.0624,
+      "step": 10004
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.655581947743468e-05,
+      "loss": 0.0583,
+      "step": 10005
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.65498812351544e-05,
+      "loss": 0.0659,
+      "step": 10006
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.654394299287411e-05,
+      "loss": 0.0454,
+      "step": 10007
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.6538004750593826e-05,
+      "loss": 0.3152,
+      "step": 10008
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.653206650831354e-05,
+      "loss": 0.0493,
+      "step": 10009
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.652612826603326e-05,
+      "loss": 0.0819,
+      "step": 10010
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.652019002375297e-05,
+      "loss": 0.0347,
+      "step": 10011
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.651425178147269e-05,
+      "loss": 0.0459,
+      "step": 10012
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.65083135391924e-05,
+      "loss": 0.1282,
+      "step": 10013
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.6502375296912115e-05,
+      "loss": 0.1576,
+      "step": 10014
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.649643705463183e-05,
+      "loss": 0.1039,
+      "step": 10015
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.649049881235155e-05,
+      "loss": 0.0998,
+      "step": 10016
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.648456057007126e-05,
+      "loss": 0.0759,
+      "step": 10017
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.6478622327790976e-05,
+      "loss": 0.0988,
+      "step": 10018
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.6472684085510696e-05,
+      "loss": 0.0869,
+      "step": 10019
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.646674584323041e-05,
+      "loss": 0.0708,
+      "step": 10020
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.646080760095012e-05,
+      "loss": 0.0918,
+      "step": 10021
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.645486935866984e-05,
+      "loss": 0.1217,
+      "step": 10022
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.644893111638955e-05,
+      "loss": 0.1068,
+      "step": 10023
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.6442992874109264e-05,
+      "loss": 0.0954,
+      "step": 10024
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.643705463182898e-05,
+      "loss": 0.1097,
+      "step": 10025
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.64311163895487e-05,
+      "loss": 0.0907,
+      "step": 10026
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.642517814726841e-05,
+      "loss": 0.0746,
+      "step": 10027
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.6419239904988125e-05,
+      "loss": 0.0669,
+      "step": 10028
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.6413301662707845e-05,
+      "loss": 0.0567,
+      "step": 10029
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.640736342042756e-05,
+      "loss": 0.0975,
+      "step": 10030
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.640142517814727e-05,
+      "loss": 0.0656,
+      "step": 10031
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.6395486935866986e-05,
+      "loss": 0.0637,
+      "step": 10032
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.63895486935867e-05,
+      "loss": 0.0561,
+      "step": 10033
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.638361045130641e-05,
+      "loss": 0.1483,
+      "step": 10034
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.637767220902613e-05,
+      "loss": 0.074,
+      "step": 10035
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.637173396674585e-05,
+      "loss": 0.0803,
+      "step": 10036
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.636579572446556e-05,
+      "loss": 0.0537,
+      "step": 10037
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.6359857482185274e-05,
+      "loss": 0.0867,
+      "step": 10038
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.6353919239904995e-05,
+      "loss": 0.0581,
+      "step": 10039
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.634798099762471e-05,
+      "loss": 0.0518,
+      "step": 10040
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.634204275534442e-05,
+      "loss": 0.0446,
+      "step": 10041
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.6336104513064135e-05,
+      "loss": 0.2347,
+      "step": 10042
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.633016627078385e-05,
+      "loss": 0.0491,
+      "step": 10043
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.632422802850356e-05,
+      "loss": 0.0928,
+      "step": 10044
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.6318289786223276e-05,
+      "loss": 0.0767,
+      "step": 10045
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.6312351543942997e-05,
+      "loss": 0.0548,
+      "step": 10046
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.630641330166271e-05,
+      "loss": 0.0628,
+      "step": 10047
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.6300475059382424e-05,
+      "loss": 0.1128,
+      "step": 10048
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.6294536817102144e-05,
+      "loss": 0.0497,
+      "step": 10049
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.628859857482186e-05,
+      "loss": 0.0955,
+      "step": 10050
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.628266033254157e-05,
+      "loss": 0.0736,
+      "step": 10051
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.6276722090261285e-05,
+      "loss": 0.0897,
+      "step": 10052
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.6270783847981e-05,
+      "loss": 0.1016,
+      "step": 10053
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.626484560570071e-05,
+      "loss": 0.0713,
+      "step": 10054
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.6258907363420426e-05,
+      "loss": 0.0368,
+      "step": 10055
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.6252969121140146e-05,
+      "loss": 0.0866,
+      "step": 10056
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.624703087885986e-05,
+      "loss": 0.0566,
+      "step": 10057
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.624109263657957e-05,
+      "loss": 0.0292,
+      "step": 10058
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.6235154394299293e-05,
+      "loss": 0.0777,
+      "step": 10059
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.622921615201901e-05,
+      "loss": 0.09,
+      "step": 10060
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.622327790973872e-05,
+      "loss": 0.0786,
+      "step": 10061
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.6217339667458434e-05,
+      "loss": 0.0965,
+      "step": 10062
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.621140142517815e-05,
+      "loss": 0.1832,
+      "step": 10063
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.620546318289786e-05,
+      "loss": 0.143,
+      "step": 10064
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.6199524940617575e-05,
+      "loss": 0.1465,
+      "step": 10065
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.6193586698337295e-05,
+      "loss": 0.1112,
+      "step": 10066
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.618764845605701e-05,
+      "loss": 0.091,
+      "step": 10067
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.618171021377672e-05,
+      "loss": 0.0637,
+      "step": 10068
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.617577197149644e-05,
+      "loss": 0.0724,
+      "step": 10069
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.6169833729216156e-05,
+      "loss": 0.1217,
+      "step": 10070
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.616389548693587e-05,
+      "loss": 0.1075,
+      "step": 10071
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.6157957244655584e-05,
+      "loss": 0.1287,
+      "step": 10072
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.6152019002375304e-05,
+      "loss": 0.0759,
+      "step": 10073
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.614608076009501e-05,
+      "loss": 0.0821,
+      "step": 10074
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.6140142517814724e-05,
+      "loss": 0.1053,
+      "step": 10075
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.6134204275534445e-05,
+      "loss": 0.0995,
+      "step": 10076
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.612826603325416e-05,
+      "loss": 0.0815,
+      "step": 10077
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.612232779097387e-05,
+      "loss": 0.0921,
+      "step": 10078
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.611638954869359e-05,
+      "loss": 0.0766,
+      "step": 10079
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.6110451306413306e-05,
+      "loss": 0.0716,
+      "step": 10080
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.610451306413302e-05,
+      "loss": 0.0774,
+      "step": 10081
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.609857482185273e-05,
+      "loss": 0.063,
+      "step": 10082
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.609263657957245e-05,
+      "loss": 0.0639,
+      "step": 10083
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.608669833729216e-05,
+      "loss": 0.2027,
+      "step": 10084
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.6080760095011874e-05,
+      "loss": 0.0754,
+      "step": 10085
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.6074821852731594e-05,
+      "loss": 0.0955,
+      "step": 10086
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.606888361045131e-05,
+      "loss": 0.0732,
+      "step": 10087
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.606294536817102e-05,
+      "loss": 0.1017,
+      "step": 10088
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.605700712589074e-05,
+      "loss": 0.0762,
+      "step": 10089
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.6051068883610455e-05,
+      "loss": 0.0561,
+      "step": 10090
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.604513064133017e-05,
+      "loss": 0.0832,
+      "step": 10091
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.603919239904988e-05,
+      "loss": 0.0757,
+      "step": 10092
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.60332541567696e-05,
+      "loss": 0.0748,
+      "step": 10093
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.602731591448931e-05,
+      "loss": 0.0644,
+      "step": 10094
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.602137767220902e-05,
+      "loss": 0.0689,
+      "step": 10095
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.601543942992874e-05,
+      "loss": 0.0703,
+      "step": 10096
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.600950118764846e-05,
+      "loss": 0.0457,
+      "step": 10097
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.600356294536817e-05,
+      "loss": 0.0457,
+      "step": 10098
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.599762470308789e-05,
+      "loss": 0.0802,
+      "step": 10099
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.5991686460807604e-05,
+      "loss": 0.0555,
+      "step": 10100
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.598574821852732e-05,
+      "loss": 0.0711,
+      "step": 10101
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.597980997624703e-05,
+      "loss": 0.0412,
+      "step": 10102
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.597387173396675e-05,
+      "loss": 0.0783,
+      "step": 10103
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.5967933491686466e-05,
+      "loss": 0.0919,
+      "step": 10104
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.596199524940617e-05,
+      "loss": 0.074,
+      "step": 10105
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.595605700712589e-05,
+      "loss": 0.061,
+      "step": 10106
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.5950118764845606e-05,
+      "loss": 0.0445,
+      "step": 10107
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.594418052256532e-05,
+      "loss": 0.0878,
+      "step": 10108
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.593824228028504e-05,
+      "loss": 0.0567,
+      "step": 10109
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.5932304038004754e-05,
+      "loss": 0.0677,
+      "step": 10110
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.592636579572447e-05,
+      "loss": 0.0952,
+      "step": 10111
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.592042755344418e-05,
+      "loss": 0.0595,
+      "step": 10112
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.59144893111639e-05,
+      "loss": 0.1583,
+      "step": 10113
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.5908551068883615e-05,
+      "loss": 0.1088,
+      "step": 10114
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.590261282660332e-05,
+      "loss": 0.1704,
+      "step": 10115
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.589667458432304e-05,
+      "loss": 0.1237,
+      "step": 10116
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.5890736342042756e-05,
+      "loss": 0.1097,
+      "step": 10117
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.588479809976247e-05,
+      "loss": 0.0936,
+      "step": 10118
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.587885985748219e-05,
+      "loss": 0.106,
+      "step": 10119
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.58729216152019e-05,
+      "loss": 0.0687,
+      "step": 10120
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.586698337292162e-05,
+      "loss": 0.1478,
+      "step": 10121
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.586104513064133e-05,
+      "loss": 0.0861,
+      "step": 10122
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.585510688836105e-05,
+      "loss": 0.0839,
+      "step": 10123
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.5849168646080764e-05,
+      "loss": 0.0852,
+      "step": 10124
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.584323040380048e-05,
+      "loss": 0.0358,
+      "step": 10125
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.583729216152019e-05,
+      "loss": 0.069,
+      "step": 10126
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.5831353919239905e-05,
+      "loss": 0.0938,
+      "step": 10127
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.582541567695962e-05,
+      "loss": 0.0997,
+      "step": 10128
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.581947743467934e-05,
+      "loss": 0.1181,
+      "step": 10129
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.581353919239905e-05,
+      "loss": 0.0946,
+      "step": 10130
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.5807600950118766e-05,
+      "loss": 0.0972,
+      "step": 10131
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.580166270783848e-05,
+      "loss": 0.09,
+      "step": 10132
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.57957244655582e-05,
+      "loss": 0.0758,
+      "step": 10133
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.5789786223277914e-05,
+      "loss": 0.0794,
+      "step": 10134
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.578384798099763e-05,
+      "loss": 0.0711,
+      "step": 10135
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.577790973871734e-05,
+      "loss": 0.0912,
+      "step": 10136
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.5771971496437054e-05,
+      "loss": 0.0711,
+      "step": 10137
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.576603325415677e-05,
+      "loss": 0.0654,
+      "step": 10138
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.576009501187649e-05,
+      "loss": 0.0782,
+      "step": 10139
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.57541567695962e-05,
+      "loss": 0.0402,
+      "step": 10140
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.5748218527315915e-05,
+      "loss": 0.0887,
+      "step": 10141
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.574228028503563e-05,
+      "loss": 0.0726,
+      "step": 10142
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.573634204275535e-05,
+      "loss": 0.0982,
+      "step": 10143
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.573040380047506e-05,
+      "loss": 0.0458,
+      "step": 10144
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.5724465558194777e-05,
+      "loss": 0.0661,
+      "step": 10145
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.571852731591449e-05,
+      "loss": 0.0491,
+      "step": 10146
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.5712589073634204e-05,
+      "loss": 0.0645,
+      "step": 10147
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.570665083135392e-05,
+      "loss": 0.0465,
+      "step": 10148
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.570071258907364e-05,
+      "loss": 0.077,
+      "step": 10149
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.569477434679335e-05,
+      "loss": 0.1085,
+      "step": 10150
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.5688836104513065e-05,
+      "loss": 0.0379,
+      "step": 10151
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.5682897862232785e-05,
+      "loss": 0.0529,
+      "step": 10152
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.56769596199525e-05,
+      "loss": 0.0632,
+      "step": 10153
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.567102137767221e-05,
+      "loss": 0.0707,
+      "step": 10154
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.5665083135391926e-05,
+      "loss": 0.0577,
+      "step": 10155
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.5659144893111646e-05,
+      "loss": 0.0458,
+      "step": 10156
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.565320665083135e-05,
+      "loss": 0.0652,
+      "step": 10157
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.564726840855107e-05,
+      "loss": 0.1083,
+      "step": 10158
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.564133016627079e-05,
+      "loss": 0.0484,
+      "step": 10159
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.56353919239905e-05,
+      "loss": 0.0596,
+      "step": 10160
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.5629453681710214e-05,
+      "loss": 0.0665,
+      "step": 10161
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.5623515439429935e-05,
+      "loss": 0.1113,
+      "step": 10162
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.561757719714965e-05,
+      "loss": 0.1835,
+      "step": 10163
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.561163895486936e-05,
+      "loss": 0.0897,
+      "step": 10164
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.5605700712589075e-05,
+      "loss": 0.1192,
+      "step": 10165
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.5599762470308796e-05,
+      "loss": 0.0967,
+      "step": 10166
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.55938242280285e-05,
+      "loss": 0.0571,
+      "step": 10167
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.5587885985748216e-05,
+      "loss": 0.089,
+      "step": 10168
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.5581947743467936e-05,
+      "loss": 0.0922,
+      "step": 10169
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.557600950118765e-05,
+      "loss": 0.0604,
+      "step": 10170
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.5570071258907364e-05,
+      "loss": 0.0884,
+      "step": 10171
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.5564133016627084e-05,
+      "loss": 0.0876,
+      "step": 10172
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.55581947743468e-05,
+      "loss": 0.0812,
+      "step": 10173
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.555225653206651e-05,
+      "loss": 0.0833,
+      "step": 10174
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.5546318289786225e-05,
+      "loss": 0.1426,
+      "step": 10175
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.5540380047505945e-05,
+      "loss": 0.0886,
+      "step": 10176
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.553444180522565e-05,
+      "loss": 0.0731,
+      "step": 10177
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.5528503562945365e-05,
+      "loss": 0.062,
+      "step": 10178
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.5522565320665086e-05,
+      "loss": 0.0817,
+      "step": 10179
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.55166270783848e-05,
+      "loss": 0.0706,
+      "step": 10180
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.551068883610451e-05,
+      "loss": 0.0678,
+      "step": 10181
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.550475059382423e-05,
+      "loss": 0.0614,
+      "step": 10182
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.549881235154395e-05,
+      "loss": 0.0727,
+      "step": 10183
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.549287410926366e-05,
+      "loss": 0.0774,
+      "step": 10184
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.5486935866983374e-05,
+      "loss": 0.1405,
+      "step": 10185
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.5480997624703094e-05,
+      "loss": 0.0601,
+      "step": 10186
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.547505938242281e-05,
+      "loss": 0.1598,
+      "step": 10187
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.5469121140142515e-05,
+      "loss": 0.0505,
+      "step": 10188
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.5463182897862235e-05,
+      "loss": 0.0913,
+      "step": 10189
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.545724465558195e-05,
+      "loss": 0.0951,
+      "step": 10190
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.545130641330166e-05,
+      "loss": 0.1019,
+      "step": 10191
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.544536817102138e-05,
+      "loss": 0.0689,
+      "step": 10192
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.5439429928741096e-05,
+      "loss": 0.1016,
+      "step": 10193
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.543349168646081e-05,
+      "loss": 0.069,
+      "step": 10194
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.542755344418052e-05,
+      "loss": 0.0803,
+      "step": 10195
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.5421615201900244e-05,
+      "loss": 0.1094,
+      "step": 10196
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.541567695961996e-05,
+      "loss": 0.1015,
+      "step": 10197
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.5409738717339664e-05,
+      "loss": 0.0982,
+      "step": 10198
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.5403800475059384e-05,
+      "loss": 0.0294,
+      "step": 10199
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.53978622327791e-05,
+      "loss": 0.0561,
+      "step": 10200
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.539192399049881e-05,
+      "loss": 0.0392,
+      "step": 10201
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.538598574821853e-05,
+      "loss": 0.0502,
+      "step": 10202
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.5380047505938246e-05,
+      "loss": 0.0628,
+      "step": 10203
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.537410926365796e-05,
+      "loss": 0.1258,
+      "step": 10204
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.536817102137767e-05,
+      "loss": 0.0486,
+      "step": 10205
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.536223277909739e-05,
+      "loss": 0.0932,
+      "step": 10206
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.535629453681711e-05,
+      "loss": 0.1082,
+      "step": 10207
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.5350356294536813e-05,
+      "loss": 0.1141,
+      "step": 10208
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.5344418052256534e-05,
+      "loss": 0.0574,
+      "step": 10209
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.533847980997625e-05,
+      "loss": 0.0462,
+      "step": 10210
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.533254156769596e-05,
+      "loss": 0.0411,
+      "step": 10211
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.532660332541568e-05,
+      "loss": 0.0737,
+      "step": 10212
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.5320665083135395e-05,
+      "loss": 0.1301,
+      "step": 10213
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.531472684085511e-05,
+      "loss": 0.1196,
+      "step": 10214
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.530878859857482e-05,
+      "loss": 0.1359,
+      "step": 10215
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.530285035629454e-05,
+      "loss": 0.089,
+      "step": 10216
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.5296912114014256e-05,
+      "loss": 0.0473,
+      "step": 10217
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.529097387173397e-05,
+      "loss": 0.0607,
+      "step": 10218
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.528503562945368e-05,
+      "loss": 0.0794,
+      "step": 10219
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.52790973871734e-05,
+      "loss": 0.0908,
+      "step": 10220
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.527315914489311e-05,
+      "loss": 0.0765,
+      "step": 10221
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.526722090261283e-05,
+      "loss": 0.2265,
+      "step": 10222
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.5261282660332544e-05,
+      "loss": 0.0766,
+      "step": 10223
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.525534441805226e-05,
+      "loss": 0.0831,
+      "step": 10224
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.524940617577197e-05,
+      "loss": 0.1059,
+      "step": 10225
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.524346793349169e-05,
+      "loss": 0.0781,
+      "step": 10226
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.5237529691211405e-05,
+      "loss": 0.062,
+      "step": 10227
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.523159144893112e-05,
+      "loss": 0.108,
+      "step": 10228
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.522565320665083e-05,
+      "loss": 0.0596,
+      "step": 10229
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.5219714964370546e-05,
+      "loss": 0.0739,
+      "step": 10230
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.521377672209026e-05,
+      "loss": 0.0713,
+      "step": 10231
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.520783847980998e-05,
+      "loss": 0.0606,
+      "step": 10232
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.5201900237529694e-05,
+      "loss": 0.0605,
+      "step": 10233
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.519596199524941e-05,
+      "loss": 0.0677,
+      "step": 10234
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.519002375296912e-05,
+      "loss": 0.0733,
+      "step": 10235
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.518408551068884e-05,
+      "loss": 0.152,
+      "step": 10236
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.5178147268408555e-05,
+      "loss": 0.1317,
+      "step": 10237
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.517220902612827e-05,
+      "loss": 0.0598,
+      "step": 10238
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.516627078384799e-05,
+      "loss": 0.0778,
+      "step": 10239
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.5160332541567695e-05,
+      "loss": 0.0729,
+      "step": 10240
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.515439429928741e-05,
+      "loss": 0.077,
+      "step": 10241
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.514845605700713e-05,
+      "loss": 0.1007,
+      "step": 10242
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.514251781472684e-05,
+      "loss": 0.0485,
+      "step": 10243
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.5136579572446557e-05,
+      "loss": 0.051,
+      "step": 10244
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.513064133016627e-05,
+      "loss": 0.084,
+      "step": 10245
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.512470308788599e-05,
+      "loss": 0.0618,
+      "step": 10246
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.5118764845605704e-05,
+      "loss": 0.0932,
+      "step": 10247
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.511282660332542e-05,
+      "loss": 0.0669,
+      "step": 10248
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.510688836104514e-05,
+      "loss": 0.0498,
+      "step": 10249
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.5100950118764845e-05,
+      "loss": 0.0512,
+      "step": 10250
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.509501187648456e-05,
+      "loss": 0.0505,
+      "step": 10251
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.508907363420428e-05,
+      "loss": 0.0637,
+      "step": 10252
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.508313539192399e-05,
+      "loss": 0.0665,
+      "step": 10253
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.5077197149643706e-05,
+      "loss": 0.0486,
+      "step": 10254
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.507125890736342e-05,
+      "loss": 0.031,
+      "step": 10255
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.506532066508314e-05,
+      "loss": 0.1244,
+      "step": 10256
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.5059382422802853e-05,
+      "loss": 0.0896,
+      "step": 10257
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.505344418052257e-05,
+      "loss": 0.0598,
+      "step": 10258
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.504750593824229e-05,
+      "loss": 0.0502,
+      "step": 10259
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.5041567695961994e-05,
+      "loss": 0.0598,
+      "step": 10260
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.503562945368171e-05,
+      "loss": 0.0659,
+      "step": 10261
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.502969121140143e-05,
+      "loss": 0.0408,
+      "step": 10262
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.502375296912114e-05,
+      "loss": 0.1555,
+      "step": 10263
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.5017814726840855e-05,
+      "loss": 0.128,
+      "step": 10264
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.501187648456057e-05,
+      "loss": 0.1301,
+      "step": 10265
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.500593824228029e-05,
+      "loss": 0.1363,
+      "step": 10266
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.5e-05,
+      "loss": 0.1153,
+      "step": 10267
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.4994061757719716e-05,
+      "loss": 0.0942,
+      "step": 10268
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.498812351543944e-05,
+      "loss": 0.0476,
+      "step": 10269
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.498218527315915e-05,
+      "loss": 0.0786,
+      "step": 10270
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.497624703087886e-05,
+      "loss": 0.1136,
+      "step": 10271
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.497030878859858e-05,
+      "loss": 0.0891,
+      "step": 10272
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.496437054631829e-05,
+      "loss": 0.0972,
+      "step": 10273
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.4958432304038005e-05,
+      "loss": 0.0832,
+      "step": 10274
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.4952494061757725e-05,
+      "loss": 0.095,
+      "step": 10275
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.494655581947744e-05,
+      "loss": 0.1389,
+      "step": 10276
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.494061757719715e-05,
+      "loss": 0.1043,
+      "step": 10277
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.4934679334916866e-05,
+      "loss": 0.0374,
+      "step": 10278
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.4928741092636586e-05,
+      "loss": 0.0761,
+      "step": 10279
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.49228028503563e-05,
+      "loss": 0.1015,
+      "step": 10280
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.4916864608076007e-05,
+      "loss": 0.0681,
+      "step": 10281
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.491092636579573e-05,
+      "loss": 0.0944,
+      "step": 10282
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.490498812351544e-05,
+      "loss": 0.0612,
+      "step": 10283
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.4899049881235154e-05,
+      "loss": 0.1023,
+      "step": 10284
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.4893111638954874e-05,
+      "loss": 0.1058,
+      "step": 10285
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.488717339667459e-05,
+      "loss": 0.1447,
+      "step": 10286
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.48812351543943e-05,
+      "loss": 0.048,
+      "step": 10287
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.4875296912114015e-05,
+      "loss": 0.037,
+      "step": 10288
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.4869358669833735e-05,
+      "loss": 0.0605,
+      "step": 10289
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.486342042755345e-05,
+      "loss": 0.1035,
+      "step": 10290
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.4857482185273156e-05,
+      "loss": 0.065,
+      "step": 10291
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.4851543942992876e-05,
+      "loss": 0.1057,
+      "step": 10292
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.484560570071259e-05,
+      "loss": 0.0652,
+      "step": 10293
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.4839667458432303e-05,
+      "loss": 0.0795,
+      "step": 10294
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.4833729216152024e-05,
+      "loss": 0.0672,
+      "step": 10295
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.482779097387174e-05,
+      "loss": 0.0748,
+      "step": 10296
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.482185273159145e-05,
+      "loss": 0.0612,
+      "step": 10297
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.4815914489311164e-05,
+      "loss": 0.0561,
+      "step": 10298
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.4809976247030885e-05,
+      "loss": 0.1124,
+      "step": 10299
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.48040380047506e-05,
+      "loss": 0.0819,
+      "step": 10300
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.479809976247031e-05,
+      "loss": 0.0947,
+      "step": 10301
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.4792161520190026e-05,
+      "loss": 0.1171,
+      "step": 10302
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.478622327790974e-05,
+      "loss": 0.064,
+      "step": 10303
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.478028503562945e-05,
+      "loss": 0.1145,
+      "step": 10304
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.477434679334917e-05,
+      "loss": 0.05,
+      "step": 10305
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.476840855106889e-05,
+      "loss": 0.0507,
+      "step": 10306
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.47624703087886e-05,
+      "loss": 0.0564,
+      "step": 10307
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.4756532066508314e-05,
+      "loss": 0.0278,
+      "step": 10308
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.4750593824228034e-05,
+      "loss": 0.0844,
+      "step": 10309
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.474465558194775e-05,
+      "loss": 0.0779,
+      "step": 10310
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.473871733966746e-05,
+      "loss": 0.0982,
+      "step": 10311
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.4732779097387175e-05,
+      "loss": 0.1991,
+      "step": 10312
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.472684085510689e-05,
+      "loss": 0.1857,
+      "step": 10313
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.47209026128266e-05,
+      "loss": 0.1573,
+      "step": 10314
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.471496437054632e-05,
+      "loss": 0.1677,
+      "step": 10315
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.4709026128266036e-05,
+      "loss": 0.1302,
+      "step": 10316
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.470308788598575e-05,
+      "loss": 0.1025,
+      "step": 10317
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.469714964370546e-05,
+      "loss": 0.0698,
+      "step": 10318
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.4691211401425184e-05,
+      "loss": 0.0665,
+      "step": 10319
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.46852731591449e-05,
+      "loss": 0.0947,
+      "step": 10320
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.467933491686461e-05,
+      "loss": 0.1236,
+      "step": 10321
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.4673396674584324e-05,
+      "loss": 0.0643,
+      "step": 10322
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.466745843230404e-05,
+      "loss": 0.0547,
+      "step": 10323
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.466152019002375e-05,
+      "loss": 0.0832,
+      "step": 10324
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.465558194774347e-05,
+      "loss": 0.0961,
+      "step": 10325
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.4649643705463185e-05,
+      "loss": 0.0809,
+      "step": 10326
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.46437054631829e-05,
+      "loss": 0.0514,
+      "step": 10327
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.463776722090261e-05,
+      "loss": 0.0668,
+      "step": 10328
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.463182897862233e-05,
+      "loss": 0.0798,
+      "step": 10329
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.4625890736342047e-05,
+      "loss": 0.0708,
+      "step": 10330
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.461995249406176e-05,
+      "loss": 0.0737,
+      "step": 10331
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.461401425178148e-05,
+      "loss": 0.0721,
+      "step": 10332
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.460807600950119e-05,
+      "loss": 0.0876,
+      "step": 10333
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.46021377672209e-05,
+      "loss": 0.0738,
+      "step": 10334
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.459619952494062e-05,
+      "loss": 0.0588,
+      "step": 10335
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.4590261282660335e-05,
+      "loss": 0.0547,
+      "step": 10336
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.458432304038005e-05,
+      "loss": 0.0822,
+      "step": 10337
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.457838479809976e-05,
+      "loss": 0.0783,
+      "step": 10338
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.457244655581948e-05,
+      "loss": 0.067,
+      "step": 10339
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.4566508313539196e-05,
+      "loss": 0.0738,
+      "step": 10340
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.456057007125891e-05,
+      "loss": 0.0687,
+      "step": 10341
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.455463182897863e-05,
+      "loss": 0.1118,
+      "step": 10342
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.4548693586698337e-05,
+      "loss": 0.0698,
+      "step": 10343
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.454275534441805e-05,
+      "loss": 0.063,
+      "step": 10344
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.453681710213777e-05,
+      "loss": 0.0653,
+      "step": 10345
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.4530878859857484e-05,
+      "loss": 0.0825,
+      "step": 10346
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.45249406175772e-05,
+      "loss": 0.0663,
+      "step": 10347
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.451900237529691e-05,
+      "loss": 0.0701,
+      "step": 10348
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.451306413301663e-05,
+      "loss": 0.0347,
+      "step": 10349
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.4507125890736345e-05,
+      "loss": 0.0763,
+      "step": 10350
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.450118764845606e-05,
+      "loss": 0.0384,
+      "step": 10351
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.449524940617578e-05,
+      "loss": 0.0652,
+      "step": 10352
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.448931116389549e-05,
+      "loss": 0.0584,
+      "step": 10353
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.44833729216152e-05,
+      "loss": 0.0565,
+      "step": 10354
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.447743467933492e-05,
+      "loss": 0.0715,
+      "step": 10355
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.4471496437054633e-05,
+      "loss": 0.0291,
+      "step": 10356
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.446555819477435e-05,
+      "loss": 0.071,
+      "step": 10357
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.445961995249406e-05,
+      "loss": 0.0379,
+      "step": 10358
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.445368171021378e-05,
+      "loss": 0.0449,
+      "step": 10359
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.4447743467933495e-05,
+      "loss": 0.0941,
+      "step": 10360
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.444180522565321e-05,
+      "loss": 0.1315,
+      "step": 10361
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.443586698337293e-05,
+      "loss": 0.0518,
+      "step": 10362
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.442992874109264e-05,
+      "loss": 0.1781,
+      "step": 10363
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.442399049881235e-05,
+      "loss": 0.1361,
+      "step": 10364
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.441805225653207e-05,
+      "loss": 0.1254,
+      "step": 10365
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.441211401425178e-05,
+      "loss": 0.0787,
+      "step": 10366
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.4406175771971496e-05,
+      "loss": 0.1192,
+      "step": 10367
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.440023752969121e-05,
+      "loss": 0.0928,
+      "step": 10368
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.439429928741093e-05,
+      "loss": 0.077,
+      "step": 10369
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.4388361045130644e-05,
+      "loss": 0.0802,
+      "step": 10370
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.438242280285036e-05,
+      "loss": 0.0908,
+      "step": 10371
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.437648456057008e-05,
+      "loss": 0.1075,
+      "step": 10372
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.437054631828979e-05,
+      "loss": 0.1115,
+      "step": 10373
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.43646080760095e-05,
+      "loss": 0.1037,
+      "step": 10374
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.435866983372922e-05,
+      "loss": 0.073,
+      "step": 10375
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.435273159144893e-05,
+      "loss": 0.1018,
+      "step": 10376
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.4346793349168646e-05,
+      "loss": 0.0934,
+      "step": 10377
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.434085510688836e-05,
+      "loss": 0.0754,
+      "step": 10378
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.433491686460808e-05,
+      "loss": 0.0789,
+      "step": 10379
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.432897862232779e-05,
+      "loss": 0.0677,
+      "step": 10380
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.432304038004751e-05,
+      "loss": 0.0781,
+      "step": 10381
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.431710213776723e-05,
+      "loss": 0.0971,
+      "step": 10382
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.431116389548694e-05,
+      "loss": 0.0604,
+      "step": 10383
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.4305225653206654e-05,
+      "loss": 0.0827,
+      "step": 10384
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.429928741092637e-05,
+      "loss": 0.0484,
+      "step": 10385
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.429334916864608e-05,
+      "loss": 0.0957,
+      "step": 10386
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.4287410926365795e-05,
+      "loss": 0.0775,
+      "step": 10387
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.428147268408551e-05,
+      "loss": 0.0888,
+      "step": 10388
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.427553444180523e-05,
+      "loss": 0.0526,
+      "step": 10389
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.426959619952494e-05,
+      "loss": 0.068,
+      "step": 10390
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.4263657957244656e-05,
+      "loss": 0.0846,
+      "step": 10391
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.4257719714964377e-05,
+      "loss": 0.0607,
+      "step": 10392
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.425178147268409e-05,
+      "loss": 0.0818,
+      "step": 10393
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.4245843230403804e-05,
+      "loss": 0.1068,
+      "step": 10394
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.423990498812352e-05,
+      "loss": 0.1063,
+      "step": 10395
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.423396674584323e-05,
+      "loss": 0.0737,
+      "step": 10396
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.4228028503562945e-05,
+      "loss": 0.0618,
+      "step": 10397
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.4222090261282665e-05,
+      "loss": 0.0651,
+      "step": 10398
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.421615201900238e-05,
+      "loss": 0.0424,
+      "step": 10399
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.421021377672209e-05,
+      "loss": 0.0738,
+      "step": 10400
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.4204275534441806e-05,
+      "loss": 0.0636,
+      "step": 10401
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.4198337292161526e-05,
+      "loss": 0.037,
+      "step": 10402
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.419239904988124e-05,
+      "loss": 0.068,
+      "step": 10403
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.418646080760095e-05,
+      "loss": 0.0207,
+      "step": 10404
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.418052256532067e-05,
+      "loss": 0.0534,
+      "step": 10405
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.417458432304038e-05,
+      "loss": 0.0564,
+      "step": 10406
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.4168646080760094e-05,
+      "loss": 0.0747,
+      "step": 10407
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.4162707838479814e-05,
+      "loss": 0.1914,
+      "step": 10408
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.415676959619953e-05,
+      "loss": 0.0266,
+      "step": 10409
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.415083135391924e-05,
+      "loss": 0.0944,
+      "step": 10410
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.4144893111638955e-05,
+      "loss": 0.0785,
+      "step": 10411
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.4138954869358675e-05,
+      "loss": 0.2818,
+      "step": 10412
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.413301662707839e-05,
+      "loss": 0.15,
+      "step": 10413
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.41270783847981e-05,
+      "loss": 0.1865,
+      "step": 10414
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.412114014251782e-05,
+      "loss": 0.1804,
+      "step": 10415
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.411520190023753e-05,
+      "loss": 0.0939,
+      "step": 10416
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.410926365795724e-05,
+      "loss": 0.1469,
+      "step": 10417
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.4103325415676964e-05,
+      "loss": 0.0993,
+      "step": 10418
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.409738717339668e-05,
+      "loss": 0.0703,
+      "step": 10419
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.409144893111639e-05,
+      "loss": 0.3492,
+      "step": 10420
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.4085510688836104e-05,
+      "loss": 0.105,
+      "step": 10421
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.4079572446555825e-05,
+      "loss": 0.1001,
+      "step": 10422
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.407363420427554e-05,
+      "loss": 0.1348,
+      "step": 10423
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.406769596199525e-05,
+      "loss": 0.1052,
+      "step": 10424
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.406175771971497e-05,
+      "loss": 0.0582,
+      "step": 10425
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.405581947743468e-05,
+      "loss": 0.1032,
+      "step": 10426
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.404988123515439e-05,
+      "loss": 0.1005,
+      "step": 10427
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.404394299287411e-05,
+      "loss": 0.1059,
+      "step": 10428
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.4038004750593827e-05,
+      "loss": 0.0954,
+      "step": 10429
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.403206650831354e-05,
+      "loss": 0.1244,
+      "step": 10430
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.4026128266033254e-05,
+      "loss": 0.1188,
+      "step": 10431
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.4020190023752974e-05,
+      "loss": 0.0586,
+      "step": 10432
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.401425178147269e-05,
+      "loss": 0.0839,
+      "step": 10433
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.40083135391924e-05,
+      "loss": 0.0989,
+      "step": 10434
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.400237529691212e-05,
+      "loss": 0.1034,
+      "step": 10435
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.3996437054631835e-05,
+      "loss": 0.1153,
+      "step": 10436
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.399049881235154e-05,
+      "loss": 0.0858,
+      "step": 10437
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.398456057007126e-05,
+      "loss": 0.0591,
+      "step": 10438
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.3978622327790976e-05,
+      "loss": 0.0804,
+      "step": 10439
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.397268408551069e-05,
+      "loss": 0.0447,
+      "step": 10440
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.39667458432304e-05,
+      "loss": 0.0608,
+      "step": 10441
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.3960807600950123e-05,
+      "loss": 0.0751,
+      "step": 10442
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.395486935866984e-05,
+      "loss": 0.0724,
+      "step": 10443
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.394893111638955e-05,
+      "loss": 0.0992,
+      "step": 10444
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.394299287410927e-05,
+      "loss": 0.0658,
+      "step": 10445
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.3937054631828984e-05,
+      "loss": 0.0482,
+      "step": 10446
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.393111638954869e-05,
+      "loss": 0.076,
+      "step": 10447
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.392517814726841e-05,
+      "loss": 0.1085,
+      "step": 10448
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.3919239904988125e-05,
+      "loss": 0.0582,
+      "step": 10449
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.391330166270784e-05,
+      "loss": 0.0649,
+      "step": 10450
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.390736342042755e-05,
+      "loss": 0.0585,
+      "step": 10451
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.390142517814727e-05,
+      "loss": 0.0766,
+      "step": 10452
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.3895486935866986e-05,
+      "loss": 0.0533,
+      "step": 10453
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.38895486935867e-05,
+      "loss": 0.0786,
+      "step": 10454
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.388361045130642e-05,
+      "loss": 0.0736,
+      "step": 10455
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.3877672209026134e-05,
+      "loss": 0.0752,
+      "step": 10456
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.387173396674584e-05,
+      "loss": 0.0823,
+      "step": 10457
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.386579572446556e-05,
+      "loss": 0.0674,
+      "step": 10458
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.3859857482185275e-05,
+      "loss": 0.0697,
+      "step": 10459
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.385391923990499e-05,
+      "loss": 0.0861,
+      "step": 10460
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.38479809976247e-05,
+      "loss": 0.0562,
+      "step": 10461
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.384204275534442e-05,
+      "loss": 0.1472,
+      "step": 10462
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.3836104513064136e-05,
+      "loss": 0.1858,
+      "step": 10463
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.383016627078385e-05,
+      "loss": 0.1309,
+      "step": 10464
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.382422802850357e-05,
+      "loss": 0.1701,
+      "step": 10465
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.381828978622328e-05,
+      "loss": 0.1209,
+      "step": 10466
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.3812351543943e-05,
+      "loss": 0.098,
+      "step": 10467
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.380641330166271e-05,
+      "loss": 0.085,
+      "step": 10468
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.3800475059382424e-05,
+      "loss": 0.0827,
+      "step": 10469
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.379453681710214e-05,
+      "loss": 0.0782,
+      "step": 10470
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.378859857482185e-05,
+      "loss": 0.0703,
+      "step": 10471
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.378266033254157e-05,
+      "loss": 0.087,
+      "step": 10472
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.3776722090261285e-05,
+      "loss": 0.0703,
+      "step": 10473
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.3770783847981e-05,
+      "loss": 0.0972,
+      "step": 10474
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.376484560570072e-05,
+      "loss": 0.0556,
+      "step": 10475
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.375890736342043e-05,
+      "loss": 0.0867,
+      "step": 10476
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.3752969121140146e-05,
+      "loss": 0.1259,
+      "step": 10477
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.374703087885986e-05,
+      "loss": 0.0603,
+      "step": 10478
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.374109263657957e-05,
+      "loss": 0.0655,
+      "step": 10479
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.373515439429929e-05,
+      "loss": 0.0779,
+      "step": 10480
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.3729216152019e-05,
+      "loss": 0.0939,
+      "step": 10481
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.372327790973872e-05,
+      "loss": 0.072,
+      "step": 10482
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.3717339667458434e-05,
+      "loss": 0.1055,
+      "step": 10483
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.371140142517815e-05,
+      "loss": 0.1128,
+      "step": 10484
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.370546318289787e-05,
+      "loss": 0.1201,
+      "step": 10485
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.369952494061758e-05,
+      "loss": 0.0682,
+      "step": 10486
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.3693586698337296e-05,
+      "loss": 0.073,
+      "step": 10487
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.368764845605701e-05,
+      "loss": 0.1538,
+      "step": 10488
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.368171021377672e-05,
+      "loss": 0.1099,
+      "step": 10489
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.3675771971496436e-05,
+      "loss": 0.0675,
+      "step": 10490
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.366983372921615e-05,
+      "loss": 0.0949,
+      "step": 10491
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.366389548693587e-05,
+      "loss": 0.1049,
+      "step": 10492
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.3657957244655584e-05,
+      "loss": 0.0528,
+      "step": 10493
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.36520190023753e-05,
+      "loss": 0.1058,
+      "step": 10494
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.364608076009502e-05,
+      "loss": 0.0759,
+      "step": 10495
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.364014251781473e-05,
+      "loss": 0.0521,
+      "step": 10496
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.3634204275534445e-05,
+      "loss": 0.0791,
+      "step": 10497
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.362826603325416e-05,
+      "loss": 0.0827,
+      "step": 10498
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.362232779097387e-05,
+      "loss": 0.0399,
+      "step": 10499
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.3616389548693586e-05,
+      "loss": 0.0549,
+      "step": 10500
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.36104513064133e-05,
+      "loss": 0.0733,
+      "step": 10501
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.360451306413302e-05,
+      "loss": 0.0564,
+      "step": 10502
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.359857482185273e-05,
+      "loss": 0.0641,
+      "step": 10503
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.359263657957245e-05,
+      "loss": 0.0879,
+      "step": 10504
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.358669833729217e-05,
+      "loss": 0.0585,
+      "step": 10505
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.358076009501188e-05,
+      "loss": 0.0445,
+      "step": 10506
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.3574821852731594e-05,
+      "loss": 0.0532,
+      "step": 10507
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.356888361045131e-05,
+      "loss": 0.1053,
+      "step": 10508
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.356294536817102e-05,
+      "loss": 0.0638,
+      "step": 10509
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.3557007125890735e-05,
+      "loss": 0.0383,
+      "step": 10510
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.355106888361045e-05,
+      "loss": 0.0459,
+      "step": 10511
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.354513064133017e-05,
+      "loss": 0.0995,
+      "step": 10512
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.353919239904988e-05,
+      "loss": 0.1407,
+      "step": 10513
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.3533254156769596e-05,
+      "loss": 0.1264,
+      "step": 10514
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.3527315914489316e-05,
+      "loss": 0.1373,
+      "step": 10515
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.352137767220903e-05,
+      "loss": 0.113,
+      "step": 10516
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.3515439429928744e-05,
+      "loss": 0.1362,
+      "step": 10517
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.350950118764846e-05,
+      "loss": 0.0933,
+      "step": 10518
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.350356294536817e-05,
+      "loss": 0.1076,
+      "step": 10519
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.3497624703087884e-05,
+      "loss": 0.0883,
+      "step": 10520
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.3491686460807605e-05,
+      "loss": 0.0676,
+      "step": 10521
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.348574821852732e-05,
+      "loss": 0.0917,
+      "step": 10522
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.347980997624703e-05,
+      "loss": 0.1236,
+      "step": 10523
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.3473871733966745e-05,
+      "loss": 0.1074,
+      "step": 10524
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.3467933491686466e-05,
+      "loss": 0.1195,
+      "step": 10525
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.346199524940618e-05,
+      "loss": 0.0827,
+      "step": 10526
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.345605700712589e-05,
+      "loss": 0.0993,
+      "step": 10527
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.3450118764845607e-05,
+      "loss": 0.0674,
+      "step": 10528
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.344418052256533e-05,
+      "loss": 0.1082,
+      "step": 10529
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.3438242280285034e-05,
+      "loss": 0.0596,
+      "step": 10530
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.3432304038004754e-05,
+      "loss": 0.058,
+      "step": 10531
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.342636579572447e-05,
+      "loss": 0.1338,
+      "step": 10532
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.342042755344418e-05,
+      "loss": 0.091,
+      "step": 10533
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.3414489311163895e-05,
+      "loss": 0.135,
+      "step": 10534
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.3408551068883615e-05,
+      "loss": 0.0497,
+      "step": 10535
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.340261282660333e-05,
+      "loss": 0.1059,
+      "step": 10536
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.339667458432304e-05,
+      "loss": 0.1586,
+      "step": 10537
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.339073634204276e-05,
+      "loss": 0.0871,
+      "step": 10538
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.3384798099762476e-05,
+      "loss": 0.0781,
+      "step": 10539
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.337885985748218e-05,
+      "loss": 0.0966,
+      "step": 10540
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.3372921615201903e-05,
+      "loss": 0.068,
+      "step": 10541
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.336698337292162e-05,
+      "loss": 0.0551,
+      "step": 10542
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.336104513064133e-05,
+      "loss": 0.07,
+      "step": 10543
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.3355106888361044e-05,
+      "loss": 0.0498,
+      "step": 10544
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.3349168646080765e-05,
+      "loss": 0.0891,
+      "step": 10545
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.334323040380048e-05,
+      "loss": 0.0906,
+      "step": 10546
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.333729216152019e-05,
+      "loss": 0.0831,
+      "step": 10547
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.333135391923991e-05,
+      "loss": 0.0683,
+      "step": 10548
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.3325415676959626e-05,
+      "loss": 0.0673,
+      "step": 10549
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.331947743467934e-05,
+      "loss": 0.0703,
+      "step": 10550
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.331353919239905e-05,
+      "loss": 0.0461,
+      "step": 10551
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.3307600950118766e-05,
+      "loss": 0.0575,
+      "step": 10552
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.330166270783848e-05,
+      "loss": 0.0403,
+      "step": 10553
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.3295724465558194e-05,
+      "loss": 0.0533,
+      "step": 10554
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.3289786223277914e-05,
+      "loss": 0.0851,
+      "step": 10555
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.328384798099763e-05,
+      "loss": 0.0358,
+      "step": 10556
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.327790973871734e-05,
+      "loss": 0.0557,
+      "step": 10557
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.327197149643706e-05,
+      "loss": 0.1093,
+      "step": 10558
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.3266033254156775e-05,
+      "loss": 0.0646,
+      "step": 10559
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.326009501187649e-05,
+      "loss": 0.1219,
+      "step": 10560
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.32541567695962e-05,
+      "loss": 0.0768,
+      "step": 10561
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.3248218527315916e-05,
+      "loss": 0.1813,
+      "step": 10562
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.324228028503563e-05,
+      "loss": 0.1702,
+      "step": 10563
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.323634204275534e-05,
+      "loss": 0.1489,
+      "step": 10564
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.323040380047506e-05,
+      "loss": 0.1151,
+      "step": 10565
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.322446555819478e-05,
+      "loss": 0.0884,
+      "step": 10566
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.321852731591449e-05,
+      "loss": 0.1087,
+      "step": 10567
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.321258907363421e-05,
+      "loss": 0.0589,
+      "step": 10568
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.3206650831353924e-05,
+      "loss": 0.0913,
+      "step": 10569
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.320071258907364e-05,
+      "loss": 0.0687,
+      "step": 10570
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.319477434679335e-05,
+      "loss": 0.121,
+      "step": 10571
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.3188836104513065e-05,
+      "loss": 0.1067,
+      "step": 10572
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.318289786223278e-05,
+      "loss": 0.073,
+      "step": 10573
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.317695961995249e-05,
+      "loss": 0.1158,
+      "step": 10574
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.317102137767221e-05,
+      "loss": 0.0764,
+      "step": 10575
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.3165083135391926e-05,
+      "loss": 0.0658,
+      "step": 10576
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.315914489311164e-05,
+      "loss": 0.1131,
+      "step": 10577
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.315320665083136e-05,
+      "loss": 0.0451,
+      "step": 10578
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.3147268408551074e-05,
+      "loss": 0.0558,
+      "step": 10579
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.314133016627079e-05,
+      "loss": 0.0752,
+      "step": 10580
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.31353919239905e-05,
+      "loss": 0.1116,
+      "step": 10581
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.3129453681710214e-05,
+      "loss": 0.0686,
+      "step": 10582
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.312351543942993e-05,
+      "loss": 0.1023,
+      "step": 10583
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.311757719714964e-05,
+      "loss": 0.1136,
+      "step": 10584
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.311163895486936e-05,
+      "loss": 0.07,
+      "step": 10585
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.3105700712589076e-05,
+      "loss": 0.0791,
+      "step": 10586
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.309976247030879e-05,
+      "loss": 0.0886,
+      "step": 10587
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.309382422802851e-05,
+      "loss": 0.0701,
+      "step": 10588
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.308788598574822e-05,
+      "loss": 0.0541,
+      "step": 10589
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.308194774346794e-05,
+      "loss": 0.0879,
+      "step": 10590
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.307600950118765e-05,
+      "loss": 0.0587,
+      "step": 10591
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.3070071258907364e-05,
+      "loss": 0.0826,
+      "step": 10592
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.306413301662708e-05,
+      "loss": 0.0923,
+      "step": 10593
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.305819477434679e-05,
+      "loss": 0.0788,
+      "step": 10594
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.305225653206651e-05,
+      "loss": 0.1092,
+      "step": 10595
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.3046318289786225e-05,
+      "loss": 0.0664,
+      "step": 10596
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.304038004750594e-05,
+      "loss": 0.0592,
+      "step": 10597
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.303444180522566e-05,
+      "loss": 0.055,
+      "step": 10598
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.302850356294537e-05,
+      "loss": 0.0558,
+      "step": 10599
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.3022565320665086e-05,
+      "loss": 0.0844,
+      "step": 10600
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.30166270783848e-05,
+      "loss": 0.1119,
+      "step": 10601
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.301068883610451e-05,
+      "loss": 0.0568,
+      "step": 10602
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.300475059382423e-05,
+      "loss": 0.0475,
+      "step": 10603
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.299881235154394e-05,
+      "loss": 0.0844,
+      "step": 10604
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.299287410926366e-05,
+      "loss": 0.104,
+      "step": 10605
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.2986935866983374e-05,
+      "loss": 0.1057,
+      "step": 10606
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.298099762470309e-05,
+      "loss": 0.0479,
+      "step": 10607
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.297505938242281e-05,
+      "loss": 0.0213,
+      "step": 10608
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.296912114014252e-05,
+      "loss": 0.0474,
+      "step": 10609
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.2963182897862235e-05,
+      "loss": 0.2277,
+      "step": 10610
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.295724465558195e-05,
+      "loss": 0.0528,
+      "step": 10611
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.295130641330167e-05,
+      "loss": 0.0608,
+      "step": 10612
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.2945368171021376e-05,
+      "loss": 0.1185,
+      "step": 10613
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.293942992874109e-05,
+      "loss": 0.1584,
+      "step": 10614
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.293349168646081e-05,
+      "loss": 0.1263,
+      "step": 10615
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.2927553444180524e-05,
+      "loss": 0.1345,
+      "step": 10616
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.292161520190024e-05,
+      "loss": 0.1294,
+      "step": 10617
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.291567695961996e-05,
+      "loss": 0.1253,
+      "step": 10618
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.290973871733967e-05,
+      "loss": 0.1018,
+      "step": 10619
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.2903800475059385e-05,
+      "loss": 0.0558,
+      "step": 10620
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.28978622327791e-05,
+      "loss": 0.1302,
+      "step": 10621
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.289192399049882e-05,
+      "loss": 0.1108,
+      "step": 10622
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.2885985748218525e-05,
+      "loss": 0.1099,
+      "step": 10623
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.288004750593824e-05,
+      "loss": 0.0808,
+      "step": 10624
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.287410926365796e-05,
+      "loss": 0.0796,
+      "step": 10625
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.286817102137767e-05,
+      "loss": 0.1081,
+      "step": 10626
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.2862232779097387e-05,
+      "loss": 0.1156,
+      "step": 10627
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.285629453681711e-05,
+      "loss": 0.0967,
+      "step": 10628
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.285035629453682e-05,
+      "loss": 0.0935,
+      "step": 10629
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.2844418052256534e-05,
+      "loss": 0.147,
+      "step": 10630
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.283847980997625e-05,
+      "loss": 0.0847,
+      "step": 10631
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.283254156769597e-05,
+      "loss": 0.0626,
+      "step": 10632
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.282660332541568e-05,
+      "loss": 0.0677,
+      "step": 10633
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.282066508313539e-05,
+      "loss": 0.0874,
+      "step": 10634
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.281472684085511e-05,
+      "loss": 0.0648,
+      "step": 10635
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.280878859857482e-05,
+      "loss": 0.1277,
+      "step": 10636
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.2802850356294536e-05,
+      "loss": 0.0757,
+      "step": 10637
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.2796912114014256e-05,
+      "loss": 0.0756,
+      "step": 10638
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.279097387173397e-05,
+      "loss": 0.0592,
+      "step": 10639
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.2785035629453683e-05,
+      "loss": 0.0754,
+      "step": 10640
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.27790973871734e-05,
+      "loss": 0.1019,
+      "step": 10641
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.277315914489312e-05,
+      "loss": 0.1175,
+      "step": 10642
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.276722090261283e-05,
+      "loss": 0.0707,
+      "step": 10643
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.2761282660332545e-05,
+      "loss": 0.0547,
+      "step": 10644
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.275534441805226e-05,
+      "loss": 0.0511,
+      "step": 10645
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.274940617577197e-05,
+      "loss": 0.0869,
+      "step": 10646
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.2743467933491685e-05,
+      "loss": 0.0599,
+      "step": 10647
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.2737529691211406e-05,
+      "loss": 0.061,
+      "step": 10648
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.273159144893112e-05,
+      "loss": 0.0896,
+      "step": 10649
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.272565320665083e-05,
+      "loss": 0.0811,
+      "step": 10650
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.2719714964370546e-05,
+      "loss": 0.054,
+      "step": 10651
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.271377672209027e-05,
+      "loss": 0.0481,
+      "step": 10652
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.270783847980998e-05,
+      "loss": 0.0629,
+      "step": 10653
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.2701900237529694e-05,
+      "loss": 0.0555,
+      "step": 10654
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.269596199524941e-05,
+      "loss": 0.0552,
+      "step": 10655
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.269002375296912e-05,
+      "loss": 0.0662,
+      "step": 10656
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.2684085510688835e-05,
+      "loss": 0.0621,
+      "step": 10657
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.2678147268408555e-05,
+      "loss": 0.0481,
+      "step": 10658
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.267220902612827e-05,
+      "loss": 0.0858,
+      "step": 10659
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.266627078384798e-05,
+      "loss": 0.0327,
+      "step": 10660
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.2660332541567696e-05,
+      "loss": 0.1439,
+      "step": 10661
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.2654394299287416e-05,
+      "loss": 0.0887,
+      "step": 10662
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.264845605700713e-05,
+      "loss": 0.1228,
+      "step": 10663
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.264251781472684e-05,
+      "loss": 0.1451,
+      "step": 10664
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.263657957244656e-05,
+      "loss": 0.1438,
+      "step": 10665
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.263064133016627e-05,
+      "loss": 0.1371,
+      "step": 10666
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.2624703087885984e-05,
+      "loss": 0.0812,
+      "step": 10667
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.2618764845605704e-05,
+      "loss": 0.1143,
+      "step": 10668
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.261282660332542e-05,
+      "loss": 0.0925,
+      "step": 10669
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.260688836104513e-05,
+      "loss": 0.101,
+      "step": 10670
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.260095011876485e-05,
+      "loss": 0.1009,
+      "step": 10671
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.2595011876484565e-05,
+      "loss": 0.1333,
+      "step": 10672
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.258907363420428e-05,
+      "loss": 0.0857,
+      "step": 10673
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.258313539192399e-05,
+      "loss": 0.0801,
+      "step": 10674
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.2577197149643706e-05,
+      "loss": 0.0974,
+      "step": 10675
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.257125890736342e-05,
+      "loss": 0.0934,
+      "step": 10676
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.256532066508313e-05,
+      "loss": 0.1569,
+      "step": 10677
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.2559382422802854e-05,
+      "loss": 0.1038,
+      "step": 10678
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.255344418052257e-05,
+      "loss": 0.0837,
+      "step": 10679
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.254750593824228e-05,
+      "loss": 0.0957,
+      "step": 10680
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.2541567695962e-05,
+      "loss": 0.0899,
+      "step": 10681
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.2535629453681715e-05,
+      "loss": 0.078,
+      "step": 10682
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.252969121140143e-05,
+      "loss": 0.078,
+      "step": 10683
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.252375296912114e-05,
+      "loss": 0.0862,
+      "step": 10684
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.2517814726840856e-05,
+      "loss": 0.0514,
+      "step": 10685
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.251187648456057e-05,
+      "loss": 0.074,
+      "step": 10686
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.250593824228028e-05,
+      "loss": 0.1007,
+      "step": 10687
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.25e-05,
+      "loss": 0.0618,
+      "step": 10688
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.249406175771972e-05,
+      "loss": 0.0453,
+      "step": 10689
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.248812351543943e-05,
+      "loss": 0.0543,
+      "step": 10690
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.248218527315915e-05,
+      "loss": 0.0455,
+      "step": 10691
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.2476247030878864e-05,
+      "loss": 0.1019,
+      "step": 10692
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.247030878859858e-05,
+      "loss": 0.1007,
+      "step": 10693
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.246437054631829e-05,
+      "loss": 0.0532,
+      "step": 10694
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.245843230403801e-05,
+      "loss": 0.0596,
+      "step": 10695
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.245249406175772e-05,
+      "loss": 0.0438,
+      "step": 10696
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.244655581947743e-05,
+      "loss": 0.0946,
+      "step": 10697
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.244061757719715e-05,
+      "loss": 0.0718,
+      "step": 10698
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.2434679334916866e-05,
+      "loss": 0.0501,
+      "step": 10699
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.242874109263658e-05,
+      "loss": 0.0607,
+      "step": 10700
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.24228028503563e-05,
+      "loss": 0.0814,
+      "step": 10701
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.2416864608076014e-05,
+      "loss": 0.0291,
+      "step": 10702
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.241092636579573e-05,
+      "loss": 0.081,
+      "step": 10703
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.240498812351544e-05,
+      "loss": 0.0396,
+      "step": 10704
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.239904988123516e-05,
+      "loss": 0.1214,
+      "step": 10705
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.239311163895487e-05,
+      "loss": 0.1152,
+      "step": 10706
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.238717339667458e-05,
+      "loss": 0.1005,
+      "step": 10707
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.23812351543943e-05,
+      "loss": 0.0784,
+      "step": 10708
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.2375296912114015e-05,
+      "loss": 0.1369,
+      "step": 10709
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.236935866983373e-05,
+      "loss": 0.0858,
+      "step": 10710
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.236342042755345e-05,
+      "loss": 0.0982,
+      "step": 10711
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.235748218527316e-05,
+      "loss": 0.0743,
+      "step": 10712
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.2351543942992876e-05,
+      "loss": 0.0414,
+      "step": 10713
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.234560570071259e-05,
+      "loss": 0.0985,
+      "step": 10714
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.233966745843231e-05,
+      "loss": 0.0457,
+      "step": 10715
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.233372921615202e-05,
+      "loss": 0.0661,
+      "step": 10716
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.232779097387173e-05,
+      "loss": 0.0883,
+      "step": 10717
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.232185273159145e-05,
+      "loss": 0.0798,
+      "step": 10718
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.2315914489311165e-05,
+      "loss": 0.0893,
+      "step": 10719
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.230997624703088e-05,
+      "loss": 0.0701,
+      "step": 10720
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.23040380047506e-05,
+      "loss": 0.0727,
+      "step": 10721
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.229809976247031e-05,
+      "loss": 0.076,
+      "step": 10722
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.2292161520190026e-05,
+      "loss": 0.0928,
+      "step": 10723
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.228622327790974e-05,
+      "loss": 0.0605,
+      "step": 10724
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.228028503562946e-05,
+      "loss": 0.0531,
+      "step": 10725
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.227434679334917e-05,
+      "loss": 0.069,
+      "step": 10726
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.226840855106888e-05,
+      "loss": 0.0618,
+      "step": 10727
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.22624703087886e-05,
+      "loss": 0.0893,
+      "step": 10728
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.2256532066508314e-05,
+      "loss": 0.0742,
+      "step": 10729
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.225059382422803e-05,
+      "loss": 0.0635,
+      "step": 10730
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.224465558194775e-05,
+      "loss": 0.0687,
+      "step": 10731
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.223871733966746e-05,
+      "loss": 0.0765,
+      "step": 10732
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.2232779097387175e-05,
+      "loss": 0.0602,
+      "step": 10733
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.222684085510689e-05,
+      "loss": 0.0543,
+      "step": 10734
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.222090261282661e-05,
+      "loss": 0.0356,
+      "step": 10735
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.221496437054632e-05,
+      "loss": 0.0557,
+      "step": 10736
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.220902612826603e-05,
+      "loss": 0.0724,
+      "step": 10737
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.220308788598575e-05,
+      "loss": 0.0514,
+      "step": 10738
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.2197149643705463e-05,
+      "loss": 0.0352,
+      "step": 10739
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.219121140142518e-05,
+      "loss": 0.066,
+      "step": 10740
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.21852731591449e-05,
+      "loss": 0.0342,
+      "step": 10741
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.217933491686461e-05,
+      "loss": 0.0434,
+      "step": 10742
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.2173396674584325e-05,
+      "loss": 0.0595,
+      "step": 10743
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.216745843230404e-05,
+      "loss": 0.0683,
+      "step": 10744
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.216152019002376e-05,
+      "loss": 0.0345,
+      "step": 10745
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.215558194774347e-05,
+      "loss": 0.0291,
+      "step": 10746
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.2149643705463186e-05,
+      "loss": 0.0957,
+      "step": 10747
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.21437054631829e-05,
+      "loss": 0.0674,
+      "step": 10748
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.213776722090261e-05,
+      "loss": 0.0431,
+      "step": 10749
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.2131828978622326e-05,
+      "loss": 0.0253,
+      "step": 10750
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.212589073634205e-05,
+      "loss": 0.0454,
+      "step": 10751
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.211995249406176e-05,
+      "loss": 0.0627,
+      "step": 10752
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.2114014251781474e-05,
+      "loss": 0.1035,
+      "step": 10753
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.210807600950119e-05,
+      "loss": 0.0848,
+      "step": 10754
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.210213776722091e-05,
+      "loss": 0.127,
+      "step": 10755
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.209619952494062e-05,
+      "loss": 0.0878,
+      "step": 10756
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.2090261282660335e-05,
+      "loss": 0.0901,
+      "step": 10757
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.208432304038005e-05,
+      "loss": 0.0681,
+      "step": 10758
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.207838479809976e-05,
+      "loss": 0.0675,
+      "step": 10759
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.2072446555819476e-05,
+      "loss": 0.0902,
+      "step": 10760
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.2066508313539196e-05,
+      "loss": 0.0819,
+      "step": 10761
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.206057007125891e-05,
+      "loss": 0.072,
+      "step": 10762
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.205463182897862e-05,
+      "loss": 0.0762,
+      "step": 10763
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.204869358669834e-05,
+      "loss": 0.0926,
+      "step": 10764
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.204275534441806e-05,
+      "loss": 0.0594,
+      "step": 10765
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.203681710213777e-05,
+      "loss": 0.0929,
+      "step": 10766
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.2030878859857484e-05,
+      "loss": 0.0701,
+      "step": 10767
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.20249406175772e-05,
+      "loss": 0.0565,
+      "step": 10768
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.201900237529691e-05,
+      "loss": 0.0405,
+      "step": 10769
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.2013064133016625e-05,
+      "loss": 0.0488,
+      "step": 10770
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.2007125890736345e-05,
+      "loss": 0.069,
+      "step": 10771
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.200118764845606e-05,
+      "loss": 0.0345,
+      "step": 10772
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.199524940617577e-05,
+      "loss": 0.21,
+      "step": 10773
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.1989311163895486e-05,
+      "loss": 0.1231,
+      "step": 10774
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.1983372921615207e-05,
+      "loss": 0.0634,
+      "step": 10775
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.197743467933492e-05,
+      "loss": 0.054,
+      "step": 10776
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.1971496437054634e-05,
+      "loss": 0.0919,
+      "step": 10777
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.1965558194774354e-05,
+      "loss": 0.0547,
+      "step": 10778
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.195961995249406e-05,
+      "loss": 0.0572,
+      "step": 10779
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.1953681710213774e-05,
+      "loss": 0.0634,
+      "step": 10780
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.1947743467933495e-05,
+      "loss": 0.07,
+      "step": 10781
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.194180522565321e-05,
+      "loss": 0.071,
+      "step": 10782
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.193586698337292e-05,
+      "loss": 0.0729,
+      "step": 10783
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.1929928741092636e-05,
+      "loss": 0.0522,
+      "step": 10784
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.1923990498812356e-05,
+      "loss": 0.0684,
+      "step": 10785
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.191805225653207e-05,
+      "loss": 0.0424,
+      "step": 10786
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.191211401425178e-05,
+      "loss": 0.1056,
+      "step": 10787
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.1906175771971503e-05,
+      "loss": 0.0684,
+      "step": 10788
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.190023752969121e-05,
+      "loss": 0.0335,
+      "step": 10789
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.1894299287410924e-05,
+      "loss": 0.0436,
+      "step": 10790
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.1888361045130644e-05,
+      "loss": 0.0566,
+      "step": 10791
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.188242280285036e-05,
+      "loss": 0.0619,
+      "step": 10792
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.187648456057007e-05,
+      "loss": 0.0456,
+      "step": 10793
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.187054631828979e-05,
+      "loss": 0.0489,
+      "step": 10794
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.1864608076009505e-05,
+      "loss": 0.0583,
+      "step": 10795
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.185866983372922e-05,
+      "loss": 0.0764,
+      "step": 10796
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.185273159144893e-05,
+      "loss": 0.0309,
+      "step": 10797
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.184679334916865e-05,
+      "loss": 0.1141,
+      "step": 10798
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.184085510688836e-05,
+      "loss": 0.061,
+      "step": 10799
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.183491686460807e-05,
+      "loss": 0.038,
+      "step": 10800
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.1828978622327794e-05,
+      "loss": 0.0579,
+      "step": 10801
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.182304038004751e-05,
+      "loss": 0.0412,
+      "step": 10802
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.181710213776722e-05,
+      "loss": 0.0283,
+      "step": 10803
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.181116389548694e-05,
+      "loss": 0.032,
+      "step": 10804
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.1805225653206655e-05,
+      "loss": 0.1153,
+      "step": 10805
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.179928741092637e-05,
+      "loss": 0.1192,
+      "step": 10806
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.179334916864608e-05,
+      "loss": 0.0987,
+      "step": 10807
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.17874109263658e-05,
+      "loss": 0.1057,
+      "step": 10808
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.1781472684085516e-05,
+      "loss": 0.1297,
+      "step": 10809
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.177553444180522e-05,
+      "loss": 0.0642,
+      "step": 10810
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.176959619952494e-05,
+      "loss": 0.061,
+      "step": 10811
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.1763657957244656e-05,
+      "loss": 0.0668,
+      "step": 10812
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.175771971496437e-05,
+      "loss": 0.0741,
+      "step": 10813
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.175178147268409e-05,
+      "loss": 0.0585,
+      "step": 10814
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.1745843230403804e-05,
+      "loss": 0.0633,
+      "step": 10815
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.173990498812352e-05,
+      "loss": 0.1068,
+      "step": 10816
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.173396674584323e-05,
+      "loss": 0.0778,
+      "step": 10817
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.172802850356295e-05,
+      "loss": 0.0755,
+      "step": 10818
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.1722090261282665e-05,
+      "loss": 0.1041,
+      "step": 10819
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.171615201900237e-05,
+      "loss": 0.0833,
+      "step": 10820
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.171021377672209e-05,
+      "loss": 0.0833,
+      "step": 10821
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.1704275534441806e-05,
+      "loss": 0.0542,
+      "step": 10822
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.169833729216152e-05,
+      "loss": 0.0498,
+      "step": 10823
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.169239904988124e-05,
+      "loss": 0.0698,
+      "step": 10824
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.168646080760095e-05,
+      "loss": 0.0813,
+      "step": 10825
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.168052256532067e-05,
+      "loss": 0.0951,
+      "step": 10826
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.167458432304038e-05,
+      "loss": 0.0682,
+      "step": 10827
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.16686460807601e-05,
+      "loss": 0.0609,
+      "step": 10828
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.1662707838479814e-05,
+      "loss": 0.0635,
+      "step": 10829
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.165676959619953e-05,
+      "loss": 0.0484,
+      "step": 10830
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.165083135391924e-05,
+      "loss": 0.0589,
+      "step": 10831
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.1644893111638955e-05,
+      "loss": 0.1014,
+      "step": 10832
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.163895486935867e-05,
+      "loss": 0.0675,
+      "step": 10833
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.163301662707839e-05,
+      "loss": 0.0617,
+      "step": 10834
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.16270783847981e-05,
+      "loss": 0.0684,
+      "step": 10835
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.1621140142517816e-05,
+      "loss": 0.0691,
+      "step": 10836
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.161520190023753e-05,
+      "loss": 0.0334,
+      "step": 10837
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.160926365795725e-05,
+      "loss": 0.0473,
+      "step": 10838
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.1603325415676964e-05,
+      "loss": 0.0437,
+      "step": 10839
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.159738717339668e-05,
+      "loss": 0.0693,
+      "step": 10840
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.159144893111639e-05,
+      "loss": 0.0422,
+      "step": 10841
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.1585510688836105e-05,
+      "loss": 0.0391,
+      "step": 10842
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.157957244655582e-05,
+      "loss": 0.0679,
+      "step": 10843
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.157363420427554e-05,
+      "loss": 0.0422,
+      "step": 10844
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.156769596199525e-05,
+      "loss": 0.05,
+      "step": 10845
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.1561757719714966e-05,
+      "loss": 0.0333,
+      "step": 10846
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.155581947743468e-05,
+      "loss": 0.0628,
+      "step": 10847
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.15498812351544e-05,
+      "loss": 0.0466,
+      "step": 10848
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.154394299287411e-05,
+      "loss": 0.0626,
+      "step": 10849
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.153800475059383e-05,
+      "loss": 0.0265,
+      "step": 10850
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.153206650831354e-05,
+      "loss": 0.0648,
+      "step": 10851
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.1526128266033254e-05,
+      "loss": 0.0953,
+      "step": 10852
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.152019002375297e-05,
+      "loss": 0.0258,
+      "step": 10853
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.151425178147269e-05,
+      "loss": 0.0203,
+      "step": 10854
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.15083135391924e-05,
+      "loss": 0.1129,
+      "step": 10855
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.1502375296912115e-05,
+      "loss": 0.0915,
+      "step": 10856
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.149643705463183e-05,
+      "loss": 0.0758,
+      "step": 10857
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.149049881235155e-05,
+      "loss": 0.0655,
+      "step": 10858
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.148456057007126e-05,
+      "loss": 0.0867,
+      "step": 10859
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.1478622327790976e-05,
+      "loss": 0.0609,
+      "step": 10860
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.1472684085510696e-05,
+      "loss": 0.0575,
+      "step": 10861
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.14667458432304e-05,
+      "loss": 0.0836,
+      "step": 10862
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.146080760095012e-05,
+      "loss": 0.0904,
+      "step": 10863
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.145486935866984e-05,
+      "loss": 0.0627,
+      "step": 10864
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.144893111638955e-05,
+      "loss": 0.0514,
+      "step": 10865
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.1442992874109264e-05,
+      "loss": 0.0679,
+      "step": 10866
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.143705463182898e-05,
+      "loss": 0.0901,
+      "step": 10867
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.14311163895487e-05,
+      "loss": 0.0513,
+      "step": 10868
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.142517814726841e-05,
+      "loss": 0.1044,
+      "step": 10869
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.1419239904988125e-05,
+      "loss": 0.056,
+      "step": 10870
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.1413301662707846e-05,
+      "loss": 0.0486,
+      "step": 10871
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.140736342042755e-05,
+      "loss": 0.0846,
+      "step": 10872
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.1401425178147266e-05,
+      "loss": 0.0803,
+      "step": 10873
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.1395486935866987e-05,
+      "loss": 0.0795,
+      "step": 10874
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.13895486935867e-05,
+      "loss": 0.0958,
+      "step": 10875
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.1383610451306414e-05,
+      "loss": 0.0846,
+      "step": 10876
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.137767220902613e-05,
+      "loss": 0.0671,
+      "step": 10877
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.137173396674585e-05,
+      "loss": 0.0451,
+      "step": 10878
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.136579572446556e-05,
+      "loss": 0.0763,
+      "step": 10879
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.1359857482185275e-05,
+      "loss": 0.067,
+      "step": 10880
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.1353919239904995e-05,
+      "loss": 0.0863,
+      "step": 10881
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.13479809976247e-05,
+      "loss": 0.0526,
+      "step": 10882
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.1342042755344416e-05,
+      "loss": 0.0627,
+      "step": 10883
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.1336104513064136e-05,
+      "loss": 0.048,
+      "step": 10884
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.133016627078385e-05,
+      "loss": 0.0459,
+      "step": 10885
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.132422802850356e-05,
+      "loss": 0.059,
+      "step": 10886
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.131828978622328e-05,
+      "loss": 0.0629,
+      "step": 10887
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.1312351543943e-05,
+      "loss": 0.0678,
+      "step": 10888
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.130641330166271e-05,
+      "loss": 0.0604,
+      "step": 10889
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.1300475059382424e-05,
+      "loss": 0.0399,
+      "step": 10890
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.1294536817102145e-05,
+      "loss": 0.0562,
+      "step": 10891
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.128859857482186e-05,
+      "loss": 0.0499,
+      "step": 10892
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.1282660332541565e-05,
+      "loss": 0.0493,
+      "step": 10893
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.1276722090261285e-05,
+      "loss": 0.1096,
+      "step": 10894
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.1270783847981e-05,
+      "loss": 0.0515,
+      "step": 10895
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.126484560570071e-05,
+      "loss": 0.0417,
+      "step": 10896
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.1258907363420426e-05,
+      "loss": 0.0383,
+      "step": 10897
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.1252969121140146e-05,
+      "loss": 0.0234,
+      "step": 10898
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.124703087885986e-05,
+      "loss": 0.027,
+      "step": 10899
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.1241092636579574e-05,
+      "loss": 0.0235,
+      "step": 10900
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.1235154394299294e-05,
+      "loss": 0.0442,
+      "step": 10901
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.122921615201901e-05,
+      "loss": 0.0341,
+      "step": 10902
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.1223277909738714e-05,
+      "loss": 0.0508,
+      "step": 10903
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.1217339667458435e-05,
+      "loss": 0.0896,
+      "step": 10904
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.121140142517815e-05,
+      "loss": 0.1236,
+      "step": 10905
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.120546318289786e-05,
+      "loss": 0.103,
+      "step": 10906
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.1199524940617575e-05,
+      "loss": 0.156,
+      "step": 10907
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.1193586698337296e-05,
+      "loss": 0.096,
+      "step": 10908
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.118764845605701e-05,
+      "loss": 0.0608,
+      "step": 10909
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.118171021377672e-05,
+      "loss": 0.0836,
+      "step": 10910
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.117577197149644e-05,
+      "loss": 0.1292,
+      "step": 10911
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.116983372921616e-05,
+      "loss": 0.0496,
+      "step": 10912
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.116389548693587e-05,
+      "loss": 0.0742,
+      "step": 10913
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.1157957244655584e-05,
+      "loss": 0.0809,
+      "step": 10914
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.11520190023753e-05,
+      "loss": 0.0883,
+      "step": 10915
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.114608076009501e-05,
+      "loss": 0.0695,
+      "step": 10916
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.114014251781473e-05,
+      "loss": 0.0748,
+      "step": 10917
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.1134204275534445e-05,
+      "loss": 0.0508,
+      "step": 10918
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.112826603325416e-05,
+      "loss": 0.0645,
+      "step": 10919
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.112232779097387e-05,
+      "loss": 0.0768,
+      "step": 10920
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.111638954869359e-05,
+      "loss": 0.0616,
+      "step": 10921
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.1110451306413306e-05,
+      "loss": 0.07,
+      "step": 10922
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.110451306413302e-05,
+      "loss": 0.0604,
+      "step": 10923
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.109857482185273e-05,
+      "loss": 0.1058,
+      "step": 10924
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.109263657957245e-05,
+      "loss": 0.0622,
+      "step": 10925
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.108669833729216e-05,
+      "loss": 0.0677,
+      "step": 10926
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.108076009501188e-05,
+      "loss": 0.0719,
+      "step": 10927
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.1074821852731594e-05,
+      "loss": 0.0554,
+      "step": 10928
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.106888361045131e-05,
+      "loss": 0.0541,
+      "step": 10929
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.106294536817102e-05,
+      "loss": 0.0488,
+      "step": 10930
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.105700712589074e-05,
+      "loss": 0.0466,
+      "step": 10931
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.1051068883610456e-05,
+      "loss": 0.08,
+      "step": 10932
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.104513064133017e-05,
+      "loss": 0.06,
+      "step": 10933
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.103919239904988e-05,
+      "loss": 0.0487,
+      "step": 10934
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.1033254156769596e-05,
+      "loss": 0.0773,
+      "step": 10935
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.102731591448931e-05,
+      "loss": 0.0834,
+      "step": 10936
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.102137767220903e-05,
+      "loss": 0.0518,
+      "step": 10937
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.1015439429928744e-05,
+      "loss": 0.0544,
+      "step": 10938
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.100950118764846e-05,
+      "loss": 0.0353,
+      "step": 10939
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.100356294536817e-05,
+      "loss": 0.0535,
+      "step": 10940
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.099762470308789e-05,
+      "loss": 0.029,
+      "step": 10941
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.0991686460807605e-05,
+      "loss": 0.0591,
+      "step": 10942
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.098574821852732e-05,
+      "loss": 0.0522,
+      "step": 10943
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.097980997624704e-05,
+      "loss": 0.0756,
+      "step": 10944
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.0973871733966746e-05,
+      "loss": 0.0263,
+      "step": 10945
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.096793349168646e-05,
+      "loss": 0.0692,
+      "step": 10946
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.096199524940618e-05,
+      "loss": 0.0515,
+      "step": 10947
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.095605700712589e-05,
+      "loss": 0.0597,
+      "step": 10948
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.095011876484561e-05,
+      "loss": 0.0549,
+      "step": 10949
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.094418052256532e-05,
+      "loss": 0.0724,
+      "step": 10950
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.093824228028504e-05,
+      "loss": 0.0343,
+      "step": 10951
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.0932304038004754e-05,
+      "loss": 0.0291,
+      "step": 10952
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.092636579572447e-05,
+      "loss": 0.0297,
+      "step": 10953
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.092042755344419e-05,
+      "loss": 0.0383,
+      "step": 10954
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.0914489311163895e-05,
+      "loss": 0.128,
+      "step": 10955
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.090855106888361e-05,
+      "loss": 0.1061,
+      "step": 10956
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.090261282660333e-05,
+      "loss": 0.1052,
+      "step": 10957
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.089667458432304e-05,
+      "loss": 0.0825,
+      "step": 10958
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.0890736342042756e-05,
+      "loss": 0.1376,
+      "step": 10959
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.088479809976247e-05,
+      "loss": 0.0896,
+      "step": 10960
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.087885985748219e-05,
+      "loss": 0.0613,
+      "step": 10961
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.0872921615201904e-05,
+      "loss": 0.0719,
+      "step": 10962
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.086698337292162e-05,
+      "loss": 0.0835,
+      "step": 10963
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.086104513064134e-05,
+      "loss": 0.1017,
+      "step": 10964
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.0855106888361044e-05,
+      "loss": 0.0908,
+      "step": 10965
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.084916864608076e-05,
+      "loss": 0.0323,
+      "step": 10966
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.084323040380048e-05,
+      "loss": 0.0796,
+      "step": 10967
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.083729216152019e-05,
+      "loss": 0.0717,
+      "step": 10968
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.0831353919239905e-05,
+      "loss": 0.0715,
+      "step": 10969
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.082541567695962e-05,
+      "loss": 0.0724,
+      "step": 10970
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.081947743467934e-05,
+      "loss": 0.0952,
+      "step": 10971
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.081353919239905e-05,
+      "loss": 0.0795,
+      "step": 10972
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.0807600950118767e-05,
+      "loss": 0.0468,
+      "step": 10973
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.080166270783849e-05,
+      "loss": 0.0776,
+      "step": 10974
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.07957244655582e-05,
+      "loss": 0.0567,
+      "step": 10975
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.078978622327791e-05,
+      "loss": 0.0458,
+      "step": 10976
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.078384798099763e-05,
+      "loss": 0.072,
+      "step": 10977
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.077790973871734e-05,
+      "loss": 0.0537,
+      "step": 10978
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.0771971496437055e-05,
+      "loss": 0.0387,
+      "step": 10979
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.076603325415677e-05,
+      "loss": 0.0583,
+      "step": 10980
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.076009501187649e-05,
+      "loss": 0.0598,
+      "step": 10981
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.07541567695962e-05,
+      "loss": 0.0429,
+      "step": 10982
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.0748218527315916e-05,
+      "loss": 0.0874,
+      "step": 10983
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.0742280285035636e-05,
+      "loss": 0.0407,
+      "step": 10984
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.073634204275535e-05,
+      "loss": 0.0511,
+      "step": 10985
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.073040380047506e-05,
+      "loss": 0.0596,
+      "step": 10986
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.072446555819478e-05,
+      "loss": 0.0427,
+      "step": 10987
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.071852731591449e-05,
+      "loss": 0.0359,
+      "step": 10988
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.0712589073634204e-05,
+      "loss": 0.0546,
+      "step": 10989
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.070665083135392e-05,
+      "loss": 0.04,
+      "step": 10990
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.070071258907364e-05,
+      "loss": 0.0381,
+      "step": 10991
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.069477434679335e-05,
+      "loss": 0.0647,
+      "step": 10992
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.0688836104513065e-05,
+      "loss": 0.0376,
+      "step": 10993
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.0682897862232786e-05,
+      "loss": 0.0467,
+      "step": 10994
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.06769596199525e-05,
+      "loss": 0.0684,
+      "step": 10995
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.0671021377672206e-05,
+      "loss": 0.054,
+      "step": 10996
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.0665083135391926e-05,
+      "loss": 0.1426,
+      "step": 10997
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.065914489311164e-05,
+      "loss": 0.0454,
+      "step": 10998
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.0653206650831354e-05,
+      "loss": 0.1056,
+      "step": 10999
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.064726840855107e-05,
+      "loss": 0.0424,
+      "step": 11000
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.064133016627079e-05,
+      "loss": 0.0177,
+      "step": 11001
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.06353919239905e-05,
+      "loss": 0.0747,
+      "step": 11002
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.0629453681710215e-05,
+      "loss": 0.0387,
+      "step": 11003
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.0623515439429935e-05,
+      "loss": 0.0882,
+      "step": 11004
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.061757719714965e-05,
+      "loss": 0.0921,
+      "step": 11005
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.061163895486936e-05,
+      "loss": 0.0945,
+      "step": 11006
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.0605700712589076e-05,
+      "loss": 0.0929,
+      "step": 11007
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.059976247030879e-05,
+      "loss": 0.091,
+      "step": 11008
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.05938242280285e-05,
+      "loss": 0.0733,
+      "step": 11009
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.0587885985748216e-05,
+      "loss": 0.0759,
+      "step": 11010
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.058194774346794e-05,
+      "loss": 0.0669,
+      "step": 11011
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.057600950118765e-05,
+      "loss": 0.0625,
+      "step": 11012
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.0570071258907364e-05,
+      "loss": 0.0647,
+      "step": 11013
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.0564133016627084e-05,
+      "loss": 0.093,
+      "step": 11014
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.05581947743468e-05,
+      "loss": 0.1044,
+      "step": 11015
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.055225653206651e-05,
+      "loss": 0.0586,
+      "step": 11016
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.0546318289786225e-05,
+      "loss": 0.0816,
+      "step": 11017
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.054038004750594e-05,
+      "loss": 0.0472,
+      "step": 11018
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.053444180522565e-05,
+      "loss": 0.0465,
+      "step": 11019
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.0528503562945366e-05,
+      "loss": 0.0491,
+      "step": 11020
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.0522565320665086e-05,
+      "loss": 0.0683,
+      "step": 11021
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.05166270783848e-05,
+      "loss": 0.066,
+      "step": 11022
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.051068883610451e-05,
+      "loss": 0.0687,
+      "step": 11023
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.0504750593824234e-05,
+      "loss": 0.0576,
+      "step": 11024
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.049881235154395e-05,
+      "loss": 0.0587,
+      "step": 11025
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.049287410926366e-05,
+      "loss": 0.0749,
+      "step": 11026
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.0486935866983374e-05,
+      "loss": 0.0548,
+      "step": 11027
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.048099762470309e-05,
+      "loss": 0.1048,
+      "step": 11028
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.04750593824228e-05,
+      "loss": 0.0452,
+      "step": 11029
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.0469121140142515e-05,
+      "loss": 0.0407,
+      "step": 11030
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.0463182897862236e-05,
+      "loss": 0.052,
+      "step": 11031
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.045724465558195e-05,
+      "loss": 0.0606,
+      "step": 11032
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.045130641330166e-05,
+      "loss": 0.0493,
+      "step": 11033
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.044536817102138e-05,
+      "loss": 0.0722,
+      "step": 11034
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.04394299287411e-05,
+      "loss": 0.0712,
+      "step": 11035
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.043349168646081e-05,
+      "loss": 0.0635,
+      "step": 11036
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.0427553444180524e-05,
+      "loss": 0.0496,
+      "step": 11037
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.042161520190024e-05,
+      "loss": 0.0432,
+      "step": 11038
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.041567695961995e-05,
+      "loss": 0.0542,
+      "step": 11039
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.040973871733967e-05,
+      "loss": 0.0226,
+      "step": 11040
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.0403800475059385e-05,
+      "loss": 0.0181,
+      "step": 11041
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.03978622327791e-05,
+      "loss": 0.0421,
+      "step": 11042
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.039192399049881e-05,
+      "loss": 0.059,
+      "step": 11043
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.038598574821853e-05,
+      "loss": 0.0697,
+      "step": 11044
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.0380047505938246e-05,
+      "loss": 0.0411,
+      "step": 11045
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.037410926365796e-05,
+      "loss": 0.0692,
+      "step": 11046
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.036817102137767e-05,
+      "loss": 0.0517,
+      "step": 11047
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.036223277909739e-05,
+      "loss": 0.0563,
+      "step": 11048
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.03562945368171e-05,
+      "loss": 0.0993,
+      "step": 11049
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.035035629453682e-05,
+      "loss": 0.0569,
+      "step": 11050
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.0344418052256534e-05,
+      "loss": 0.031,
+      "step": 11051
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.033847980997625e-05,
+      "loss": 0.077,
+      "step": 11052
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.033254156769596e-05,
+      "loss": 0.044,
+      "step": 11053
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.032660332541568e-05,
+      "loss": 0.0906,
+      "step": 11054
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.0320665083135395e-05,
+      "loss": 0.1287,
+      "step": 11055
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.031472684085511e-05,
+      "loss": 0.091,
+      "step": 11056
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.030878859857482e-05,
+      "loss": 0.0869,
+      "step": 11057
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.030285035629454e-05,
+      "loss": 0.0884,
+      "step": 11058
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.029691211401425e-05,
+      "loss": 0.1088,
+      "step": 11059
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.029097387173397e-05,
+      "loss": 0.092,
+      "step": 11060
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.0285035629453684e-05,
+      "loss": 0.0691,
+      "step": 11061
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.02790973871734e-05,
+      "loss": 0.0618,
+      "step": 11062
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.027315914489311e-05,
+      "loss": 0.0685,
+      "step": 11063
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.026722090261283e-05,
+      "loss": 0.0834,
+      "step": 11064
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.0261282660332545e-05,
+      "loss": 0.0786,
+      "step": 11065
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.025534441805226e-05,
+      "loss": 0.0972,
+      "step": 11066
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.024940617577198e-05,
+      "loss": 0.0651,
+      "step": 11067
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.024346793349169e-05,
+      "loss": 0.0756,
+      "step": 11068
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.02375296912114e-05,
+      "loss": 0.0644,
+      "step": 11069
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.023159144893112e-05,
+      "loss": 0.0707,
+      "step": 11070
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.022565320665083e-05,
+      "loss": 0.1432,
+      "step": 11071
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.0219714964370547e-05,
+      "loss": 0.0474,
+      "step": 11072
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.021377672209026e-05,
+      "loss": 0.0691,
+      "step": 11073
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.020783847980998e-05,
+      "loss": 0.0829,
+      "step": 11074
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.0201900237529694e-05,
+      "loss": 0.0609,
+      "step": 11075
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.019596199524941e-05,
+      "loss": 0.1215,
+      "step": 11076
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.019002375296913e-05,
+      "loss": 0.0518,
+      "step": 11077
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.018408551068884e-05,
+      "loss": 0.0729,
+      "step": 11078
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.017814726840855e-05,
+      "loss": 0.0533,
+      "step": 11079
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.017220902612827e-05,
+      "loss": 0.0799,
+      "step": 11080
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.016627078384798e-05,
+      "loss": 0.0569,
+      "step": 11081
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.0160332541567696e-05,
+      "loss": 0.0669,
+      "step": 11082
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.015439429928741e-05,
+      "loss": 0.07,
+      "step": 11083
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.014845605700713e-05,
+      "loss": 0.0642,
+      "step": 11084
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.0142517814726843e-05,
+      "loss": 0.0565,
+      "step": 11085
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.013657957244656e-05,
+      "loss": 0.0623,
+      "step": 11086
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.013064133016628e-05,
+      "loss": 0.0608,
+      "step": 11087
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.012470308788599e-05,
+      "loss": 0.0606,
+      "step": 11088
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.0118764845605705e-05,
+      "loss": 0.0624,
+      "step": 11089
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.011282660332542e-05,
+      "loss": 0.0506,
+      "step": 11090
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.010688836104513e-05,
+      "loss": 0.0388,
+      "step": 11091
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.0100950118764845e-05,
+      "loss": 0.0374,
+      "step": 11092
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.009501187648456e-05,
+      "loss": 0.0769,
+      "step": 11093
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.008907363420428e-05,
+      "loss": 0.0521,
+      "step": 11094
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.008313539192399e-05,
+      "loss": 0.0363,
+      "step": 11095
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.0077197149643706e-05,
+      "loss": 0.0767,
+      "step": 11096
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.007125890736343e-05,
+      "loss": 0.0573,
+      "step": 11097
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.006532066508314e-05,
+      "loss": 0.0352,
+      "step": 11098
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.0059382422802854e-05,
+      "loss": 0.0631,
+      "step": 11099
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.005344418052257e-05,
+      "loss": 0.0592,
+      "step": 11100
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.004750593824228e-05,
+      "loss": 0.0284,
+      "step": 11101
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.0041567695961995e-05,
+      "loss": 0.0577,
+      "step": 11102
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.003562945368171e-05,
+      "loss": 0.0606,
+      "step": 11103
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.002969121140143e-05,
+      "loss": 0.1483,
+      "step": 11104
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.002375296912114e-05,
+      "loss": 0.1192,
+      "step": 11105
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.0017814726840856e-05,
+      "loss": 0.0965,
+      "step": 11106
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.0011876484560576e-05,
+      "loss": 0.1062,
+      "step": 11107
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.000593824228029e-05,
+      "loss": 0.0763,
+      "step": 11108
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4e-05,
+      "loss": 0.0729,
+      "step": 11109
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 3.999406175771972e-05,
+      "loss": 0.0529,
+      "step": 11110
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 3.998812351543943e-05,
+      "loss": 0.0611,
+      "step": 11111
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 3.9982185273159144e-05,
+      "loss": 0.0661,
+      "step": 11112
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 3.997624703087886e-05,
+      "loss": 0.0892,
+      "step": 11113
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 3.997030878859858e-05,
+      "loss": 0.1042,
+      "step": 11114
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 3.996437054631829e-05,
+      "loss": 0.1001,
+      "step": 11115
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 3.9958432304038005e-05,
+      "loss": 0.0777,
+      "step": 11116
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 3.9952494061757725e-05,
+      "loss": 0.0785,
+      "step": 11117
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 3.994655581947744e-05,
+      "loss": 0.1625,
+      "step": 11118
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 3.994061757719715e-05,
+      "loss": 0.0869,
+      "step": 11119
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 3.9934679334916866e-05,
+      "loss": 0.0936,
+      "step": 11120
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 3.992874109263658e-05,
+      "loss": 0.0615,
+      "step": 11121
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 3.992280285035629e-05,
+      "loss": 0.0721,
+      "step": 11122
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 3.991686460807601e-05,
+      "loss": 0.0801,
+      "step": 11123
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 3.991092636579573e-05,
+      "loss": 0.0488,
+      "step": 11124
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 3.990498812351544e-05,
+      "loss": 0.0647,
+      "step": 11125
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 3.9899049881235154e-05,
+      "loss": 0.0718,
+      "step": 11126
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 3.9893111638954875e-05,
+      "loss": 0.062,
+      "step": 11127
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 3.988717339667459e-05,
+      "loss": 0.0674,
+      "step": 11128
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 3.98812351543943e-05,
+      "loss": 0.0464,
+      "step": 11129
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 3.9875296912114016e-05,
+      "loss": 0.0493,
+      "step": 11130
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 3.986935866983373e-05,
+      "loss": 0.0703,
+      "step": 11131
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 3.986342042755344e-05,
+      "loss": 0.105,
+      "step": 11132
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 3.9857482185273156e-05,
+      "loss": 0.0668,
+      "step": 11133
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 3.985154394299288e-05,
+      "loss": 0.0428,
+      "step": 11134
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 3.984560570071259e-05,
+      "loss": 0.0528,
+      "step": 11135
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 3.9839667458432304e-05,
+      "loss": 0.0691,
+      "step": 11136
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 3.9833729216152024e-05,
+      "loss": 0.0677,
+      "step": 11137
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 3.982779097387174e-05,
+      "loss": 0.0655,
+      "step": 11138
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 3.982185273159145e-05,
+      "loss": 0.0606,
+      "step": 11139
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 3.9815914489311165e-05,
+      "loss": 0.0365,
+      "step": 11140
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 3.9809976247030885e-05,
+      "loss": 0.0583,
+      "step": 11141
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 3.980403800475059e-05,
+      "loss": 0.0708,
+      "step": 11142
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 3.9798099762470306e-05,
+      "loss": 0.116,
+      "step": 11143
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 3.9792161520190026e-05,
+      "loss": 0.045,
+      "step": 11144
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 3.978622327790974e-05,
+      "loss": 0.0376,
+      "step": 11145
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 3.978028503562945e-05,
+      "loss": 0.0534,
+      "step": 11146
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 3.9774346793349174e-05,
+      "loss": 0.065,
+      "step": 11147
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 3.976840855106889e-05,
+      "loss": 0.0354,
+      "step": 11148
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 3.97624703087886e-05,
+      "loss": 0.0494,
+      "step": 11149
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 3.9756532066508314e-05,
+      "loss": 0.0591,
+      "step": 11150
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 3.9750593824228035e-05,
+      "loss": 0.0265,
+      "step": 11151
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 3.974465558194774e-05,
+      "loss": 0.1285,
+      "step": 11152
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 3.9738717339667455e-05,
+      "loss": 0.0289,
+      "step": 11153
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 3.9732779097387175e-05,
+      "loss": 0.0501,
+      "step": 11154
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 3.972684085510689e-05,
+      "loss": 0.1224,
+      "step": 11155
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 3.97209026128266e-05,
+      "loss": 0.1134,
+      "step": 11156
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 3.971496437054632e-05,
+      "loss": 0.1084,
+      "step": 11157
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 3.9709026128266036e-05,
+      "loss": 0.0993,
+      "step": 11158
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 3.970308788598575e-05,
+      "loss": 0.1689,
+      "step": 11159
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 3.9697149643705464e-05,
+      "loss": 0.0608,
+      "step": 11160
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 3.9691211401425184e-05,
+      "loss": 0.0861,
+      "step": 11161
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 3.968527315914489e-05,
+      "loss": 0.1009,
+      "step": 11162
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 3.967933491686461e-05,
+      "loss": 0.0659,
+      "step": 11163
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 3.9673396674584325e-05,
+      "loss": 0.0683,
+      "step": 11164
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 3.966745843230404e-05,
+      "loss": 0.0763,
+      "step": 11165
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 3.966152019002375e-05,
+      "loss": 0.0589,
+      "step": 11166
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 3.965558194774347e-05,
+      "loss": 0.052,
+      "step": 11167
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 3.9649643705463186e-05,
+      "loss": 0.055,
+      "step": 11168
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 3.96437054631829e-05,
+      "loss": 0.0861,
+      "step": 11169
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 3.963776722090261e-05,
+      "loss": 0.0711,
+      "step": 11170
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 3.963182897862233e-05,
+      "loss": 0.042,
+      "step": 11171
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 3.962589073634205e-05,
+      "loss": 0.0667,
+      "step": 11172
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 3.961995249406176e-05,
+      "loss": 0.0699,
+      "step": 11173
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 3.9614014251781474e-05,
+      "loss": 0.0906,
+      "step": 11174
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 3.960807600950119e-05,
+      "loss": 0.0488,
+      "step": 11175
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 3.96021377672209e-05,
+      "loss": 0.1238,
+      "step": 11176
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 3.959619952494062e-05,
+      "loss": 0.0384,
+      "step": 11177
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 3.9590261282660335e-05,
+      "loss": 0.0554,
+      "step": 11178
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 3.958432304038005e-05,
+      "loss": 0.5307,
+      "step": 11179
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 3.957838479809976e-05,
+      "loss": 0.0486,
+      "step": 11180
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 3.957244655581948e-05,
+      "loss": 0.0603,
+      "step": 11181
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 3.9566508313539196e-05,
+      "loss": 0.0716,
+      "step": 11182
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 3.956057007125891e-05,
+      "loss": 0.074,
+      "step": 11183
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 3.9554631828978623e-05,
+      "loss": 0.0712,
+      "step": 11184
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 3.954869358669834e-05,
+      "loss": 0.0824,
+      "step": 11185
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 3.954275534441805e-05,
+      "loss": 0.0313,
+      "step": 11186
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 3.953681710213777e-05,
+      "loss": 0.0753,
+      "step": 11187
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 3.9530878859857485e-05,
+      "loss": 0.1458,
+      "step": 11188
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 3.95249406175772e-05,
+      "loss": 0.0575,
+      "step": 11189
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 3.951900237529692e-05,
+      "loss": 0.0441,
+      "step": 11190
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 3.951306413301663e-05,
+      "loss": 0.0524,
+      "step": 11191
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 3.9507125890736346e-05,
+      "loss": 0.057,
+      "step": 11192
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 3.950118764845606e-05,
+      "loss": 0.0309,
+      "step": 11193
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 3.949524940617577e-05,
+      "loss": 0.0891,
+      "step": 11194
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 3.9489311163895486e-05,
+      "loss": 0.078,
+      "step": 11195
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 3.94833729216152e-05,
+      "loss": 0.086,
+      "step": 11196
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 3.947743467933492e-05,
+      "loss": 0.0433,
+      "step": 11197
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 3.9471496437054634e-05,
+      "loss": 0.0485,
+      "step": 11198
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 3.946555819477435e-05,
+      "loss": 0.0423,
+      "step": 11199
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 3.945961995249407e-05,
+      "loss": 0.0295,
+      "step": 11200
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 3.945368171021378e-05,
+      "loss": 0.0768,
+      "step": 11201
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 3.9447743467933495e-05,
+      "loss": 0.0324,
+      "step": 11202
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 3.944180522565321e-05,
+      "loss": 0.0536,
+      "step": 11203
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 3.943586698337292e-05,
+      "loss": 0.0647,
+      "step": 11204
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 3.9429928741092636e-05,
+      "loss": 0.1513,
+      "step": 11205
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 3.942399049881235e-05,
+      "loss": 0.1188,
+      "step": 11206
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 3.941805225653207e-05,
+      "loss": 0.0832,
+      "step": 11207
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 3.941211401425178e-05,
+      "loss": 0.0815,
+      "step": 11208
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 3.94061757719715e-05,
+      "loss": 0.0859,
+      "step": 11209
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 3.940023752969122e-05,
+      "loss": 0.0839,
+      "step": 11210
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 3.939429928741093e-05,
+      "loss": 0.0856,
+      "step": 11211
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 3.9388361045130644e-05,
+      "loss": 0.0946,
+      "step": 11212
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 3.938242280285036e-05,
+      "loss": 0.0478,
+      "step": 11213
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 3.937648456057007e-05,
+      "loss": 0.0741,
+      "step": 11214
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 3.9370546318289785e-05,
+      "loss": 0.0663,
+      "step": 11215
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 3.93646080760095e-05,
+      "loss": 0.0955,
+      "step": 11216
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 3.935866983372922e-05,
+      "loss": 0.062,
+      "step": 11217
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 3.935273159144893e-05,
+      "loss": 0.0867,
+      "step": 11218
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 3.9346793349168646e-05,
+      "loss": 0.0661,
+      "step": 11219
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 3.9340855106888367e-05,
+      "loss": 0.0496,
+      "step": 11220
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 3.933491686460808e-05,
+      "loss": 0.0747,
+      "step": 11221
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 3.9328978622327794e-05,
+      "loss": 0.0643,
+      "step": 11222
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 3.932304038004751e-05,
+      "loss": 0.0601,
+      "step": 11223
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 3.931710213776723e-05,
+      "loss": 0.0906,
+      "step": 11224
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 3.9311163895486934e-05,
+      "loss": 0.0835,
+      "step": 11225
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 3.930522565320665e-05,
+      "loss": 0.0825,
+      "step": 11226
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 3.929928741092637e-05,
+      "loss": 0.0642,
+      "step": 11227
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 3.929334916864608e-05,
+      "loss": 0.0691,
+      "step": 11228
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 3.9287410926365796e-05,
+      "loss": 0.0549,
+      "step": 11229
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 3.9281472684085516e-05,
+      "loss": 0.0409,
+      "step": 11230
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 3.927553444180523e-05,
+      "loss": 0.0791,
+      "step": 11231
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 3.926959619952494e-05,
+      "loss": 0.0516,
+      "step": 11232
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 3.926365795724466e-05,
+      "loss": 0.0424,
+      "step": 11233
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 3.925771971496438e-05,
+      "loss": 0.0532,
+      "step": 11234
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 3.9251781472684084e-05,
+      "loss": 0.067,
+      "step": 11235
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 3.92458432304038e-05,
+      "loss": 0.0601,
+      "step": 11236
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 3.923990498812352e-05,
+      "loss": 0.0609,
+      "step": 11237
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 3.923396674584323e-05,
+      "loss": 0.0869,
+      "step": 11238
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 3.9228028503562945e-05,
+      "loss": 0.0416,
+      "step": 11239
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 3.9222090261282665e-05,
+      "loss": 0.0439,
+      "step": 11240
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 3.921615201900238e-05,
+      "loss": 0.0468,
+      "step": 11241
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 3.921021377672209e-05,
+      "loss": 0.034,
+      "step": 11242
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 3.9204275534441806e-05,
+      "loss": 0.0235,
+      "step": 11243
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 3.9198337292161526e-05,
+      "loss": 0.0426,
+      "step": 11244
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 3.919239904988123e-05,
+      "loss": 0.0721,
+      "step": 11245
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 3.918646080760095e-05,
+      "loss": 0.0558,
+      "step": 11246
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 3.918052256532067e-05,
+      "loss": 0.0422,
+      "step": 11247
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 3.917458432304038e-05,
+      "loss": 0.0277,
+      "step": 11248
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 3.9168646080760094e-05,
+      "loss": 0.057,
+      "step": 11249
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 3.9162707838479815e-05,
+      "loss": 0.0378,
+      "step": 11250
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 3.915676959619953e-05,
+      "loss": 0.0212,
+      "step": 11251
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 3.915083135391924e-05,
+      "loss": 0.0252,
+      "step": 11252
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 3.9144893111638955e-05,
+      "loss": 0.0436,
+      "step": 11253
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 3.9138954869358676e-05,
+      "loss": 0.0332,
+      "step": 11254
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 3.913301662707839e-05,
+      "loss": 0.1462,
+      "step": 11255
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 3.9127078384798096e-05,
+      "loss": 0.0865,
+      "step": 11256
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 3.9121140142517816e-05,
+      "loss": 0.0925,
+      "step": 11257
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 3.911520190023753e-05,
+      "loss": 0.1135,
+      "step": 11258
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 3.9109263657957244e-05,
+      "loss": 0.0633,
+      "step": 11259
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 3.9103325415676964e-05,
+      "loss": 0.0566,
+      "step": 11260
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 3.909738717339668e-05,
+      "loss": 0.0713,
+      "step": 11261
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 3.909144893111639e-05,
+      "loss": 0.063,
+      "step": 11262
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 3.9085510688836105e-05,
+      "loss": 0.0784,
+      "step": 11263
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 3.9079572446555825e-05,
+      "loss": 0.067,
+      "step": 11264
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 3.907363420427554e-05,
+      "loss": 0.0718,
+      "step": 11265
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 3.9067695961995246e-05,
+      "loss": 0.0703,
+      "step": 11266
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 3.9061757719714966e-05,
+      "loss": 0.0623,
+      "step": 11267
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 3.905581947743468e-05,
+      "loss": 0.0548,
+      "step": 11268
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 3.904988123515439e-05,
+      "loss": 0.0594,
+      "step": 11269
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 3.904394299287411e-05,
+      "loss": 0.0486,
+      "step": 11270
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 3.903800475059383e-05,
+      "loss": 0.0665,
+      "step": 11271
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 3.903206650831354e-05,
+      "loss": 0.0544,
+      "step": 11272
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 3.9026128266033254e-05,
+      "loss": 0.0721,
+      "step": 11273
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 3.9020190023752974e-05,
+      "loss": 0.0648,
+      "step": 11274
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 3.901425178147269e-05,
+      "loss": 0.0918,
+      "step": 11275
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 3.9008313539192395e-05,
+      "loss": 0.0848,
+      "step": 11276
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 3.9002375296912115e-05,
+      "loss": 0.0517,
+      "step": 11277
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 3.899643705463183e-05,
+      "loss": 0.0781,
+      "step": 11278
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 3.899049881235154e-05,
+      "loss": 0.0608,
+      "step": 11279
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 3.898456057007126e-05,
+      "loss": 0.1875,
+      "step": 11280
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 3.8978622327790976e-05,
+      "loss": 0.0546,
+      "step": 11281
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 3.897268408551069e-05,
+      "loss": 0.0341,
+      "step": 11282
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 3.8966745843230403e-05,
+      "loss": 0.074,
+      "step": 11283
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 3.8960807600950124e-05,
+      "loss": 0.128,
+      "step": 11284
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 3.895486935866984e-05,
+      "loss": 0.0412,
+      "step": 11285
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 3.894893111638955e-05,
+      "loss": 0.0768,
+      "step": 11286
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 3.8942992874109265e-05,
+      "loss": 0.0604,
+      "step": 11287
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 3.893705463182898e-05,
+      "loss": 0.0544,
+      "step": 11288
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 3.893111638954869e-05,
+      "loss": 0.0476,
+      "step": 11289
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 3.892517814726841e-05,
+      "loss": 0.0551,
+      "step": 11290
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 3.8919239904988126e-05,
+      "loss": 0.0471,
+      "step": 11291
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 3.891330166270784e-05,
+      "loss": 0.0655,
+      "step": 11292
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 3.890736342042755e-05,
+      "loss": 0.0736,
+      "step": 11293
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 3.890142517814727e-05,
+      "loss": 0.0493,
+      "step": 11294
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 3.889548693586699e-05,
+      "loss": 0.0774,
+      "step": 11295
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 3.88895486935867e-05,
+      "loss": 0.08,
+      "step": 11296
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 3.8883610451306414e-05,
+      "loss": 0.0513,
+      "step": 11297
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 3.887767220902613e-05,
+      "loss": 0.0697,
+      "step": 11298
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 3.887173396674584e-05,
+      "loss": 0.0721,
+      "step": 11299
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 3.886579572446556e-05,
+      "loss": 0.0661,
+      "step": 11300
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 3.8859857482185275e-05,
+      "loss": 0.1294,
+      "step": 11301
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 3.885391923990499e-05,
+      "loss": 0.0484,
+      "step": 11302
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 3.88479809976247e-05,
+      "loss": 0.0534,
+      "step": 11303
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 3.884204275534442e-05,
+      "loss": 0.0558,
+      "step": 11304
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 3.8836104513064136e-05,
+      "loss": 0.1149,
+      "step": 11305
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 3.883016627078385e-05,
+      "loss": 0.1145,
+      "step": 11306
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 3.882422802850357e-05,
+      "loss": 0.1091,
+      "step": 11307
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 3.881828978622328e-05,
+      "loss": 0.0851,
+      "step": 11308
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 3.881235154394299e-05,
+      "loss": 0.1052,
+      "step": 11309
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 3.880641330166271e-05,
+      "loss": 0.0642,
+      "step": 11310
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 3.8800475059382424e-05,
+      "loss": 0.0816,
+      "step": 11311
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 3.879453681710214e-05,
+      "loss": 0.0895,
+      "step": 11312
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 3.878859857482186e-05,
+      "loss": 0.0758,
+      "step": 11313
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 3.878266033254157e-05,
+      "loss": 0.0884,
+      "step": 11314
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 3.8776722090261285e-05,
+      "loss": 0.0463,
+      "step": 11315
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 3.8770783847981e-05,
+      "loss": 0.0882,
+      "step": 11316
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 3.876484560570072e-05,
+      "loss": 0.0609,
+      "step": 11317
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 3.8758907363420426e-05,
+      "loss": 0.0536,
+      "step": 11318
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 3.875296912114014e-05,
+      "loss": 0.043,
+      "step": 11319
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 3.874703087885986e-05,
+      "loss": 0.0667,
+      "step": 11320
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 3.8741092636579574e-05,
+      "loss": 0.1212,
+      "step": 11321
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 3.873515439429929e-05,
+      "loss": 0.0746,
+      "step": 11322
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 3.872921615201901e-05,
+      "loss": 0.0861,
+      "step": 11323
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 3.872327790973872e-05,
+      "loss": 0.0734,
+      "step": 11324
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 3.8717339667458435e-05,
+      "loss": 0.0595,
+      "step": 11325
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 3.871140142517815e-05,
+      "loss": 0.0859,
+      "step": 11326
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 3.870546318289787e-05,
+      "loss": 0.0753,
+      "step": 11327
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 3.8699524940617576e-05,
+      "loss": 0.078,
+      "step": 11328
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 3.869358669833729e-05,
+      "loss": 0.0609,
+      "step": 11329
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 3.868764845605701e-05,
+      "loss": 0.0799,
+      "step": 11330
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 3.868171021377672e-05,
+      "loss": 0.0477,
+      "step": 11331
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 3.867577197149644e-05,
+      "loss": 0.0774,
+      "step": 11332
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 3.866983372921616e-05,
+      "loss": 0.0412,
+      "step": 11333
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 3.866389548693587e-05,
+      "loss": 0.049,
+      "step": 11334
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 3.8657957244655584e-05,
+      "loss": 0.06,
+      "step": 11335
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 3.86520190023753e-05,
+      "loss": 0.0923,
+      "step": 11336
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 3.864608076009502e-05,
+      "loss": 0.0565,
+      "step": 11337
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 3.864014251781473e-05,
+      "loss": 0.0887,
+      "step": 11338
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 3.863420427553444e-05,
+      "loss": 0.0866,
+      "step": 11339
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 3.862826603325416e-05,
+      "loss": 0.0577,
+      "step": 11340
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 3.862232779097387e-05,
+      "loss": 0.0409,
+      "step": 11341
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 3.8616389548693586e-05,
+      "loss": 0.0923,
+      "step": 11342
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 3.8610451306413306e-05,
+      "loss": 0.0516,
+      "step": 11343
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 3.860451306413302e-05,
+      "loss": 0.0684,
+      "step": 11344
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 3.8598574821852734e-05,
+      "loss": 0.096,
+      "step": 11345
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 3.859263657957245e-05,
+      "loss": 0.0631,
+      "step": 11346
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 3.858669833729217e-05,
+      "loss": 0.0425,
+      "step": 11347
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 3.858076009501188e-05,
+      "loss": 0.0638,
+      "step": 11348
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 3.857482185273159e-05,
+      "loss": 0.0619,
+      "step": 11349
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 3.856888361045131e-05,
+      "loss": 0.0467,
+      "step": 11350
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 3.856294536817102e-05,
+      "loss": 0.0459,
+      "step": 11351
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 3.8557007125890735e-05,
+      "loss": 0.0137,
+      "step": 11352
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 3.8551068883610456e-05,
+      "loss": 0.0529,
+      "step": 11353
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 3.854513064133017e-05,
+      "loss": 0.0785,
+      "step": 11354
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 3.853919239904988e-05,
+      "loss": 0.1501,
+      "step": 11355
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 3.8533254156769597e-05,
+      "loss": 0.0849,
+      "step": 11356
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 3.852731591448932e-05,
+      "loss": 0.1248,
+      "step": 11357
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 3.852137767220903e-05,
+      "loss": 0.1013,
+      "step": 11358
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 3.851543942992874e-05,
+      "loss": 0.1082,
+      "step": 11359
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 3.850950118764846e-05,
+      "loss": 0.0639,
+      "step": 11360
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 3.850356294536817e-05,
+      "loss": 0.1071,
+      "step": 11361
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 3.8497624703087885e-05,
+      "loss": 0.0637,
+      "step": 11362
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 3.8491686460807605e-05,
+      "loss": 0.0944,
+      "step": 11363
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 3.848574821852732e-05,
+      "loss": 0.0889,
+      "step": 11364
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 3.847980997624703e-05,
+      "loss": 0.0579,
+      "step": 11365
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 3.8473871733966746e-05,
+      "loss": 0.0793,
+      "step": 11366
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 3.8467933491686466e-05,
+      "loss": 0.0765,
+      "step": 11367
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 3.846199524940618e-05,
+      "loss": 0.0614,
+      "step": 11368
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 3.845605700712589e-05,
+      "loss": 0.0527,
+      "step": 11369
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 3.845011876484561e-05,
+      "loss": 0.0714,
+      "step": 11370
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 3.844418052256532e-05,
+      "loss": 0.0706,
+      "step": 11371
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 3.8438242280285034e-05,
+      "loss": 0.0552,
+      "step": 11372
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 3.8432304038004754e-05,
+      "loss": 0.0767,
+      "step": 11373
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 3.842636579572447e-05,
+      "loss": 0.078,
+      "step": 11374
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 3.842042755344418e-05,
+      "loss": 0.1315,
+      "step": 11375
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 3.8414489311163895e-05,
+      "loss": 0.0621,
+      "step": 11376
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 3.8408551068883616e-05,
+      "loss": 0.0877,
+      "step": 11377
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 3.840261282660333e-05,
+      "loss": 0.0715,
+      "step": 11378
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 3.839667458432304e-05,
+      "loss": 0.0764,
+      "step": 11379
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 3.8390736342042756e-05,
+      "loss": 0.1467,
+      "step": 11380
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 3.838479809976247e-05,
+      "loss": 0.0539,
+      "step": 11381
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 3.8378859857482183e-05,
+      "loss": 0.0504,
+      "step": 11382
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 3.8372921615201904e-05,
+      "loss": 0.0481,
+      "step": 11383
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 3.836698337292162e-05,
+      "loss": 0.0468,
+      "step": 11384
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 3.836104513064133e-05,
+      "loss": 0.0422,
+      "step": 11385
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 3.8355106888361045e-05,
+      "loss": 0.0448,
+      "step": 11386
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 3.8349168646080765e-05,
+      "loss": 0.0691,
+      "step": 11387
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 3.834323040380048e-05,
+      "loss": 0.0311,
+      "step": 11388
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 3.833729216152019e-05,
+      "loss": 0.0365,
+      "step": 11389
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 3.833135391923991e-05,
+      "loss": 0.0691,
+      "step": 11390
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 3.832541567695962e-05,
+      "loss": 0.0465,
+      "step": 11391
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 3.831947743467933e-05,
+      "loss": 0.0644,
+      "step": 11392
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 3.831353919239905e-05,
+      "loss": 0.087,
+      "step": 11393
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 3.830760095011877e-05,
+      "loss": 0.0998,
+      "step": 11394
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 3.830166270783848e-05,
+      "loss": 0.0657,
+      "step": 11395
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 3.8295724465558194e-05,
+      "loss": 0.0889,
+      "step": 11396
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 3.8289786223277914e-05,
+      "loss": 0.0411,
+      "step": 11397
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 3.828384798099763e-05,
+      "loss": 0.0423,
+      "step": 11398
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 3.827790973871734e-05,
+      "loss": 0.0363,
+      "step": 11399
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 3.827197149643706e-05,
+      "loss": 0.0548,
+      "step": 11400
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 3.826603325415677e-05,
+      "loss": 0.0322,
+      "step": 11401
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 3.826009501187648e-05,
+      "loss": 0.0404,
+      "step": 11402
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 3.82541567695962e-05,
+      "loss": 0.1089,
+      "step": 11403
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 3.8248218527315916e-05,
+      "loss": 0.043,
+      "step": 11404
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 3.824228028503563e-05,
+      "loss": 0.0967,
+      "step": 11405
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 3.823634204275534e-05,
+      "loss": 0.1328,
+      "step": 11406
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 3.8230403800475064e-05,
+      "loss": 0.1213,
+      "step": 11407
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 3.822446555819478e-05,
+      "loss": 0.088,
+      "step": 11408
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 3.821852731591449e-05,
+      "loss": 0.1192,
+      "step": 11409
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 3.821258907363421e-05,
+      "loss": 0.0516,
+      "step": 11410
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 3.820665083135392e-05,
+      "loss": 0.1116,
+      "step": 11411
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 3.820071258907363e-05,
+      "loss": 0.0679,
+      "step": 11412
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 3.819477434679335e-05,
+      "loss": 0.064,
+      "step": 11413
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 3.8188836104513066e-05,
+      "loss": 0.1062,
+      "step": 11414
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 3.818289786223278e-05,
+      "loss": 0.048,
+      "step": 11415
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 3.817695961995249e-05,
+      "loss": 0.0769,
+      "step": 11416
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 3.817102137767221e-05,
+      "loss": 0.0829,
+      "step": 11417
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 3.8165083135391927e-05,
+      "loss": 0.0555,
+      "step": 11418
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 3.815914489311164e-05,
+      "loss": 0.0833,
+      "step": 11419
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 3.815320665083136e-05,
+      "loss": 0.0845,
+      "step": 11420
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 3.8147268408551074e-05,
+      "loss": 0.0559,
+      "step": 11421
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 3.814133016627078e-05,
+      "loss": 0.0806,
+      "step": 11422
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 3.81353919239905e-05,
+      "loss": 0.0947,
+      "step": 11423
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 3.8129453681710215e-05,
+      "loss": 0.0577,
+      "step": 11424
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 3.812351543942993e-05,
+      "loss": 0.1064,
+      "step": 11425
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 3.811757719714964e-05,
+      "loss": 0.0892,
+      "step": 11426
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 3.811163895486936e-05,
+      "loss": 0.0916,
+      "step": 11427
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 3.8105700712589076e-05,
+      "loss": 0.0624,
+      "step": 11428
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 3.809976247030879e-05,
+      "loss": 0.0746,
+      "step": 11429
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 3.809382422802851e-05,
+      "loss": 0.0509,
+      "step": 11430
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 3.8087885985748223e-05,
+      "loss": 0.0371,
+      "step": 11431
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 3.808194774346793e-05,
+      "loss": 0.0651,
+      "step": 11432
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 3.807600950118765e-05,
+      "loss": 0.0561,
+      "step": 11433
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 3.8070071258907364e-05,
+      "loss": 0.0567,
+      "step": 11434
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 3.806413301662708e-05,
+      "loss": 0.0811,
+      "step": 11435
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 3.80581947743468e-05,
+      "loss": 0.0826,
+      "step": 11436
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 3.805225653206651e-05,
+      "loss": 0.0512,
+      "step": 11437
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 3.8046318289786225e-05,
+      "loss": 0.0509,
+      "step": 11438
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 3.804038004750594e-05,
+      "loss": 0.0842,
+      "step": 11439
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 3.803444180522566e-05,
+      "loss": 0.0643,
+      "step": 11440
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 3.802850356294537e-05,
+      "loss": 0.0455,
+      "step": 11441
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 3.802256532066508e-05,
+      "loss": 0.0524,
+      "step": 11442
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 3.80166270783848e-05,
+      "loss": 0.0423,
+      "step": 11443
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 3.8010688836104514e-05,
+      "loss": 0.085,
+      "step": 11444
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 3.800475059382423e-05,
+      "loss": 0.0425,
+      "step": 11445
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 3.799881235154395e-05,
+      "loss": 0.0288,
+      "step": 11446
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 3.799287410926366e-05,
+      "loss": 0.0566,
+      "step": 11447
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 3.7986935866983375e-05,
+      "loss": 0.061,
+      "step": 11448
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 3.798099762470309e-05,
+      "loss": 0.1008,
+      "step": 11449
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 3.797505938242281e-05,
+      "loss": 0.0388,
+      "step": 11450
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 3.796912114014252e-05,
+      "loss": 0.0444,
+      "step": 11451
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 3.7963182897862236e-05,
+      "loss": 0.0545,
+      "step": 11452
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 3.795724465558195e-05,
+      "loss": 0.0256,
+      "step": 11453
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 3.795130641330166e-05,
+      "loss": 0.0686,
+      "step": 11454
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 3.7945368171021377e-05,
+      "loss": 0.1269,
+      "step": 11455
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 3.79394299287411e-05,
+      "loss": 0.185,
+      "step": 11456
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 3.793349168646081e-05,
+      "loss": 0.1017,
+      "step": 11457
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 3.7927553444180524e-05,
+      "loss": 0.0945,
+      "step": 11458
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 3.792161520190024e-05,
+      "loss": 0.1022,
+      "step": 11459
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 3.791567695961996e-05,
+      "loss": 0.0917,
+      "step": 11460
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 3.790973871733967e-05,
+      "loss": 0.0661,
+      "step": 11461
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 3.7903800475059385e-05,
+      "loss": 0.0621,
+      "step": 11462
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 3.78978622327791e-05,
+      "loss": 0.0974,
+      "step": 11463
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 3.789192399049881e-05,
+      "loss": 0.0966,
+      "step": 11464
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 3.7885985748218526e-05,
+      "loss": 0.0721,
+      "step": 11465
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 3.7880047505938246e-05,
+      "loss": 0.0759,
+      "step": 11466
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 3.787410926365796e-05,
+      "loss": 0.0621,
+      "step": 11467
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 3.7868171021377673e-05,
+      "loss": 0.0562,
+      "step": 11468
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 3.786223277909739e-05,
+      "loss": 0.0657,
+      "step": 11469
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 3.785629453681711e-05,
+      "loss": 0.0687,
+      "step": 11470
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 3.785035629453682e-05,
+      "loss": 0.047,
+      "step": 11471
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 3.7844418052256534e-05,
+      "loss": 0.0789,
+      "step": 11472
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 3.783847980997625e-05,
+      "loss": 0.0769,
+      "step": 11473
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 3.783254156769596e-05,
+      "loss": 0.0809,
+      "step": 11474
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 3.7826603325415675e-05,
+      "loss": 0.0803,
+      "step": 11475
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 3.7820665083135396e-05,
+      "loss": 0.0679,
+      "step": 11476
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 3.781472684085511e-05,
+      "loss": 0.0639,
+      "step": 11477
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 3.780878859857482e-05,
+      "loss": 0.0782,
+      "step": 11478
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 3.7802850356294536e-05,
+      "loss": 0.0389,
+      "step": 11479
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 3.779691211401426e-05,
+      "loss": 0.0527,
+      "step": 11480
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 3.779097387173397e-05,
+      "loss": 0.1415,
+      "step": 11481
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 3.7785035629453684e-05,
+      "loss": 0.0478,
+      "step": 11482
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 3.7779097387173404e-05,
+      "loss": 0.0352,
+      "step": 11483
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 3.777315914489311e-05,
+      "loss": 0.0571,
+      "step": 11484
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 3.7767220902612825e-05,
+      "loss": 0.0875,
+      "step": 11485
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 3.7761282660332545e-05,
+      "loss": 0.0477,
+      "step": 11486
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 3.775534441805226e-05,
+      "loss": 0.0601,
+      "step": 11487
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 3.774940617577197e-05,
+      "loss": 0.0742,
+      "step": 11488
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 3.7743467933491686e-05,
+      "loss": 0.0567,
+      "step": 11489
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 3.7737529691211406e-05,
+      "loss": 0.0433,
+      "step": 11490
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 3.773159144893112e-05,
+      "loss": 0.0463,
+      "step": 11491
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 3.772565320665083e-05,
+      "loss": 0.1185,
+      "step": 11492
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 3.7719714964370554e-05,
+      "loss": 0.1335,
+      "step": 11493
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 3.771377672209026e-05,
+      "loss": 0.049,
+      "step": 11494
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 3.7707838479809974e-05,
+      "loss": 0.0392,
+      "step": 11495
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 3.7701900237529694e-05,
+      "loss": 0.0222,
+      "step": 11496
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 3.769596199524941e-05,
+      "loss": 0.0358,
+      "step": 11497
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 3.769002375296912e-05,
+      "loss": 0.0292,
+      "step": 11498
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 3.7684085510688835e-05,
+      "loss": 0.0387,
+      "step": 11499
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 3.7678147268408555e-05,
+      "loss": 0.0758,
+      "step": 11500
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 3.767220902612827e-05,
+      "loss": 0.1128,
+      "step": 11501
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 3.766627078384798e-05,
+      "loss": 0.0964,
+      "step": 11502
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 3.76603325415677e-05,
+      "loss": 0.1029,
+      "step": 11503
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 3.7654394299287417e-05,
+      "loss": 0.0996,
+      "step": 11504
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 3.764845605700712e-05,
+      "loss": 0.0995,
+      "step": 11505
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 3.7642517814726844e-05,
+      "loss": 0.1095,
+      "step": 11506
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 3.763657957244656e-05,
+      "loss": 0.1309,
+      "step": 11507
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 3.763064133016627e-05,
+      "loss": 0.1344,
+      "step": 11508
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 3.7624703087885984e-05,
+      "loss": 0.1015,
+      "step": 11509
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 3.7618764845605705e-05,
+      "loss": 0.1127,
+      "step": 11510
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 3.761282660332542e-05,
+      "loss": 0.0654,
+      "step": 11511
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 3.760688836104513e-05,
+      "loss": 0.068,
+      "step": 11512
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 3.760095011876485e-05,
+      "loss": 0.0794,
+      "step": 11513
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 3.7595011876484566e-05,
+      "loss": 0.0894,
+      "step": 11514
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 3.758907363420427e-05,
+      "loss": 0.0618,
+      "step": 11515
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 3.758313539192399e-05,
+      "loss": 0.1,
+      "step": 11516
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 3.757719714964371e-05,
+      "loss": 0.1918,
+      "step": 11517
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 3.757125890736342e-05,
+      "loss": 0.0721,
+      "step": 11518
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 3.7565320665083134e-05,
+      "loss": 0.092,
+      "step": 11519
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 3.7559382422802854e-05,
+      "loss": 0.0587,
+      "step": 11520
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 3.755344418052257e-05,
+      "loss": 0.056,
+      "step": 11521
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 3.754750593824228e-05,
+      "loss": 0.0826,
+      "step": 11522
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 3.7541567695962e-05,
+      "loss": 0.0557,
+      "step": 11523
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 3.7535629453681715e-05,
+      "loss": 0.0868,
+      "step": 11524
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 3.752969121140142e-05,
+      "loss": 0.1033,
+      "step": 11525
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 3.752375296912114e-05,
+      "loss": 0.0802,
+      "step": 11526
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 3.7517814726840856e-05,
+      "loss": 0.0597,
+      "step": 11527
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 3.751187648456057e-05,
+      "loss": 0.0493,
+      "step": 11528
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 3.750593824228028e-05,
+      "loss": 0.075,
+      "step": 11529
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.1078,
+      "step": 11530
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 3.749406175771972e-05,
+      "loss": 0.0765,
+      "step": 11531
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 3.748812351543943e-05,
+      "loss": 0.0681,
+      "step": 11532
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 3.748218527315915e-05,
+      "loss": 0.0874,
+      "step": 11533
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 3.7476247030878865e-05,
+      "loss": 0.1653,
+      "step": 11534
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 3.747030878859858e-05,
+      "loss": 0.0713,
+      "step": 11535
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 3.746437054631829e-05,
+      "loss": 0.0314,
+      "step": 11536
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 3.7458432304038005e-05,
+      "loss": 0.094,
+      "step": 11537
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 3.745249406175772e-05,
+      "loss": 0.0424,
+      "step": 11538
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 3.744655581947743e-05,
+      "loss": 0.0574,
+      "step": 11539
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 3.744061757719715e-05,
+      "loss": 0.0457,
+      "step": 11540
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 3.7434679334916866e-05,
+      "loss": 0.0753,
+      "step": 11541
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 3.742874109263658e-05,
+      "loss": 0.0347,
+      "step": 11542
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 3.74228028503563e-05,
+      "loss": 0.048,
+      "step": 11543
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 3.7416864608076014e-05,
+      "loss": 0.0438,
+      "step": 11544
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 3.741092636579573e-05,
+      "loss": 0.0656,
+      "step": 11545
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 3.740498812351544e-05,
+      "loss": 0.0457,
+      "step": 11546
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 3.7399049881235155e-05,
+      "loss": 0.0875,
+      "step": 11547
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 3.739311163895487e-05,
+      "loss": 0.1507,
+      "step": 11548
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 3.738717339667458e-05,
+      "loss": 0.0747,
+      "step": 11549
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 3.73812351543943e-05,
+      "loss": 0.0732,
+      "step": 11550
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 3.7375296912114016e-05,
+      "loss": 0.0693,
+      "step": 11551
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 3.736935866983373e-05,
+      "loss": 0.0507,
+      "step": 11552
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 3.736342042755345e-05,
+      "loss": 0.0111,
+      "step": 11553
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 3.735748218527316e-05,
+      "loss": 0.101,
+      "step": 11554
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 3.735154394299288e-05,
+      "loss": 0.1388,
+      "step": 11555
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 3.734560570071259e-05,
+      "loss": 0.0782,
+      "step": 11556
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 3.7339667458432304e-05,
+      "loss": 0.1397,
+      "step": 11557
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 3.733372921615202e-05,
+      "loss": 0.0597,
+      "step": 11558
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 3.732779097387174e-05,
+      "loss": 0.0901,
+      "step": 11559
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 3.732185273159145e-05,
+      "loss": 0.0438,
+      "step": 11560
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 3.7315914489311165e-05,
+      "loss": 0.0609,
+      "step": 11561
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 3.730997624703088e-05,
+      "loss": 0.0653,
+      "step": 11562
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 3.73040380047506e-05,
+      "loss": 0.0905,
+      "step": 11563
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 3.729809976247031e-05,
+      "loss": 0.0625,
+      "step": 11564
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 3.7292161520190026e-05,
+      "loss": 0.0451,
+      "step": 11565
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 3.728622327790974e-05,
+      "loss": 0.0614,
+      "step": 11566
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 3.7280285035629453e-05,
+      "loss": 0.0942,
+      "step": 11567
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 3.727434679334917e-05,
+      "loss": 0.0728,
+      "step": 11568
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 3.726840855106889e-05,
+      "loss": 0.0768,
+      "step": 11569
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 3.72624703087886e-05,
+      "loss": 0.0668,
+      "step": 11570
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 3.7256532066508315e-05,
+      "loss": 0.0891,
+      "step": 11571
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 3.725059382422803e-05,
+      "loss": 0.0566,
+      "step": 11572
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 3.724465558194775e-05,
+      "loss": 0.0846,
+      "step": 11573
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 3.723871733966746e-05,
+      "loss": 0.0972,
+      "step": 11574
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 3.7232779097387176e-05,
+      "loss": 0.0699,
+      "step": 11575
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 3.722684085510689e-05,
+      "loss": 0.0518,
+      "step": 11576
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 3.72209026128266e-05,
+      "loss": 0.0551,
+      "step": 11577
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 3.7214964370546316e-05,
+      "loss": 0.06,
+      "step": 11578
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 3.720902612826604e-05,
+      "loss": 0.0512,
+      "step": 11579
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 3.720308788598575e-05,
+      "loss": 0.0591,
+      "step": 11580
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 3.7197149643705464e-05,
+      "loss": 0.0432,
+      "step": 11581
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 3.719121140142518e-05,
+      "loss": 0.0613,
+      "step": 11582
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 3.71852731591449e-05,
+      "loss": 0.0386,
+      "step": 11583
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 3.717933491686461e-05,
+      "loss": 0.042,
+      "step": 11584
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 3.7173396674584325e-05,
+      "loss": 0.0445,
+      "step": 11585
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 3.7167458432304045e-05,
+      "loss": 0.0349,
+      "step": 11586
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 3.716152019002376e-05,
+      "loss": 0.0576,
+      "step": 11587
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 3.7155581947743466e-05,
+      "loss": 0.0774,
+      "step": 11588
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 3.7149643705463186e-05,
+      "loss": 0.0473,
+      "step": 11589
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 3.71437054631829e-05,
+      "loss": 0.0475,
+      "step": 11590
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 3.713776722090261e-05,
+      "loss": 0.1946,
+      "step": 11591
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 3.713182897862233e-05,
+      "loss": 0.0292,
+      "step": 11592
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 3.712589073634205e-05,
+      "loss": 0.048,
+      "step": 11593
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 3.711995249406176e-05,
+      "loss": 0.0139,
+      "step": 11594
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 3.7114014251781474e-05,
+      "loss": 0.1749,
+      "step": 11595
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 3.7108076009501195e-05,
+      "loss": 0.0642,
+      "step": 11596
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 3.710213776722091e-05,
+      "loss": 0.0849,
+      "step": 11597
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 3.7096199524940615e-05,
+      "loss": 0.1176,
+      "step": 11598
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 3.7090261282660335e-05,
+      "loss": 0.0599,
+      "step": 11599
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 3.708432304038005e-05,
+      "loss": 0.0811,
+      "step": 11600
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 3.707838479809976e-05,
+      "loss": 0.0522,
+      "step": 11601
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 3.7072446555819476e-05,
+      "loss": 0.0486,
+      "step": 11602
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 3.7066508313539197e-05,
+      "loss": 0.0457,
+      "step": 11603
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 3.706057007125891e-05,
+      "loss": 0.0441,
+      "step": 11604
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 3.7054631828978624e-05,
+      "loss": 0.05,
+      "step": 11605
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 3.7048693586698344e-05,
+      "loss": 0.0544,
+      "step": 11606
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 3.704275534441806e-05,
+      "loss": 0.0416,
+      "step": 11607
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 3.7036817102137764e-05,
+      "loss": 0.0527,
+      "step": 11608
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 3.7030878859857485e-05,
+      "loss": 0.0876,
+      "step": 11609
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 3.70249406175772e-05,
+      "loss": 0.0532,
+      "step": 11610
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 3.701900237529691e-05,
+      "loss": 0.0438,
+      "step": 11611
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 3.7013064133016626e-05,
+      "loss": 0.0699,
+      "step": 11612
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 3.7007125890736346e-05,
+      "loss": 0.0547,
+      "step": 11613
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 3.700118764845606e-05,
+      "loss": 0.0881,
+      "step": 11614
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 3.699524940617577e-05,
+      "loss": 0.0627,
+      "step": 11615
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 3.6989311163895493e-05,
+      "loss": 0.0466,
+      "step": 11616
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 3.698337292161521e-05,
+      "loss": 0.0715,
+      "step": 11617
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 3.697743467933492e-05,
+      "loss": 0.0511,
+      "step": 11618
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 3.6971496437054634e-05,
+      "loss": 0.124,
+      "step": 11619
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 3.696555819477435e-05,
+      "loss": 0.1064,
+      "step": 11620
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 3.695961995249406e-05,
+      "loss": 0.0609,
+      "step": 11621
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 3.6953681710213775e-05,
+      "loss": 0.0894,
+      "step": 11622
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 3.6947743467933495e-05,
+      "loss": 0.0484,
+      "step": 11623
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 3.694180522565321e-05,
+      "loss": 0.0816,
+      "step": 11624
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 3.693586698337292e-05,
+      "loss": 0.0406,
+      "step": 11625
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 3.692992874109264e-05,
+      "loss": 0.0492,
+      "step": 11626
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 3.6923990498812356e-05,
+      "loss": 0.0676,
+      "step": 11627
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 3.691805225653207e-05,
+      "loss": 0.0602,
+      "step": 11628
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 3.6912114014251784e-05,
+      "loss": 0.0422,
+      "step": 11629
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 3.69061757719715e-05,
+      "loss": 0.0363,
+      "step": 11630
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 3.690023752969121e-05,
+      "loss": 0.0351,
+      "step": 11631
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 3.6894299287410924e-05,
+      "loss": 0.0568,
+      "step": 11632
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 3.6888361045130645e-05,
+      "loss": 0.0469,
+      "step": 11633
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 3.688242280285036e-05,
+      "loss": 0.0447,
+      "step": 11634
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 3.687648456057007e-05,
+      "loss": 0.0532,
+      "step": 11635
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 3.687054631828979e-05,
+      "loss": 0.0362,
+      "step": 11636
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 3.6864608076009506e-05,
+      "loss": 0.0508,
+      "step": 11637
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 3.685866983372922e-05,
+      "loss": 0.0355,
+      "step": 11638
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 3.685273159144893e-05,
+      "loss": 0.0396,
+      "step": 11639
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 3.6846793349168646e-05,
+      "loss": 0.0448,
+      "step": 11640
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 3.684085510688836e-05,
+      "loss": 0.0303,
+      "step": 11641
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 3.6834916864608074e-05,
+      "loss": 0.0505,
+      "step": 11642
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 3.6828978622327794e-05,
+      "loss": 0.0484,
+      "step": 11643
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 3.682304038004751e-05,
+      "loss": 0.2564,
+      "step": 11644
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 3.681710213776722e-05,
+      "loss": 0.0069,
+      "step": 11645
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.681116389548694e-05,
+      "loss": 0.0194,
+      "step": 11646
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.6805225653206655e-05,
+      "loss": 0.1072,
+      "step": 11647
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.679928741092637e-05,
+      "loss": 0.106,
+      "step": 11648
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.679334916864608e-05,
+      "loss": 0.0912,
+      "step": 11649
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.6787410926365796e-05,
+      "loss": 0.0612,
+      "step": 11650
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.678147268408551e-05,
+      "loss": 0.0639,
+      "step": 11651
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.677553444180522e-05,
+      "loss": 0.0849,
+      "step": 11652
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.676959619952494e-05,
+      "loss": 0.0545,
+      "step": 11653
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 3.676365795724466e-05,
+      "loss": 0.0321,
+      "step": 11654
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 3.675771971496437e-05,
+      "loss": 0.0501,
+      "step": 11655
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 3.675178147268409e-05,
+      "loss": 0.0575,
+      "step": 11656
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 3.6745843230403804e-05,
+      "loss": 0.0724,
+      "step": 11657
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 3.673990498812352e-05,
+      "loss": 0.0668,
+      "step": 11658
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 3.673396674584323e-05,
+      "loss": 0.0454,
+      "step": 11659
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 3.6728028503562945e-05,
+      "loss": 0.0593,
+      "step": 11660
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 3.672209026128266e-05,
+      "loss": 0.0437,
+      "step": 11661
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 3.671615201900237e-05,
+      "loss": 0.0431,
+      "step": 11662
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 3.671021377672209e-05,
+      "loss": 0.048,
+      "step": 11663
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 3.6704275534441806e-05,
+      "loss": 0.0521,
+      "step": 11664
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 3.669833729216152e-05,
+      "loss": 0.0589,
+      "step": 11665
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 3.669239904988124e-05,
+      "loss": 0.0534,
+      "step": 11666
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 3.6686460807600954e-05,
+      "loss": 0.0534,
+      "step": 11667
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 3.668052256532067e-05,
+      "loss": 0.0457,
+      "step": 11668
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 3.667458432304038e-05,
+      "loss": 0.0743,
+      "step": 11669
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 3.6668646080760095e-05,
+      "loss": 0.0683,
+      "step": 11670
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 3.666270783847981e-05,
+      "loss": 0.0676,
+      "step": 11671
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 3.665676959619952e-05,
+      "loss": 0.0756,
+      "step": 11672
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 3.665083135391924e-05,
+      "loss": 0.0587,
+      "step": 11673
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 3.6644893111638956e-05,
+      "loss": 0.0391,
+      "step": 11674
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 3.663895486935867e-05,
+      "loss": 0.0478,
+      "step": 11675
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 3.663301662707839e-05,
+      "loss": 0.0508,
+      "step": 11676
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 3.66270783847981e-05,
+      "loss": 0.1276,
+      "step": 11677
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 3.662114014251782e-05,
+      "loss": 0.0479,
+      "step": 11678
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 3.661520190023753e-05,
+      "loss": 0.0253,
+      "step": 11679
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 3.660926365795725e-05,
+      "loss": 0.0422,
+      "step": 11680
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 3.660332541567696e-05,
+      "loss": 0.036,
+      "step": 11681
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 3.659738717339668e-05,
+      "loss": 0.0656,
+      "step": 11682
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 3.659144893111639e-05,
+      "loss": 0.031,
+      "step": 11683
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 3.6585510688836105e-05,
+      "loss": 0.0302,
+      "step": 11684
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 3.657957244655582e-05,
+      "loss": 0.0664,
+      "step": 11685
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 3.657363420427554e-05,
+      "loss": 0.0776,
+      "step": 11686
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 3.656769596199525e-05,
+      "loss": 0.0417,
+      "step": 11687
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 3.6561757719714966e-05,
+      "loss": 0.0481,
+      "step": 11688
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 3.655581947743468e-05,
+      "loss": 0.0766,
+      "step": 11689
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 3.65498812351544e-05,
+      "loss": 0.0441,
+      "step": 11690
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 3.654394299287411e-05,
+      "loss": 0.0636,
+      "step": 11691
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 3.653800475059383e-05,
+      "loss": 0.0598,
+      "step": 11692
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 3.653206650831354e-05,
+      "loss": 0.0498,
+      "step": 11693
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 3.6526128266033254e-05,
+      "loss": 0.0711,
+      "step": 11694
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 3.652019002375297e-05,
+      "loss": 0.014,
+      "step": 11695
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 3.651425178147269e-05,
+      "loss": 0.1032,
+      "step": 11696
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 3.65083135391924e-05,
+      "loss": 0.0933,
+      "step": 11697
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 3.6502375296912115e-05,
+      "loss": 0.1038,
+      "step": 11698
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 3.649643705463183e-05,
+      "loss": 0.0729,
+      "step": 11699
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 3.649049881235155e-05,
+      "loss": 0.0766,
+      "step": 11700
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 3.648456057007126e-05,
+      "loss": 0.1602,
+      "step": 11701
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 3.6478622327790977e-05,
+      "loss": 0.0664,
+      "step": 11702
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 3.647268408551069e-05,
+      "loss": 0.0725,
+      "step": 11703
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 3.6466745843230404e-05,
+      "loss": 0.1097,
+      "step": 11704
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 3.646080760095012e-05,
+      "loss": 0.2299,
+      "step": 11705
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 3.645486935866984e-05,
+      "loss": 0.0661,
+      "step": 11706
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 3.644893111638955e-05,
+      "loss": 0.0806,
+      "step": 11707
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 3.6442992874109265e-05,
+      "loss": 0.0841,
+      "step": 11708
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 3.6437054631828985e-05,
+      "loss": 0.0955,
+      "step": 11709
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 3.64311163895487e-05,
+      "loss": 0.0515,
+      "step": 11710
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 3.642517814726841e-05,
+      "loss": 0.0444,
+      "step": 11711
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 3.6419239904988126e-05,
+      "loss": 0.0509,
+      "step": 11712
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 3.641330166270784e-05,
+      "loss": 0.0923,
+      "step": 11713
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 3.640736342042755e-05,
+      "loss": 0.0578,
+      "step": 11714
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 3.640142517814727e-05,
+      "loss": 0.0838,
+      "step": 11715
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 3.639548693586699e-05,
+      "loss": 0.0434,
+      "step": 11716
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 3.63895486935867e-05,
+      "loss": 0.0812,
+      "step": 11717
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 3.6383610451306414e-05,
+      "loss": 0.0651,
+      "step": 11718
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 3.6377672209026135e-05,
+      "loss": 0.0539,
+      "step": 11719
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 3.637173396674585e-05,
+      "loss": 0.1358,
+      "step": 11720
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 3.636579572446556e-05,
+      "loss": 0.0656,
+      "step": 11721
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 3.6359857482185275e-05,
+      "loss": 0.121,
+      "step": 11722
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 3.635391923990499e-05,
+      "loss": 0.0416,
+      "step": 11723
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 3.63479809976247e-05,
+      "loss": 0.055,
+      "step": 11724
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 3.6342042755344416e-05,
+      "loss": 0.0957,
+      "step": 11725
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 3.6336104513064136e-05,
+      "loss": 0.0717,
+      "step": 11726
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 3.633016627078385e-05,
+      "loss": 0.04,
+      "step": 11727
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 3.6324228028503564e-05,
+      "loss": 0.0654,
+      "step": 11728
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 3.6318289786223284e-05,
+      "loss": 0.0503,
+      "step": 11729
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 3.6312351543943e-05,
+      "loss": 0.0797,
+      "step": 11730
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 3.630641330166271e-05,
+      "loss": 0.0771,
+      "step": 11731
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 3.6300475059382425e-05,
+      "loss": 0.042,
+      "step": 11732
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 3.629453681710214e-05,
+      "loss": 0.0309,
+      "step": 11733
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 3.628859857482185e-05,
+      "loss": 0.0668,
+      "step": 11734
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 3.6282660332541565e-05,
+      "loss": 0.0535,
+      "step": 11735
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 3.6276722090261286e-05,
+      "loss": 0.0473,
+      "step": 11736
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 3.6270783847981e-05,
+      "loss": 0.0471,
+      "step": 11737
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 3.626484560570071e-05,
+      "loss": 0.0545,
+      "step": 11738
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 3.625890736342043e-05,
+      "loss": 0.0508,
+      "step": 11739
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 3.625296912114015e-05,
+      "loss": 0.0421,
+      "step": 11740
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 3.624703087885986e-05,
+      "loss": 0.0157,
+      "step": 11741
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 3.6241092636579574e-05,
+      "loss": 0.0549,
+      "step": 11742
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 3.623515439429929e-05,
+      "loss": 0.0663,
+      "step": 11743
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 3.6229216152019e-05,
+      "loss": 0.0376,
+      "step": 11744
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 3.6223277909738715e-05,
+      "loss": 0.0755,
+      "step": 11745
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 3.6217339667458435e-05,
+      "loss": 0.0434,
+      "step": 11746
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 3.621140142517815e-05,
+      "loss": 0.1299,
+      "step": 11747
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 3.620546318289786e-05,
+      "loss": 0.0735,
+      "step": 11748
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 3.619952494061758e-05,
+      "loss": 0.1124,
+      "step": 11749
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 3.6193586698337296e-05,
+      "loss": 0.0723,
+      "step": 11750
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 3.618764845605701e-05,
+      "loss": 0.0623,
+      "step": 11751
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 3.618171021377672e-05,
+      "loss": 0.0673,
+      "step": 11752
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 3.617577197149644e-05,
+      "loss": 0.0857,
+      "step": 11753
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 3.616983372921615e-05,
+      "loss": 0.0634,
+      "step": 11754
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 3.6163895486935864e-05,
+      "loss": 0.0589,
+      "step": 11755
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 3.6157957244655584e-05,
+      "loss": 0.0724,
+      "step": 11756
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 3.61520190023753e-05,
+      "loss": 0.0528,
+      "step": 11757
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 3.614608076009501e-05,
+      "loss": 0.0563,
+      "step": 11758
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 3.614014251781473e-05,
+      "loss": 0.0751,
+      "step": 11759
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 3.6134204275534446e-05,
+      "loss": 0.0449,
+      "step": 11760
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 3.612826603325416e-05,
+      "loss": 0.0574,
+      "step": 11761
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 3.612232779097387e-05,
+      "loss": 0.0699,
+      "step": 11762
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 3.611638954869359e-05,
+      "loss": 0.0484,
+      "step": 11763
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 3.61104513064133e-05,
+      "loss": 0.0644,
+      "step": 11764
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 3.6104513064133013e-05,
+      "loss": 0.063,
+      "step": 11765
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 3.6098574821852734e-05,
+      "loss": 0.0532,
+      "step": 11766
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 3.609263657957245e-05,
+      "loss": 0.0686,
+      "step": 11767
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 3.608669833729216e-05,
+      "loss": 0.0448,
+      "step": 11768
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 3.608076009501188e-05,
+      "loss": 0.0651,
+      "step": 11769
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 3.6074821852731595e-05,
+      "loss": 0.0454,
+      "step": 11770
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 3.606888361045131e-05,
+      "loss": 0.2192,
+      "step": 11771
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 3.606294536817102e-05,
+      "loss": 0.0551,
+      "step": 11772
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 3.605700712589074e-05,
+      "loss": 0.0436,
+      "step": 11773
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 3.605106888361045e-05,
+      "loss": 0.0584,
+      "step": 11774
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 3.604513064133016e-05,
+      "loss": 0.0446,
+      "step": 11775
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 3.603919239904988e-05,
+      "loss": 0.0419,
+      "step": 11776
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 3.60332541567696e-05,
+      "loss": 0.057,
+      "step": 11777
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 3.602731591448931e-05,
+      "loss": 0.0417,
+      "step": 11778
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 3.602137767220903e-05,
+      "loss": 0.0473,
+      "step": 11779
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 3.6015439429928744e-05,
+      "loss": 0.0477,
+      "step": 11780
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 3.600950118764846e-05,
+      "loss": 0.0407,
+      "step": 11781
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 3.600356294536817e-05,
+      "loss": 0.0494,
+      "step": 11782
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 3.599762470308789e-05,
+      "loss": 0.0312,
+      "step": 11783
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 3.5991686460807605e-05,
+      "loss": 0.0415,
+      "step": 11784
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 3.598574821852731e-05,
+      "loss": 0.0476,
+      "step": 11785
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 3.597980997624703e-05,
+      "loss": 0.0579,
+      "step": 11786
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 3.5973871733966746e-05,
+      "loss": 0.0572,
+      "step": 11787
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 3.596793349168646e-05,
+      "loss": 0.0218,
+      "step": 11788
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 3.596199524940618e-05,
+      "loss": 0.0522,
+      "step": 11789
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 3.5956057007125894e-05,
+      "loss": 0.0272,
+      "step": 11790
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 3.595011876484561e-05,
+      "loss": 0.0705,
+      "step": 11791
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 3.594418052256532e-05,
+      "loss": 0.0997,
+      "step": 11792
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 3.593824228028504e-05,
+      "loss": 0.0511,
+      "step": 11793
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 3.5932304038004755e-05,
+      "loss": 0.0419,
+      "step": 11794
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 3.592636579572446e-05,
+      "loss": 0.0182,
+      "step": 11795
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 3.592042755344418e-05,
+      "loss": 0.1247,
+      "step": 11796
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 3.5914489311163895e-05,
+      "loss": 0.0971,
+      "step": 11797
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 3.590855106888361e-05,
+      "loss": 0.0881,
+      "step": 11798
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 3.590261282660333e-05,
+      "loss": 0.0799,
+      "step": 11799
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 3.589667458432304e-05,
+      "loss": 0.0881,
+      "step": 11800
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 3.5890736342042757e-05,
+      "loss": 0.0666,
+      "step": 11801
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 3.588479809976247e-05,
+      "loss": 0.0573,
+      "step": 11802
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 3.587885985748219e-05,
+      "loss": 0.0837,
+      "step": 11803
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 3.5872921615201904e-05,
+      "loss": 0.0427,
+      "step": 11804
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 3.586698337292162e-05,
+      "loss": 0.0458,
+      "step": 11805
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 3.586104513064133e-05,
+      "loss": 0.1018,
+      "step": 11806
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 3.5855106888361045e-05,
+      "loss": 0.063,
+      "step": 11807
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 3.584916864608076e-05,
+      "loss": 0.0332,
+      "step": 11808
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 3.584323040380048e-05,
+      "loss": 0.0464,
+      "step": 11809
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 3.583729216152019e-05,
+      "loss": 0.0623,
+      "step": 11810
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 3.5831353919239906e-05,
+      "loss": 0.068,
+      "step": 11811
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 3.582541567695962e-05,
+      "loss": 0.0774,
+      "step": 11812
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 3.581947743467934e-05,
+      "loss": 0.1037,
+      "step": 11813
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 3.5813539192399053e-05,
+      "loss": 0.0747,
+      "step": 11814
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 3.580760095011877e-05,
+      "loss": 0.0369,
+      "step": 11815
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 3.580166270783848e-05,
+      "loss": 0.03,
+      "step": 11816
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 3.5795724465558194e-05,
+      "loss": 0.0567,
+      "step": 11817
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 3.578978622327791e-05,
+      "loss": 0.0759,
+      "step": 11818
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 3.578384798099763e-05,
+      "loss": 0.0409,
+      "step": 11819
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 3.577790973871734e-05,
+      "loss": 0.0394,
+      "step": 11820
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 3.5771971496437055e-05,
+      "loss": 0.066,
+      "step": 11821
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 3.576603325415677e-05,
+      "loss": 0.0607,
+      "step": 11822
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 3.576009501187649e-05,
+      "loss": 0.0348,
+      "step": 11823
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 3.57541567695962e-05,
+      "loss": 0.0485,
+      "step": 11824
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 3.5748218527315916e-05,
+      "loss": 0.0407,
+      "step": 11825
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 3.574228028503563e-05,
+      "loss": 0.0409,
+      "step": 11826
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 3.5736342042755344e-05,
+      "loss": 0.0569,
+      "step": 11827
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 3.573040380047506e-05,
+      "loss": 0.058,
+      "step": 11828
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 3.572446555819478e-05,
+      "loss": 0.0428,
+      "step": 11829
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 3.571852731591449e-05,
+      "loss": 0.0592,
+      "step": 11830
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 3.5712589073634205e-05,
+      "loss": 0.0437,
+      "step": 11831
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 3.5706650831353925e-05,
+      "loss": 0.0763,
+      "step": 11832
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 3.570071258907364e-05,
+      "loss": 0.0429,
+      "step": 11833
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 3.569477434679335e-05,
+      "loss": 0.039,
+      "step": 11834
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 3.5688836104513066e-05,
+      "loss": 0.0293,
+      "step": 11835
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 3.568289786223278e-05,
+      "loss": 0.0342,
+      "step": 11836
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 3.567695961995249e-05,
+      "loss": 0.0307,
+      "step": 11837
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 3.5671021377672206e-05,
+      "loss": 0.055,
+      "step": 11838
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 3.566508313539193e-05,
+      "loss": 0.0286,
+      "step": 11839
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 3.565914489311164e-05,
+      "loss": 0.0537,
+      "step": 11840
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 3.5653206650831354e-05,
+      "loss": 0.0579,
+      "step": 11841
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 3.5647268408551074e-05,
+      "loss": 0.0423,
+      "step": 11842
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 3.564133016627079e-05,
+      "loss": 0.0321,
+      "step": 11843
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 3.56353919239905e-05,
+      "loss": 0.0327,
+      "step": 11844
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 3.5629453681710215e-05,
+      "loss": 0.0402,
+      "step": 11845
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 3.5623515439429935e-05,
+      "loss": 0.0817,
+      "step": 11846
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 3.561757719714964e-05,
+      "loss": 0.0877,
+      "step": 11847
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 3.5611638954869356e-05,
+      "loss": 0.0835,
+      "step": 11848
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 3.5605700712589076e-05,
+      "loss": 0.2479,
+      "step": 11849
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 3.559976247030879e-05,
+      "loss": 0.0738,
+      "step": 11850
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 3.55938242280285e-05,
+      "loss": 0.0536,
+      "step": 11851
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 3.5587885985748224e-05,
+      "loss": 0.0603,
+      "step": 11852
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 3.558194774346794e-05,
+      "loss": 0.0612,
+      "step": 11853
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 3.557600950118765e-05,
+      "loss": 0.1143,
+      "step": 11854
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 3.5570071258907364e-05,
+      "loss": 0.0413,
+      "step": 11855
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 3.5564133016627085e-05,
+      "loss": 0.0621,
+      "step": 11856
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 3.555819477434679e-05,
+      "loss": 0.1428,
+      "step": 11857
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 3.5552256532066505e-05,
+      "loss": 0.0567,
+      "step": 11858
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 3.5546318289786226e-05,
+      "loss": 0.0841,
+      "step": 11859
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 3.554038004750594e-05,
+      "loss": 0.0798,
+      "step": 11860
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 3.553444180522565e-05,
+      "loss": 0.0625,
+      "step": 11861
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 3.552850356294537e-05,
+      "loss": 0.0705,
+      "step": 11862
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 3.552256532066509e-05,
+      "loss": 0.0517,
+      "step": 11863
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 3.55166270783848e-05,
+      "loss": 0.0543,
+      "step": 11864
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 3.5510688836104514e-05,
+      "loss": 0.0526,
+      "step": 11865
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 3.5504750593824234e-05,
+      "loss": 0.0466,
+      "step": 11866
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 3.549881235154395e-05,
+      "loss": 0.0678,
+      "step": 11867
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 3.5492874109263655e-05,
+      "loss": 0.0981,
+      "step": 11868
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 3.5486935866983375e-05,
+      "loss": 0.0564,
+      "step": 11869
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 3.548099762470309e-05,
+      "loss": 0.0758,
+      "step": 11870
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 3.54750593824228e-05,
+      "loss": 0.0847,
+      "step": 11871
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 3.546912114014252e-05,
+      "loss": 0.0829,
+      "step": 11872
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 3.5463182897862236e-05,
+      "loss": 0.0428,
+      "step": 11873
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 3.545724465558195e-05,
+      "loss": 0.0735,
+      "step": 11874
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 3.545130641330166e-05,
+      "loss": 0.0497,
+      "step": 11875
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 3.5445368171021384e-05,
+      "loss": 0.0439,
+      "step": 11876
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 3.54394299287411e-05,
+      "loss": 0.0368,
+      "step": 11877
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 3.5433491686460804e-05,
+      "loss": 0.0541,
+      "step": 11878
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 3.5427553444180524e-05,
+      "loss": 0.0573,
+      "step": 11879
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 3.542161520190024e-05,
+      "loss": 0.0528,
+      "step": 11880
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 3.541567695961995e-05,
+      "loss": 0.0582,
+      "step": 11881
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 3.540973871733967e-05,
+      "loss": 0.0608,
+      "step": 11882
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 3.5403800475059385e-05,
+      "loss": 0.026,
+      "step": 11883
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 3.53978622327791e-05,
+      "loss": 0.0472,
+      "step": 11884
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 3.539192399049881e-05,
+      "loss": 0.0735,
+      "step": 11885
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 3.538598574821853e-05,
+      "loss": 0.0379,
+      "step": 11886
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 3.5380047505938246e-05,
+      "loss": 0.0327,
+      "step": 11887
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 3.537410926365795e-05,
+      "loss": 0.0368,
+      "step": 11888
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 3.5368171021377674e-05,
+      "loss": 0.0455,
+      "step": 11889
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 3.536223277909739e-05,
+      "loss": 0.0322,
+      "step": 11890
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 3.53562945368171e-05,
+      "loss": 0.0321,
+      "step": 11891
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 3.535035629453682e-05,
+      "loss": 0.0697,
+      "step": 11892
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 3.5344418052256535e-05,
+      "loss": 0.0403,
+      "step": 11893
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 3.533847980997625e-05,
+      "loss": 0.0386,
+      "step": 11894
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 3.533254156769596e-05,
+      "loss": 0.0503,
+      "step": 11895
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 3.532660332541568e-05,
+      "loss": 0.1263,
+      "step": 11896
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 3.5320665083135396e-05,
+      "loss": 0.0993,
+      "step": 11897
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 3.531472684085511e-05,
+      "loss": 0.1007,
+      "step": 11898
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 3.530878859857482e-05,
+      "loss": 0.0862,
+      "step": 11899
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 3.5302850356294537e-05,
+      "loss": 0.0725,
+      "step": 11900
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 3.529691211401425e-05,
+      "loss": 0.0718,
+      "step": 11901
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 3.529097387173397e-05,
+      "loss": 0.0627,
+      "step": 11902
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 3.5285035629453684e-05,
+      "loss": 0.1332,
+      "step": 11903
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 3.52790973871734e-05,
+      "loss": 0.0641,
+      "step": 11904
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 3.527315914489311e-05,
+      "loss": 0.056,
+      "step": 11905
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 3.526722090261283e-05,
+      "loss": 0.0682,
+      "step": 11906
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 3.5261282660332545e-05,
+      "loss": 0.0621,
+      "step": 11907
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 3.525534441805226e-05,
+      "loss": 0.1091,
+      "step": 11908
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 3.524940617577197e-05,
+      "loss": 0.1128,
+      "step": 11909
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 3.5243467933491686e-05,
+      "loss": 0.0443,
+      "step": 11910
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 3.52375296912114e-05,
+      "loss": 0.0615,
+      "step": 11911
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 3.523159144893112e-05,
+      "loss": 0.046,
+      "step": 11912
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 3.5225653206650833e-05,
+      "loss": 0.0494,
+      "step": 11913
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 3.521971496437055e-05,
+      "loss": 0.0718,
+      "step": 11914
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 3.521377672209026e-05,
+      "loss": 0.033,
+      "step": 11915
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 3.520783847980998e-05,
+      "loss": 0.0587,
+      "step": 11916
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 3.5201900237529695e-05,
+      "loss": 0.0524,
+      "step": 11917
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 3.519596199524941e-05,
+      "loss": 0.0528,
+      "step": 11918
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 3.519002375296912e-05,
+      "loss": 0.034,
+      "step": 11919
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 3.5184085510688835e-05,
+      "loss": 0.0802,
+      "step": 11920
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 3.517814726840855e-05,
+      "loss": 0.0564,
+      "step": 11921
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 3.517220902612827e-05,
+      "loss": 0.0676,
+      "step": 11922
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 3.516627078384798e-05,
+      "loss": 0.0421,
+      "step": 11923
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 3.5160332541567696e-05,
+      "loss": 0.0496,
+      "step": 11924
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 3.515439429928741e-05,
+      "loss": 0.0656,
+      "step": 11925
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 3.514845605700713e-05,
+      "loss": 0.0363,
+      "step": 11926
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 3.5142517814726844e-05,
+      "loss": 0.031,
+      "step": 11927
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 3.513657957244656e-05,
+      "loss": 0.0774,
+      "step": 11928
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 3.513064133016628e-05,
+      "loss": 0.044,
+      "step": 11929
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 3.5124703087885985e-05,
+      "loss": 0.0288,
+      "step": 11930
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 3.51187648456057e-05,
+      "loss": 0.0412,
+      "step": 11931
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 3.511282660332542e-05,
+      "loss": 0.0431,
+      "step": 11932
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 3.510688836104513e-05,
+      "loss": 0.057,
+      "step": 11933
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 3.5100950118764846e-05,
+      "loss": 0.0585,
+      "step": 11934
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 3.509501187648456e-05,
+      "loss": 0.0628,
+      "step": 11935
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 3.508907363420428e-05,
+      "loss": 0.0328,
+      "step": 11936
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 3.508313539192399e-05,
+      "loss": 0.0861,
+      "step": 11937
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 3.507719714964371e-05,
+      "loss": 0.036,
+      "step": 11938
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 3.507125890736343e-05,
+      "loss": 0.0388,
+      "step": 11939
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 3.5065320665083134e-05,
+      "loss": 0.0275,
+      "step": 11940
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 3.505938242280285e-05,
+      "loss": 0.0232,
+      "step": 11941
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 3.505344418052257e-05,
+      "loss": 0.0342,
+      "step": 11942
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 3.504750593824228e-05,
+      "loss": 0.0345,
+      "step": 11943
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 3.5041567695961995e-05,
+      "loss": 0.1117,
+      "step": 11944
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 3.503562945368171e-05,
+      "loss": 0.0122,
+      "step": 11945
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 3.502969121140143e-05,
+      "loss": 0.0864,
+      "step": 11946
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 3.502375296912114e-05,
+      "loss": 0.1436,
+      "step": 11947
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 3.5017814726840856e-05,
+      "loss": 0.1088,
+      "step": 11948
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 3.5011876484560577e-05,
+      "loss": 0.1076,
+      "step": 11949
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 3.500593824228028e-05,
+      "loss": 0.0666,
+      "step": 11950
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 3.5e-05,
+      "loss": 0.0921,
+      "step": 11951
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 3.499406175771972e-05,
+      "loss": 0.0497,
+      "step": 11952
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 3.498812351543943e-05,
+      "loss": 0.0578,
+      "step": 11953
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 3.4982185273159144e-05,
+      "loss": 0.0845,
+      "step": 11954
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 3.4976247030878865e-05,
+      "loss": 0.0662,
+      "step": 11955
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 3.497030878859858e-05,
+      "loss": 0.0807,
+      "step": 11956
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 3.496437054631829e-05,
+      "loss": 0.1027,
+      "step": 11957
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 3.4958432304038006e-05,
+      "loss": 0.0476,
+      "step": 11958
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 3.4952494061757726e-05,
+      "loss": 0.0669,
+      "step": 11959
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 3.494655581947744e-05,
+      "loss": 0.0672,
+      "step": 11960
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 3.4940617577197146e-05,
+      "loss": 0.049,
+      "step": 11961
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 3.493467933491687e-05,
+      "loss": 0.0389,
+      "step": 11962
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 3.492874109263658e-05,
+      "loss": 0.0629,
+      "step": 11963
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 3.4922802850356294e-05,
+      "loss": 0.0967,
+      "step": 11964
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 3.4916864608076014e-05,
+      "loss": 0.053,
+      "step": 11965
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 3.491092636579573e-05,
+      "loss": 0.068,
+      "step": 11966
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 3.490498812351544e-05,
+      "loss": 0.0461,
+      "step": 11967
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 3.4899049881235155e-05,
+      "loss": 0.0678,
+      "step": 11968
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 3.4893111638954875e-05,
+      "loss": 0.0881,
+      "step": 11969
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 3.488717339667459e-05,
+      "loss": 0.0386,
+      "step": 11970
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 3.4881235154394296e-05,
+      "loss": 0.0482,
+      "step": 11971
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 3.4875296912114016e-05,
+      "loss": 0.0635,
+      "step": 11972
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 3.486935866983373e-05,
+      "loss": 0.0451,
+      "step": 11973
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 3.486342042755344e-05,
+      "loss": 0.0728,
+      "step": 11974
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 3.4857482185273164e-05,
+      "loss": 0.0378,
+      "step": 11975
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 3.485154394299288e-05,
+      "loss": 0.1114,
+      "step": 11976
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 3.484560570071259e-05,
+      "loss": 0.0703,
+      "step": 11977
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 3.4839667458432304e-05,
+      "loss": 0.0389,
+      "step": 11978
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 3.4833729216152025e-05,
+      "loss": 0.0444,
+      "step": 11979
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 3.482779097387174e-05,
+      "loss": 0.0841,
+      "step": 11980
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 3.482185273159145e-05,
+      "loss": 0.1566,
+      "step": 11981
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 3.4815914489311165e-05,
+      "loss": 0.0896,
+      "step": 11982
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 3.480997624703088e-05,
+      "loss": 0.0295,
+      "step": 11983
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 3.480403800475059e-05,
+      "loss": 0.0337,
+      "step": 11984
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 3.479809976247031e-05,
+      "loss": 0.0427,
+      "step": 11985
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 3.4792161520190026e-05,
+      "loss": 0.0614,
+      "step": 11986
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 3.478622327790974e-05,
+      "loss": 0.0313,
+      "step": 11987
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 3.4780285035629454e-05,
+      "loss": 0.057,
+      "step": 11988
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 3.4774346793349174e-05,
+      "loss": 0.0527,
+      "step": 11989
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 3.476840855106889e-05,
+      "loss": 0.0492,
+      "step": 11990
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 3.47624703087886e-05,
+      "loss": 0.0392,
+      "step": 11991
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 3.4756532066508315e-05,
+      "loss": 0.0539,
+      "step": 11992
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 3.475059382422803e-05,
+      "loss": 0.0511,
+      "step": 11993
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 3.474465558194774e-05,
+      "loss": 0.0336,
+      "step": 11994
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 3.473871733966746e-05,
+      "loss": 0.0215,
+      "step": 11995
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 3.4732779097387176e-05,
+      "loss": 0.0891,
+      "step": 11996
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 3.472684085510689e-05,
+      "loss": 0.103,
+      "step": 11997
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 3.47209026128266e-05,
+      "loss": 0.0826,
+      "step": 11998
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 3.471496437054632e-05,
+      "loss": 0.0473,
+      "step": 11999
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 3.470902612826604e-05,
+      "loss": 0.088,
+      "step": 12000
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 3.470308788598575e-05,
+      "loss": 0.0564,
+      "step": 12001
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 3.4697149643705464e-05,
+      "loss": 0.0739,
+      "step": 12002
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 3.469121140142518e-05,
+      "loss": 0.0971,
+      "step": 12003
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 3.468527315914489e-05,
+      "loss": 0.1023,
+      "step": 12004
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 3.467933491686461e-05,
+      "loss": 0.0701,
+      "step": 12005
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 3.4673396674584325e-05,
+      "loss": 0.07,
+      "step": 12006
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 3.466745843230404e-05,
+      "loss": 0.068,
+      "step": 12007
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 3.466152019002375e-05,
+      "loss": 0.0675,
+      "step": 12008
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 3.465558194774347e-05,
+      "loss": 0.0519,
+      "step": 12009
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 3.4649643705463186e-05,
+      "loss": 0.0618,
+      "step": 12010
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 3.46437054631829e-05,
+      "loss": 0.0545,
+      "step": 12011
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 3.463776722090262e-05,
+      "loss": 0.0546,
+      "step": 12012
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 3.463182897862233e-05,
+      "loss": 0.0409,
+      "step": 12013
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 3.462589073634204e-05,
+      "loss": 0.0507,
+      "step": 12014
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 3.461995249406176e-05,
+      "loss": 0.0376,
+      "step": 12015
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 3.4614014251781475e-05,
+      "loss": 0.0487,
+      "step": 12016
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 3.460807600950119e-05,
+      "loss": 0.0544,
+      "step": 12017
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 3.46021377672209e-05,
+      "loss": 0.0519,
+      "step": 12018
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 3.459619952494062e-05,
+      "loss": 0.0637,
+      "step": 12019
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 3.4590261282660336e-05,
+      "loss": 0.056,
+      "step": 12020
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 3.458432304038005e-05,
+      "loss": 0.0356,
+      "step": 12021
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 3.457838479809977e-05,
+      "loss": 0.0607,
+      "step": 12022
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 3.4572446555819476e-05,
+      "loss": 0.0586,
+      "step": 12023
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 3.456650831353919e-05,
+      "loss": 0.0714,
+      "step": 12024
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 3.456057007125891e-05,
+      "loss": 0.1014,
+      "step": 12025
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 3.4554631828978624e-05,
+      "loss": 0.0634,
+      "step": 12026
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 3.454869358669834e-05,
+      "loss": 0.0577,
+      "step": 12027
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 3.454275534441805e-05,
+      "loss": 0.0557,
+      "step": 12028
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 3.453681710213777e-05,
+      "loss": 0.0414,
+      "step": 12029
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 3.4530878859857485e-05,
+      "loss": 0.0486,
+      "step": 12030
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 3.45249406175772e-05,
+      "loss": 0.0357,
+      "step": 12031
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 3.451900237529692e-05,
+      "loss": 0.0357,
+      "step": 12032
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 3.4513064133016626e-05,
+      "loss": 0.0605,
+      "step": 12033
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 3.450712589073634e-05,
+      "loss": 0.0297,
+      "step": 12034
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 3.450118764845606e-05,
+      "loss": 0.0623,
+      "step": 12035
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 3.449524940617577e-05,
+      "loss": 0.0363,
+      "step": 12036
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 3.448931116389549e-05,
+      "loss": 0.0639,
+      "step": 12037
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 3.44833729216152e-05,
+      "loss": 0.0516,
+      "step": 12038
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 3.447743467933492e-05,
+      "loss": 0.0463,
+      "step": 12039
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 3.4471496437054634e-05,
+      "loss": 0.0398,
+      "step": 12040
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 3.446555819477435e-05,
+      "loss": 0.1099,
+      "step": 12041
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 3.445961995249407e-05,
+      "loss": 0.034,
+      "step": 12042
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 3.445368171021378e-05,
+      "loss": 0.0372,
+      "step": 12043
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 3.444774346793349e-05,
+      "loss": 0.0447,
+      "step": 12044
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 3.444180522565321e-05,
+      "loss": 0.0356,
+      "step": 12045
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 3.443586698337292e-05,
+      "loss": 0.0637,
+      "step": 12046
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 3.4429928741092636e-05,
+      "loss": 0.0697,
+      "step": 12047
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 3.442399049881235e-05,
+      "loss": 0.1328,
+      "step": 12048
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 3.441805225653207e-05,
+      "loss": 0.0502,
+      "step": 12049
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 3.4412114014251784e-05,
+      "loss": 0.0768,
+      "step": 12050
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 3.44061757719715e-05,
+      "loss": 0.0674,
+      "step": 12051
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 3.440023752969122e-05,
+      "loss": 0.0571,
+      "step": 12052
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 3.439429928741093e-05,
+      "loss": 0.071,
+      "step": 12053
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 3.438836104513064e-05,
+      "loss": 0.0864,
+      "step": 12054
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 3.438242280285036e-05,
+      "loss": 0.1092,
+      "step": 12055
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 3.437648456057007e-05,
+      "loss": 0.0418,
+      "step": 12056
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 3.4370546318289786e-05,
+      "loss": 0.0601,
+      "step": 12057
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 3.43646080760095e-05,
+      "loss": 0.0387,
+      "step": 12058
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 3.435866983372922e-05,
+      "loss": 0.0554,
+      "step": 12059
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 3.435273159144893e-05,
+      "loss": 0.0567,
+      "step": 12060
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 3.434679334916865e-05,
+      "loss": 0.0407,
+      "step": 12061
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 3.434085510688837e-05,
+      "loss": 0.0577,
+      "step": 12062
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 3.433491686460808e-05,
+      "loss": 0.0601,
+      "step": 12063
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 3.4328978622327794e-05,
+      "loss": 0.0723,
+      "step": 12064
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 3.432304038004751e-05,
+      "loss": 0.0465,
+      "step": 12065
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 3.431710213776722e-05,
+      "loss": 0.0607,
+      "step": 12066
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 3.4311163895486935e-05,
+      "loss": 0.0534,
+      "step": 12067
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 3.430522565320665e-05,
+      "loss": 0.0513,
+      "step": 12068
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 3.429928741092637e-05,
+      "loss": 0.0582,
+      "step": 12069
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 3.429334916864608e-05,
+      "loss": 0.0533,
+      "step": 12070
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 3.4287410926365796e-05,
+      "loss": 0.0745,
+      "step": 12071
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 3.4281472684085516e-05,
+      "loss": 0.0549,
+      "step": 12072
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 3.427553444180523e-05,
+      "loss": 0.0505,
+      "step": 12073
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 3.4269596199524944e-05,
+      "loss": 0.0511,
+      "step": 12074
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 3.426365795724466e-05,
+      "loss": 0.0395,
+      "step": 12075
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 3.425771971496437e-05,
+      "loss": 0.0374,
+      "step": 12076
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 3.4251781472684084e-05,
+      "loss": 0.0734,
+      "step": 12077
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 3.4245843230403805e-05,
+      "loss": 0.051,
+      "step": 12078
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 3.423990498812352e-05,
+      "loss": 0.0627,
+      "step": 12079
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 3.423396674584323e-05,
+      "loss": 0.04,
+      "step": 12080
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 3.4228028503562945e-05,
+      "loss": 0.0711,
+      "step": 12081
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 3.4222090261282666e-05,
+      "loss": 0.0321,
+      "step": 12082
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 3.421615201900238e-05,
+      "loss": 0.0557,
+      "step": 12083
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 3.421021377672209e-05,
+      "loss": 0.0637,
+      "step": 12084
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 3.4204275534441806e-05,
+      "loss": 0.0479,
+      "step": 12085
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 3.419833729216152e-05,
+      "loss": 0.0784,
+      "step": 12086
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 3.4192399049881234e-05,
+      "loss": 0.087,
+      "step": 12087
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 3.4186460807600954e-05,
+      "loss": 0.0545,
+      "step": 12088
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 3.418052256532067e-05,
+      "loss": 0.0339,
+      "step": 12089
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 3.417458432304038e-05,
+      "loss": 0.0656,
+      "step": 12090
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 3.4168646080760095e-05,
+      "loss": 0.0265,
+      "step": 12091
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 3.4162707838479815e-05,
+      "loss": 0.0227,
+      "step": 12092
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 3.415676959619953e-05,
+      "loss": 0.0962,
+      "step": 12093
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 3.415083135391924e-05,
+      "loss": 0.0267,
+      "step": 12094
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 3.4144893111638956e-05,
+      "loss": 0.0533,
+      "step": 12095
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 3.413895486935867e-05,
+      "loss": 0.047,
+      "step": 12096
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 3.413301662707838e-05,
+      "loss": 0.0954,
+      "step": 12097
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 3.41270783847981e-05,
+      "loss": 0.0802,
+      "step": 12098
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 3.412114014251782e-05,
+      "loss": 0.0807,
+      "step": 12099
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 3.411520190023753e-05,
+      "loss": 0.1096,
+      "step": 12100
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 3.4109263657957244e-05,
+      "loss": 0.1244,
+      "step": 12101
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 3.4103325415676964e-05,
+      "loss": 0.0528,
+      "step": 12102
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 3.409738717339668e-05,
+      "loss": 0.065,
+      "step": 12103
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 3.409144893111639e-05,
+      "loss": 0.0707,
+      "step": 12104
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 3.408551068883611e-05,
+      "loss": 0.0575,
+      "step": 12105
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 3.407957244655582e-05,
+      "loss": 0.0407,
+      "step": 12106
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 3.407363420427553e-05,
+      "loss": 0.042,
+      "step": 12107
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 3.406769596199525e-05,
+      "loss": 0.0672,
+      "step": 12108
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 3.4061757719714966e-05,
+      "loss": 0.0879,
+      "step": 12109
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 3.405581947743468e-05,
+      "loss": 0.0463,
+      "step": 12110
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 3.4049881235154393e-05,
+      "loss": 0.0513,
+      "step": 12111
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 3.4043942992874114e-05,
+      "loss": 0.0369,
+      "step": 12112
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 3.403800475059383e-05,
+      "loss": 0.0725,
+      "step": 12113
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 3.403206650831354e-05,
+      "loss": 0.1029,
+      "step": 12114
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 3.402612826603326e-05,
+      "loss": 0.065,
+      "step": 12115
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 3.402019002375297e-05,
+      "loss": 0.1032,
+      "step": 12116
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 3.401425178147268e-05,
+      "loss": 0.0571,
+      "step": 12117
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 3.40083135391924e-05,
+      "loss": 0.0459,
+      "step": 12118
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 3.4002375296912116e-05,
+      "loss": 0.0429,
+      "step": 12119
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 3.399643705463183e-05,
+      "loss": 0.0358,
+      "step": 12120
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 3.399049881235154e-05,
+      "loss": 0.0472,
+      "step": 12121
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 3.398456057007126e-05,
+      "loss": 0.032,
+      "step": 12122
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 3.397862232779098e-05,
+      "loss": 0.0412,
+      "step": 12123
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 3.397268408551069e-05,
+      "loss": 0.0572,
+      "step": 12124
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 3.396674584323041e-05,
+      "loss": 0.048,
+      "step": 12125
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 3.3960807600950124e-05,
+      "loss": 0.0704,
+      "step": 12126
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 3.395486935866983e-05,
+      "loss": 0.077,
+      "step": 12127
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 3.394893111638955e-05,
+      "loss": 0.0509,
+      "step": 12128
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 3.3942992874109265e-05,
+      "loss": 0.0773,
+      "step": 12129
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 3.393705463182898e-05,
+      "loss": 0.092,
+      "step": 12130
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 3.393111638954869e-05,
+      "loss": 0.0414,
+      "step": 12131
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 3.392517814726841e-05,
+      "loss": 0.0543,
+      "step": 12132
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 3.3919239904988126e-05,
+      "loss": 0.0371,
+      "step": 12133
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 3.391330166270784e-05,
+      "loss": 0.0359,
+      "step": 12134
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 3.390736342042756e-05,
+      "loss": 0.0722,
+      "step": 12135
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 3.3901425178147274e-05,
+      "loss": 0.0484,
+      "step": 12136
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 3.389548693586698e-05,
+      "loss": 0.0217,
+      "step": 12137
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 3.38895486935867e-05,
+      "loss": 0.0593,
+      "step": 12138
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 3.3883610451306414e-05,
+      "loss": 0.0416,
+      "step": 12139
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 3.387767220902613e-05,
+      "loss": 0.0558,
+      "step": 12140
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 3.387173396674584e-05,
+      "loss": 0.0582,
+      "step": 12141
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 3.386579572446556e-05,
+      "loss": 0.0383,
+      "step": 12142
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 3.3859857482185275e-05,
+      "loss": 0.0839,
+      "step": 12143
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 3.385391923990499e-05,
+      "loss": 0.0323,
+      "step": 12144
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 3.384798099762471e-05,
+      "loss": 0.0532,
+      "step": 12145
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 3.384204275534442e-05,
+      "loss": 0.0567,
+      "step": 12146
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 3.383610451306413e-05,
+      "loss": 0.1075,
+      "step": 12147
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 3.383016627078385e-05,
+      "loss": 0.0507,
+      "step": 12148
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 3.3824228028503564e-05,
+      "loss": 0.1173,
+      "step": 12149
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 3.381828978622328e-05,
+      "loss": 0.115,
+      "step": 12150
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 3.381235154394299e-05,
+      "loss": 0.0589,
+      "step": 12151
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 3.380641330166271e-05,
+      "loss": 0.0647,
+      "step": 12152
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 3.3800475059382425e-05,
+      "loss": 0.0662,
+      "step": 12153
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 3.379453681710214e-05,
+      "loss": 0.0391,
+      "step": 12154
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 3.378859857482186e-05,
+      "loss": 0.0514,
+      "step": 12155
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 3.378266033254157e-05,
+      "loss": 0.077,
+      "step": 12156
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 3.3776722090261286e-05,
+      "loss": 0.0612,
+      "step": 12157
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 3.3770783847981e-05,
+      "loss": 0.0551,
+      "step": 12158
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 3.376484560570071e-05,
+      "loss": 0.0775,
+      "step": 12159
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 3.375890736342043e-05,
+      "loss": 0.0377,
+      "step": 12160
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 3.375296912114014e-05,
+      "loss": 0.0501,
+      "step": 12161
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 3.374703087885986e-05,
+      "loss": 0.0571,
+      "step": 12162
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 3.3741092636579574e-05,
+      "loss": 0.067,
+      "step": 12163
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 3.373515439429929e-05,
+      "loss": 0.0743,
+      "step": 12164
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 3.372921615201901e-05,
+      "loss": 0.0408,
+      "step": 12165
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 3.372327790973872e-05,
+      "loss": 0.0668,
+      "step": 12166
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 3.3717339667458435e-05,
+      "loss": 0.0662,
+      "step": 12167
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 3.371140142517815e-05,
+      "loss": 0.0503,
+      "step": 12168
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 3.370546318289786e-05,
+      "loss": 0.0376,
+      "step": 12169
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 3.3699524940617576e-05,
+      "loss": 0.0504,
+      "step": 12170
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 3.369358669833729e-05,
+      "loss": 0.0722,
+      "step": 12171
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 3.368764845605701e-05,
+      "loss": 0.0653,
+      "step": 12172
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 3.3681710213776724e-05,
+      "loss": 0.0766,
+      "step": 12173
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 3.367577197149644e-05,
+      "loss": 0.0563,
+      "step": 12174
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 3.366983372921616e-05,
+      "loss": 0.0909,
+      "step": 12175
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 3.366389548693587e-05,
+      "loss": 0.0938,
+      "step": 12176
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 3.3657957244655585e-05,
+      "loss": 0.0607,
+      "step": 12177
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 3.36520190023753e-05,
+      "loss": 0.0363,
+      "step": 12178
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 3.364608076009501e-05,
+      "loss": 0.0348,
+      "step": 12179
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 3.3640142517814725e-05,
+      "loss": 0.0341,
+      "step": 12180
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 3.363420427553444e-05,
+      "loss": 0.0347,
+      "step": 12181
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 3.362826603325416e-05,
+      "loss": 0.0369,
+      "step": 12182
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 3.362232779097387e-05,
+      "loss": 0.0521,
+      "step": 12183
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 3.3616389548693586e-05,
+      "loss": 0.0695,
+      "step": 12184
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 3.361045130641331e-05,
+      "loss": 0.0359,
+      "step": 12185
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 3.360451306413302e-05,
+      "loss": 0.0722,
+      "step": 12186
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 3.3598574821852734e-05,
+      "loss": 0.064,
+      "step": 12187
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 3.359263657957245e-05,
+      "loss": 0.0543,
+      "step": 12188
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 3.358669833729216e-05,
+      "loss": 0.067,
+      "step": 12189
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 3.3580760095011875e-05,
+      "loss": 0.0363,
+      "step": 12190
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 3.357482185273159e-05,
+      "loss": 0.0294,
+      "step": 12191
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 3.356888361045131e-05,
+      "loss": 0.0345,
+      "step": 12192
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 3.356294536817102e-05,
+      "loss": 0.0367,
+      "step": 12193
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 3.3557007125890736e-05,
+      "loss": 0.0288,
+      "step": 12194
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 3.3551068883610456e-05,
+      "loss": 0.0571,
+      "step": 12195
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 3.354513064133017e-05,
+      "loss": 0.062,
+      "step": 12196
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 3.353919239904988e-05,
+      "loss": 0.0725,
+      "step": 12197
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 3.35332541567696e-05,
+      "loss": 0.0827,
+      "step": 12198
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 3.352731591448931e-05,
+      "loss": 0.1127,
+      "step": 12199
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 3.3521377672209024e-05,
+      "loss": 0.0609,
+      "step": 12200
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 3.3515439429928744e-05,
+      "loss": 0.0796,
+      "step": 12201
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 3.350950118764846e-05,
+      "loss": 0.0753,
+      "step": 12202
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 3.350356294536817e-05,
+      "loss": 0.064,
+      "step": 12203
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 3.3497624703087885e-05,
+      "loss": 0.0653,
+      "step": 12204
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 3.3491686460807606e-05,
+      "loss": 0.0458,
+      "step": 12205
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 3.348574821852732e-05,
+      "loss": 0.0627,
+      "step": 12206
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 3.347980997624703e-05,
+      "loss": 0.0837,
+      "step": 12207
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 3.3473871733966746e-05,
+      "loss": 0.0623,
+      "step": 12208
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 3.346793349168647e-05,
+      "loss": 0.0551,
+      "step": 12209
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 3.3461995249406173e-05,
+      "loss": 0.0691,
+      "step": 12210
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 3.3456057007125894e-05,
+      "loss": 0.047,
+      "step": 12211
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 3.345011876484561e-05,
+      "loss": 0.0403,
+      "step": 12212
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 3.344418052256532e-05,
+      "loss": 0.0589,
+      "step": 12213
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 3.3438242280285035e-05,
+      "loss": 0.0829,
+      "step": 12214
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 3.3432304038004755e-05,
+      "loss": 0.0602,
+      "step": 12215
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 3.342636579572447e-05,
+      "loss": 0.0445,
+      "step": 12216
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 3.342042755344418e-05,
+      "loss": 0.0551,
+      "step": 12217
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 3.3414489311163896e-05,
+      "loss": 0.0792,
+      "step": 12218
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 3.3408551068883616e-05,
+      "loss": 0.0434,
+      "step": 12219
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 3.340261282660332e-05,
+      "loss": 0.0426,
+      "step": 12220
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 3.339667458432304e-05,
+      "loss": 0.0619,
+      "step": 12221
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 3.339073634204276e-05,
+      "loss": 0.0585,
+      "step": 12222
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 3.338479809976247e-05,
+      "loss": 0.0474,
+      "step": 12223
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 3.3378859857482184e-05,
+      "loss": 0.051,
+      "step": 12224
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 3.3372921615201904e-05,
+      "loss": 0.0349,
+      "step": 12225
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 3.336698337292162e-05,
+      "loss": 0.0342,
+      "step": 12226
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 3.336104513064133e-05,
+      "loss": 0.048,
+      "step": 12227
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 3.335510688836105e-05,
+      "loss": 0.0554,
+      "step": 12228
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 3.3349168646080765e-05,
+      "loss": 0.0293,
+      "step": 12229
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 3.334323040380047e-05,
+      "loss": 0.0506,
+      "step": 12230
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 3.333729216152019e-05,
+      "loss": 0.0631,
+      "step": 12231
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 3.3331353919239906e-05,
+      "loss": 0.0879,
+      "step": 12232
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 3.332541567695962e-05,
+      "loss": 0.0651,
+      "step": 12233
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 3.331947743467933e-05,
+      "loss": 0.0594,
+      "step": 12234
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 3.3313539192399054e-05,
+      "loss": 0.0504,
+      "step": 12235
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 3.330760095011877e-05,
+      "loss": 0.0673,
+      "step": 12236
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 3.330166270783848e-05,
+      "loss": 0.0461,
+      "step": 12237
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 3.32957244655582e-05,
+      "loss": 0.0839,
+      "step": 12238
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 3.3289786223277915e-05,
+      "loss": 0.0611,
+      "step": 12239
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 3.328384798099763e-05,
+      "loss": 0.0468,
+      "step": 12240
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 3.327790973871734e-05,
+      "loss": 0.1256,
+      "step": 12241
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 3.3271971496437055e-05,
+      "loss": 0.0576,
+      "step": 12242
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 3.326603325415677e-05,
+      "loss": 0.0074,
+      "step": 12243
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 3.326009501187648e-05,
+      "loss": 0.043,
+      "step": 12244
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 3.32541567695962e-05,
+      "loss": 0.0366,
+      "step": 12245
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 3.3248218527315917e-05,
+      "loss": 0.0192,
+      "step": 12246
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 3.324228028503563e-05,
+      "loss": 0.1363,
+      "step": 12247
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 3.323634204275535e-05,
+      "loss": 0.0912,
+      "step": 12248
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 3.3230403800475064e-05,
+      "loss": 0.0747,
+      "step": 12249
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 3.322446555819478e-05,
+      "loss": 0.0626,
+      "step": 12250
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 3.321852731591449e-05,
+      "loss": 0.0664,
+      "step": 12251
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 3.3212589073634205e-05,
+      "loss": 0.0547,
+      "step": 12252
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 3.320665083135392e-05,
+      "loss": 0.0382,
+      "step": 12253
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 3.320071258907363e-05,
+      "loss": 0.1769,
+      "step": 12254
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 3.319477434679335e-05,
+      "loss": 0.0708,
+      "step": 12255
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 3.3188836104513066e-05,
+      "loss": 0.0706,
+      "step": 12256
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 3.318289786223278e-05,
+      "loss": 0.0898,
+      "step": 12257
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 3.31769596199525e-05,
+      "loss": 0.0517,
+      "step": 12258
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 3.3171021377672213e-05,
+      "loss": 0.0702,
+      "step": 12259
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 3.316508313539193e-05,
+      "loss": 0.0913,
+      "step": 12260
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 3.315914489311164e-05,
+      "loss": 0.0379,
+      "step": 12261
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 3.3153206650831354e-05,
+      "loss": 0.05,
+      "step": 12262
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 3.314726840855107e-05,
+      "loss": 0.1282,
+      "step": 12263
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 3.314133016627078e-05,
+      "loss": 0.061,
+      "step": 12264
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 3.31353919239905e-05,
+      "loss": 0.0615,
+      "step": 12265
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 3.3129453681710215e-05,
+      "loss": 0.0487,
+      "step": 12266
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 3.312351543942993e-05,
+      "loss": 0.0588,
+      "step": 12267
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 3.311757719714965e-05,
+      "loss": 0.061,
+      "step": 12268
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 3.311163895486936e-05,
+      "loss": 0.1431,
+      "step": 12269
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 3.3105700712589076e-05,
+      "loss": 0.0534,
+      "step": 12270
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 3.309976247030879e-05,
+      "loss": 0.0752,
+      "step": 12271
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 3.3093824228028504e-05,
+      "loss": 0.0373,
+      "step": 12272
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 3.308788598574822e-05,
+      "loss": 0.0572,
+      "step": 12273
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 3.308194774346793e-05,
+      "loss": 0.0858,
+      "step": 12274
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 3.307600950118765e-05,
+      "loss": 0.0703,
+      "step": 12275
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 3.3070071258907365e-05,
+      "loss": 0.0489,
+      "step": 12276
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 3.306413301662708e-05,
+      "loss": 0.0772,
+      "step": 12277
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 3.30581947743468e-05,
+      "loss": 0.0812,
+      "step": 12278
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 3.305225653206651e-05,
+      "loss": 0.0358,
+      "step": 12279
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 3.3046318289786226e-05,
+      "loss": 0.0388,
+      "step": 12280
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 3.304038004750594e-05,
+      "loss": 0.0593,
+      "step": 12281
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 3.303444180522565e-05,
+      "loss": 0.0543,
+      "step": 12282
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 3.3028503562945367e-05,
+      "loss": 0.0628,
+      "step": 12283
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 3.302256532066508e-05,
+      "loss": 0.05,
+      "step": 12284
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 3.30166270783848e-05,
+      "loss": 0.0443,
+      "step": 12285
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 3.3010688836104514e-05,
+      "loss": 0.0384,
+      "step": 12286
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 3.300475059382423e-05,
+      "loss": 0.0517,
+      "step": 12287
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 3.299881235154395e-05,
+      "loss": 0.0431,
+      "step": 12288
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 3.299287410926366e-05,
+      "loss": 0.0602,
+      "step": 12289
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 3.2986935866983375e-05,
+      "loss": 0.0157,
+      "step": 12290
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 3.298099762470309e-05,
+      "loss": 0.0492,
+      "step": 12291
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 3.297505938242281e-05,
+      "loss": 0.0258,
+      "step": 12292
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 3.2969121140142516e-05,
+      "loss": 0.0586,
+      "step": 12293
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 3.296318289786223e-05,
+      "loss": 0.0779,
+      "step": 12294
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 3.295724465558195e-05,
+      "loss": 0.0691,
+      "step": 12295
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 3.295130641330166e-05,
+      "loss": 0.117,
+      "step": 12296
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 3.294536817102138e-05,
+      "loss": 0.101,
+      "step": 12297
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 3.29394299287411e-05,
+      "loss": 0.0602,
+      "step": 12298
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 3.293349168646081e-05,
+      "loss": 0.0848,
+      "step": 12299
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 3.2927553444180524e-05,
+      "loss": 0.0873,
+      "step": 12300
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 3.292161520190024e-05,
+      "loss": 0.0441,
+      "step": 12301
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 3.291567695961996e-05,
+      "loss": 0.0649,
+      "step": 12302
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 3.2909738717339665e-05,
+      "loss": 0.075,
+      "step": 12303
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 3.290380047505938e-05,
+      "loss": 0.0925,
+      "step": 12304
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 3.28978622327791e-05,
+      "loss": 0.0983,
+      "step": 12305
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 3.289192399049881e-05,
+      "loss": 0.0784,
+      "step": 12306
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 3.2885985748218526e-05,
+      "loss": 0.0789,
+      "step": 12307
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 3.288004750593825e-05,
+      "loss": 0.061,
+      "step": 12308
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 3.287410926365796e-05,
+      "loss": 0.0569,
+      "step": 12309
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 3.2868171021377674e-05,
+      "loss": 0.0639,
+      "step": 12310
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 3.286223277909739e-05,
+      "loss": 0.0604,
+      "step": 12311
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 3.285629453681711e-05,
+      "loss": 0.0517,
+      "step": 12312
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 3.2850356294536815e-05,
+      "loss": 0.0647,
+      "step": 12313
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 3.284441805225653e-05,
+      "loss": 0.0476,
+      "step": 12314
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 3.283847980997625e-05,
+      "loss": 0.0363,
+      "step": 12315
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 3.283254156769596e-05,
+      "loss": 0.0594,
+      "step": 12316
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 3.2826603325415676e-05,
+      "loss": 0.0514,
+      "step": 12317
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 3.2820665083135396e-05,
+      "loss": 0.0516,
+      "step": 12318
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 3.281472684085511e-05,
+      "loss": 0.0591,
+      "step": 12319
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 3.280878859857482e-05,
+      "loss": 0.0461,
+      "step": 12320
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 3.280285035629454e-05,
+      "loss": 0.0467,
+      "step": 12321
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 3.279691211401426e-05,
+      "loss": 0.1392,
+      "step": 12322
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 3.279097387173397e-05,
+      "loss": 0.0529,
+      "step": 12323
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 3.278503562945368e-05,
+      "loss": 0.0465,
+      "step": 12324
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 3.27790973871734e-05,
+      "loss": 0.0235,
+      "step": 12325
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 3.277315914489311e-05,
+      "loss": 0.0914,
+      "step": 12326
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 3.2767220902612825e-05,
+      "loss": 0.0383,
+      "step": 12327
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 3.2761282660332545e-05,
+      "loss": 0.0545,
+      "step": 12328
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 3.275534441805226e-05,
+      "loss": 0.056,
+      "step": 12329
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 3.274940617577197e-05,
+      "loss": 0.0385,
+      "step": 12330
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 3.2743467933491686e-05,
+      "loss": 0.037,
+      "step": 12331
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 3.2737529691211406e-05,
+      "loss": 0.0316,
+      "step": 12332
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 3.273159144893112e-05,
+      "loss": 0.0497,
+      "step": 12333
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 3.2725653206650834e-05,
+      "loss": 0.0364,
+      "step": 12334
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 3.271971496437055e-05,
+      "loss": 0.036,
+      "step": 12335
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 3.271377672209026e-05,
+      "loss": 0.0922,
+      "step": 12336
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 3.2707838479809974e-05,
+      "loss": 0.0445,
+      "step": 12337
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 3.2701900237529695e-05,
+      "loss": 0.0404,
+      "step": 12338
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 3.269596199524941e-05,
+      "loss": 0.0467,
+      "step": 12339
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 3.269002375296912e-05,
+      "loss": 0.0615,
+      "step": 12340
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 3.2684085510688835e-05,
+      "loss": 0.0855,
+      "step": 12341
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 3.2678147268408556e-05,
+      "loss": 0.0544,
+      "step": 12342
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 3.267220902612827e-05,
+      "loss": 0.0868,
+      "step": 12343
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 3.266627078384798e-05,
+      "loss": 0.0267,
+      "step": 12344
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 3.2660332541567697e-05,
+      "loss": 0.0345,
+      "step": 12345
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 3.265439429928741e-05,
+      "loss": 0.0622,
+      "step": 12346
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 3.2648456057007124e-05,
+      "loss": 0.0995,
+      "step": 12347
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 3.2642517814726844e-05,
+      "loss": 0.0948,
+      "step": 12348
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 3.263657957244656e-05,
+      "loss": 0.0751,
+      "step": 12349
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 3.263064133016627e-05,
+      "loss": 0.0763,
+      "step": 12350
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 3.262470308788599e-05,
+      "loss": 0.0515,
+      "step": 12351
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 3.2618764845605705e-05,
+      "loss": 0.0531,
+      "step": 12352
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 3.261282660332542e-05,
+      "loss": 0.0802,
+      "step": 12353
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 3.260688836104513e-05,
+      "loss": 0.0605,
+      "step": 12354
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 3.2600950118764846e-05,
+      "loss": 0.0458,
+      "step": 12355
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 3.259501187648456e-05,
+      "loss": 0.0646,
+      "step": 12356
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 3.258907363420427e-05,
+      "loss": 0.0725,
+      "step": 12357
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 3.2583135391923993e-05,
+      "loss": 0.0585,
+      "step": 12358
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 3.257719714964371e-05,
+      "loss": 0.0488,
+      "step": 12359
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 3.257125890736342e-05,
+      "loss": 0.0728,
+      "step": 12360
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 3.256532066508314e-05,
+      "loss": 0.0516,
+      "step": 12361
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 3.2559382422802855e-05,
+      "loss": 0.0751,
+      "step": 12362
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 3.255344418052257e-05,
+      "loss": 0.0396,
+      "step": 12363
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 3.254750593824228e-05,
+      "loss": 0.0847,
+      "step": 12364
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 3.2541567695961995e-05,
+      "loss": 0.0585,
+      "step": 12365
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 3.253562945368171e-05,
+      "loss": 0.0625,
+      "step": 12366
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 3.252969121140142e-05,
+      "loss": 0.0358,
+      "step": 12367
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 3.252375296912114e-05,
+      "loss": 0.0507,
+      "step": 12368
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 3.2517814726840856e-05,
+      "loss": 0.0628,
+      "step": 12369
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 3.251187648456057e-05,
+      "loss": 0.0781,
+      "step": 12370
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 3.250593824228029e-05,
+      "loss": 0.0502,
+      "step": 12371
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 0.0654,
+      "step": 12372
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 3.249406175771972e-05,
+      "loss": 0.0753,
+      "step": 12373
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 3.248812351543943e-05,
+      "loss": 0.0331,
+      "step": 12374
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 3.248218527315915e-05,
+      "loss": 0.0472,
+      "step": 12375
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 3.247624703087886e-05,
+      "loss": 0.0769,
+      "step": 12376
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 3.247030878859857e-05,
+      "loss": 0.0238,
+      "step": 12377
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 3.246437054631829e-05,
+      "loss": 0.0459,
+      "step": 12378
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 3.2458432304038006e-05,
+      "loss": 0.0982,
+      "step": 12379
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 3.245249406175772e-05,
+      "loss": 0.0528,
+      "step": 12380
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 3.244655581947744e-05,
+      "loss": 0.118,
+      "step": 12381
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 3.244061757719715e-05,
+      "loss": 0.0598,
+      "step": 12382
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 3.243467933491687e-05,
+      "loss": 0.0662,
+      "step": 12383
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 3.242874109263658e-05,
+      "loss": 0.0464,
+      "step": 12384
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 3.24228028503563e-05,
+      "loss": 0.0305,
+      "step": 12385
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 3.241686460807601e-05,
+      "loss": 0.0724,
+      "step": 12386
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 3.241092636579572e-05,
+      "loss": 0.0313,
+      "step": 12387
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 3.240498812351544e-05,
+      "loss": 0.0721,
+      "step": 12388
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 3.2399049881235155e-05,
+      "loss": 0.0509,
+      "step": 12389
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 3.239311163895487e-05,
+      "loss": 0.0698,
+      "step": 12390
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 3.238717339667459e-05,
+      "loss": 0.0384,
+      "step": 12391
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 3.23812351543943e-05,
+      "loss": 0.0373,
+      "step": 12392
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 3.2375296912114016e-05,
+      "loss": 0.1024,
+      "step": 12393
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 3.236935866983373e-05,
+      "loss": 0.0231,
+      "step": 12394
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 3.236342042755345e-05,
+      "loss": 0.0073,
+      "step": 12395
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 3.235748218527316e-05,
+      "loss": 0.0471,
+      "step": 12396
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 3.235154394299287e-05,
+      "loss": 0.1422,
+      "step": 12397
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 3.234560570071259e-05,
+      "loss": 0.1206,
+      "step": 12398
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 3.2339667458432304e-05,
+      "loss": 0.0672,
+      "step": 12399
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 3.233372921615202e-05,
+      "loss": 0.0869,
+      "step": 12400
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 3.232779097387174e-05,
+      "loss": 0.1132,
+      "step": 12401
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 3.232185273159145e-05,
+      "loss": 0.0361,
+      "step": 12402
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 3.2315914489311166e-05,
+      "loss": 0.0443,
+      "step": 12403
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 3.230997624703088e-05,
+      "loss": 0.057,
+      "step": 12404
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 3.23040380047506e-05,
+      "loss": 0.0525,
+      "step": 12405
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 3.229809976247031e-05,
+      "loss": 0.0842,
+      "step": 12406
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 3.229216152019002e-05,
+      "loss": 0.0773,
+      "step": 12407
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 3.228622327790974e-05,
+      "loss": 0.0759,
+      "step": 12408
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 3.2280285035629454e-05,
+      "loss": 0.0667,
+      "step": 12409
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 3.227434679334917e-05,
+      "loss": 0.038,
+      "step": 12410
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 3.226840855106889e-05,
+      "loss": 0.0568,
+      "step": 12411
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 3.22624703087886e-05,
+      "loss": 0.0711,
+      "step": 12412
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 3.2256532066508315e-05,
+      "loss": 0.1533,
+      "step": 12413
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 3.225059382422803e-05,
+      "loss": 0.0539,
+      "step": 12414
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 3.224465558194775e-05,
+      "loss": 0.0755,
+      "step": 12415
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 3.223871733966746e-05,
+      "loss": 0.0401,
+      "step": 12416
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 3.223277909738717e-05,
+      "loss": 0.058,
+      "step": 12417
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 3.222684085510689e-05,
+      "loss": 0.0503,
+      "step": 12418
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 3.22209026128266e-05,
+      "loss": 0.0731,
+      "step": 12419
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 3.221496437054632e-05,
+      "loss": 0.0447,
+      "step": 12420
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 3.220902612826604e-05,
+      "loss": 0.0418,
+      "step": 12421
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 3.220308788598575e-05,
+      "loss": 0.0898,
+      "step": 12422
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 3.2197149643705464e-05,
+      "loss": 0.0629,
+      "step": 12423
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 3.219121140142518e-05,
+      "loss": 0.0533,
+      "step": 12424
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 3.21852731591449e-05,
+      "loss": 0.0614,
+      "step": 12425
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 3.217933491686461e-05,
+      "loss": 0.1076,
+      "step": 12426
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 3.217339667458432e-05,
+      "loss": 0.0391,
+      "step": 12427
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 3.216745843230404e-05,
+      "loss": 0.0772,
+      "step": 12428
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 3.216152019002375e-05,
+      "loss": 0.0621,
+      "step": 12429
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 3.2155581947743466e-05,
+      "loss": 0.0359,
+      "step": 12430
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 3.2149643705463187e-05,
+      "loss": 0.0399,
+      "step": 12431
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 3.21437054631829e-05,
+      "loss": 0.0457,
+      "step": 12432
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 3.2137767220902614e-05,
+      "loss": 0.0275,
+      "step": 12433
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 3.213182897862233e-05,
+      "loss": 0.0326,
+      "step": 12434
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 3.212589073634205e-05,
+      "loss": 0.059,
+      "step": 12435
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 3.211995249406176e-05,
+      "loss": 0.0447,
+      "step": 12436
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 3.2114014251781475e-05,
+      "loss": 0.0425,
+      "step": 12437
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 3.210807600950119e-05,
+      "loss": 0.729,
+      "step": 12438
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 3.21021377672209e-05,
+      "loss": 0.0455,
+      "step": 12439
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 3.2096199524940616e-05,
+      "loss": 0.0311,
+      "step": 12440
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 3.2090261282660336e-05,
+      "loss": 0.0528,
+      "step": 12441
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 3.208432304038005e-05,
+      "loss": 0.0276,
+      "step": 12442
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 3.207838479809976e-05,
+      "loss": 0.077,
+      "step": 12443
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 3.2072446555819477e-05,
+      "loss": 0.0516,
+      "step": 12444
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 3.20665083135392e-05,
+      "loss": 0.0696,
+      "step": 12445
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 3.206057007125891e-05,
+      "loss": 0.0822,
+      "step": 12446
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 3.2054631828978624e-05,
+      "loss": 0.0821,
+      "step": 12447
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 3.204869358669834e-05,
+      "loss": 0.1001,
+      "step": 12448
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 3.204275534441805e-05,
+      "loss": 0.0864,
+      "step": 12449
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 3.2036817102137765e-05,
+      "loss": 0.0895,
+      "step": 12450
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 3.2030878859857485e-05,
+      "loss": 0.0583,
+      "step": 12451
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 3.20249406175772e-05,
+      "loss": 0.1027,
+      "step": 12452
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 3.201900237529691e-05,
+      "loss": 0.0908,
+      "step": 12453
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 3.2013064133016626e-05,
+      "loss": 0.0469,
+      "step": 12454
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 3.2007125890736346e-05,
+      "loss": 0.0574,
+      "step": 12455
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 3.200118764845606e-05,
+      "loss": 0.0917,
+      "step": 12456
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 3.1995249406175773e-05,
+      "loss": 0.0533,
+      "step": 12457
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 3.1989311163895494e-05,
+      "loss": 0.0609,
+      "step": 12458
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 3.19833729216152e-05,
+      "loss": 0.0633,
+      "step": 12459
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 3.1977434679334914e-05,
+      "loss": 0.0485,
+      "step": 12460
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 3.1971496437054635e-05,
+      "loss": 0.0654,
+      "step": 12461
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 3.196555819477435e-05,
+      "loss": 0.0277,
+      "step": 12462
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 3.195961995249406e-05,
+      "loss": 0.0492,
+      "step": 12463
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 3.1953681710213775e-05,
+      "loss": 0.0577,
+      "step": 12464
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 3.1947743467933496e-05,
+      "loss": 0.0647,
+      "step": 12465
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 3.194180522565321e-05,
+      "loss": 0.0702,
+      "step": 12466
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 3.193586698337292e-05,
+      "loss": 0.0408,
+      "step": 12467
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 3.192992874109264e-05,
+      "loss": 0.0476,
+      "step": 12468
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 3.192399049881235e-05,
+      "loss": 0.0663,
+      "step": 12469
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 3.1918052256532064e-05,
+      "loss": 0.0742,
+      "step": 12470
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 3.1912114014251784e-05,
+      "loss": 0.0638,
+      "step": 12471
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 3.19061757719715e-05,
+      "loss": 0.054,
+      "step": 12472
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 3.190023752969121e-05,
+      "loss": 0.0694,
+      "step": 12473
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 3.189429928741093e-05,
+      "loss": 0.0598,
+      "step": 12474
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 3.1888361045130645e-05,
+      "loss": 0.0571,
+      "step": 12475
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 3.188242280285036e-05,
+      "loss": 0.0293,
+      "step": 12476
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 3.187648456057007e-05,
+      "loss": 0.036,
+      "step": 12477
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 3.187054631828979e-05,
+      "loss": 0.0333,
+      "step": 12478
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 3.18646080760095e-05,
+      "loss": 0.0689,
+      "step": 12479
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 3.185866983372921e-05,
+      "loss": 0.046,
+      "step": 12480
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 3.185273159144893e-05,
+      "loss": 0.0471,
+      "step": 12481
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 3.184679334916865e-05,
+      "loss": 0.0786,
+      "step": 12482
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 3.184085510688836e-05,
+      "loss": 0.0297,
+      "step": 12483
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.183491686460808e-05,
+      "loss": 0.0287,
+      "step": 12484
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.1828978622327794e-05,
+      "loss": 0.0402,
+      "step": 12485
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.182304038004751e-05,
+      "loss": 0.093,
+      "step": 12486
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.181710213776722e-05,
+      "loss": 0.012,
+      "step": 12487
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.181116389548694e-05,
+      "loss": 0.0568,
+      "step": 12488
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.1805225653206655e-05,
+      "loss": 0.0725,
+      "step": 12489
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.179928741092636e-05,
+      "loss": 0.0816,
+      "step": 12490
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.179334916864608e-05,
+      "loss": 0.0498,
+      "step": 12491
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.1787410926365796e-05,
+      "loss": 0.0951,
+      "step": 12492
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.178147268408551e-05,
+      "loss": 0.0474,
+      "step": 12493
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.177553444180523e-05,
+      "loss": 0.0771,
+      "step": 12494
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.1769596199524944e-05,
+      "loss": 0.0709,
+      "step": 12495
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.176365795724466e-05,
+      "loss": 0.0472,
+      "step": 12496
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.175771971496437e-05,
+      "loss": 0.0602,
+      "step": 12497
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.175178147268409e-05,
+      "loss": 0.0467,
+      "step": 12498
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.1745843230403805e-05,
+      "loss": 0.0579,
+      "step": 12499
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.173990498812351e-05,
+      "loss": 0.0439,
+      "step": 12500
+    },
+    {
+      "epoch": 14.01,
+      "eval_loss": 1.8013914823532104,
+      "eval_runtime": 896.6971,
+      "eval_samples_per_second": 2.946,
+      "eval_steps_per_second": 0.369,
+      "eval_wer": 1.0022411741372472,
+      "step": 12500
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.173396674584323e-05,
+      "loss": 0.0633,
+      "step": 12501
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.1728028503562946e-05,
+      "loss": 0.0396,
+      "step": 12502
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.172209026128266e-05,
+      "loss": 0.0507,
+      "step": 12503
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.171615201900238e-05,
+      "loss": 0.0226,
+      "step": 12504
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.171021377672209e-05,
+      "loss": 0.0325,
+      "step": 12505
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.170427553444181e-05,
+      "loss": 0.0494,
+      "step": 12506
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.169833729216152e-05,
+      "loss": 0.0392,
+      "step": 12507
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.169239904988124e-05,
+      "loss": 0.0642,
+      "step": 12508
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.1686460807600954e-05,
+      "loss": 0.0539,
+      "step": 12509
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.168052256532066e-05,
+      "loss": 0.0442,
+      "step": 12510
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.167458432304038e-05,
+      "loss": 0.0521,
+      "step": 12511
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.1668646080760095e-05,
+      "loss": 0.0459,
+      "step": 12512
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.166270783847981e-05,
+      "loss": 0.0466,
+      "step": 12513
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.165676959619953e-05,
+      "loss": 0.0411,
+      "step": 12514
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.165083135391924e-05,
+      "loss": 0.0409,
+      "step": 12515
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.1644893111638956e-05,
+      "loss": 0.0344,
+      "step": 12516
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.163895486935867e-05,
+      "loss": 0.0564,
+      "step": 12517
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.163301662707839e-05,
+      "loss": 0.0322,
+      "step": 12518
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.1627078384798104e-05,
+      "loss": 0.037,
+      "step": 12519
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.162114014251782e-05,
+      "loss": 0.03,
+      "step": 12520
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.161520190023753e-05,
+      "loss": 0.0398,
+      "step": 12521
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.1609263657957244e-05,
+      "loss": 0.032,
+      "step": 12522
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.160332541567696e-05,
+      "loss": 0.0311,
+      "step": 12523
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.159738717339668e-05,
+      "loss": 0.2277,
+      "step": 12524
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.159144893111639e-05,
+      "loss": 0.0242,
+      "step": 12525
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.1585510688836105e-05,
+      "loss": 0.0318,
+      "step": 12526
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.157957244655582e-05,
+      "loss": 0.0337,
+      "step": 12527
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.157363420427554e-05,
+      "loss": 0.057,
+      "step": 12528
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.156769596199525e-05,
+      "loss": 0.0357,
+      "step": 12529
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.1561757719714967e-05,
+      "loss": 0.0848,
+      "step": 12530
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.155581947743468e-05,
+      "loss": 0.0201,
+      "step": 12531
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.1549881235154394e-05,
+      "loss": 0.0552,
+      "step": 12532
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.154394299287411e-05,
+      "loss": 0.0345,
+      "step": 12533
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.153800475059383e-05,
+      "loss": 0.0174,
+      "step": 12534
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.153206650831354e-05,
+      "loss": 0.0457,
+      "step": 12535
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.1526128266033255e-05,
+      "loss": 0.0342,
+      "step": 12536
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.152019002375297e-05,
+      "loss": 0.0447,
+      "step": 12537
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.151425178147269e-05,
+      "loss": 0.0087,
+      "step": 12538
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.15083135391924e-05,
+      "loss": 0.1443,
+      "step": 12539
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.1502375296912116e-05,
+      "loss": 0.0541,
+      "step": 12540
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.1496437054631836e-05,
+      "loss": 0.0923,
+      "step": 12541
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.149049881235154e-05,
+      "loss": 0.0585,
+      "step": 12542
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.148456057007126e-05,
+      "loss": 0.0509,
+      "step": 12543
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.147862232779098e-05,
+      "loss": 0.0685,
+      "step": 12544
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.147268408551069e-05,
+      "loss": 0.044,
+      "step": 12545
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.1466745843230404e-05,
+      "loss": 0.0559,
+      "step": 12546
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.146080760095012e-05,
+      "loss": 0.0674,
+      "step": 12547
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.145486935866984e-05,
+      "loss": 0.05,
+      "step": 12548
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.144893111638955e-05,
+      "loss": 0.0389,
+      "step": 12549
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.1442992874109265e-05,
+      "loss": 0.0578,
+      "step": 12550
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.1437054631828986e-05,
+      "loss": 0.0438,
+      "step": 12551
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.143111638954869e-05,
+      "loss": 0.0588,
+      "step": 12552
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.1425178147268406e-05,
+      "loss": 0.0408,
+      "step": 12553
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.1419239904988126e-05,
+      "loss": 0.0447,
+      "step": 12554
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.141330166270784e-05,
+      "loss": 0.0532,
+      "step": 12555
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.1407363420427553e-05,
+      "loss": 0.0504,
+      "step": 12556
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.140142517814727e-05,
+      "loss": 0.0426,
+      "step": 12557
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.139548693586699e-05,
+      "loss": 0.0602,
+      "step": 12558
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.13895486935867e-05,
+      "loss": 0.0671,
+      "step": 12559
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.1383610451306415e-05,
+      "loss": 0.025,
+      "step": 12560
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.1377672209026135e-05,
+      "loss": 0.0386,
+      "step": 12561
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.137173396674584e-05,
+      "loss": 0.0747,
+      "step": 12562
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.1365795724465555e-05,
+      "loss": 0.0428,
+      "step": 12563
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.1359857482185276e-05,
+      "loss": 0.0388,
+      "step": 12564
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.135391923990499e-05,
+      "loss": 0.0829,
+      "step": 12565
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.13479809976247e-05,
+      "loss": 0.0492,
+      "step": 12566
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.1342042755344416e-05,
+      "loss": 0.0638,
+      "step": 12567
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.133610451306414e-05,
+      "loss": 0.0446,
+      "step": 12568
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.133016627078385e-05,
+      "loss": 0.0464,
+      "step": 12569
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.1324228028503564e-05,
+      "loss": 0.0478,
+      "step": 12570
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.1318289786223284e-05,
+      "loss": 0.0646,
+      "step": 12571
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.1312351543943e-05,
+      "loss": 0.0519,
+      "step": 12572
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.1306413301662705e-05,
+      "loss": 0.0466,
+      "step": 12573
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.1300475059382425e-05,
+      "loss": 0.0335,
+      "step": 12574
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.129453681710214e-05,
+      "loss": 0.019,
+      "step": 12575
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.128859857482185e-05,
+      "loss": 0.031,
+      "step": 12576
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.1282660332541566e-05,
+      "loss": 0.0289,
+      "step": 12577
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.1276722090261286e-05,
+      "loss": 0.0266,
+      "step": 12578
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.1270783847981e-05,
+      "loss": 0.058,
+      "step": 12579
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.126484560570071e-05,
+      "loss": 0.066,
+      "step": 12580
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.1258907363420434e-05,
+      "loss": 0.0168,
+      "step": 12581
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.125296912114015e-05,
+      "loss": 0.031,
+      "step": 12582
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.1247030878859854e-05,
+      "loss": 0.0224,
+      "step": 12583
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.1241092636579574e-05,
+      "loss": 0.0326,
+      "step": 12584
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.123515439429929e-05,
+      "loss": 0.0362,
+      "step": 12585
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.1229216152019e-05,
+      "loss": 0.0095,
+      "step": 12586
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.1223277909738715e-05,
+      "loss": 0.104,
+      "step": 12587
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.1217339667458436e-05,
+      "loss": 0.0818,
+      "step": 12588
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.121140142517815e-05,
+      "loss": 0.1028,
+      "step": 12589
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.120546318289786e-05,
+      "loss": 0.0401,
+      "step": 12590
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.119952494061758e-05,
+      "loss": 0.074,
+      "step": 12591
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.1193586698337297e-05,
+      "loss": 0.064,
+      "step": 12592
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.1187648456057003e-05,
+      "loss": 0.0337,
+      "step": 12593
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.1181710213776724e-05,
+      "loss": 0.0498,
+      "step": 12594
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.117577197149644e-05,
+      "loss": 0.0608,
+      "step": 12595
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.116983372921615e-05,
+      "loss": 0.0341,
+      "step": 12596
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.116389548693587e-05,
+      "loss": 0.0371,
+      "step": 12597
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.1157957244655585e-05,
+      "loss": 0.0634,
+      "step": 12598
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.11520190023753e-05,
+      "loss": 0.053,
+      "step": 12599
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.114608076009501e-05,
+      "loss": 0.0673,
+      "step": 12600
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.114014251781473e-05,
+      "loss": 0.0547,
+      "step": 12601
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.1134204275534446e-05,
+      "loss": 0.0472,
+      "step": 12602
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.112826603325416e-05,
+      "loss": 0.0326,
+      "step": 12603
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.112232779097387e-05,
+      "loss": 0.0374,
+      "step": 12604
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.111638954869359e-05,
+      "loss": 0.0476,
+      "step": 12605
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.11104513064133e-05,
+      "loss": 0.0487,
+      "step": 12606
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.110451306413302e-05,
+      "loss": 0.0394,
+      "step": 12607
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.1098574821852734e-05,
+      "loss": 0.0388,
+      "step": 12608
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.109263657957245e-05,
+      "loss": 0.0302,
+      "step": 12609
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.108669833729216e-05,
+      "loss": 0.0782,
+      "step": 12610
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.108076009501188e-05,
+      "loss": 0.0468,
+      "step": 12611
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.1074821852731595e-05,
+      "loss": 0.1268,
+      "step": 12612
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.106888361045131e-05,
+      "loss": 0.0643,
+      "step": 12613
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.106294536817102e-05,
+      "loss": 0.034,
+      "step": 12614
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.1057007125890736e-05,
+      "loss": 0.0688,
+      "step": 12615
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.105106888361045e-05,
+      "loss": 0.0496,
+      "step": 12616
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.104513064133017e-05,
+      "loss": 0.0253,
+      "step": 12617
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.1039192399049884e-05,
+      "loss": 0.0415,
+      "step": 12618
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.10332541567696e-05,
+      "loss": 0.0398,
+      "step": 12619
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.102731591448931e-05,
+      "loss": 0.0519,
+      "step": 12620
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.102137767220903e-05,
+      "loss": 0.0441,
+      "step": 12621
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.1015439429928745e-05,
+      "loss": 0.0735,
+      "step": 12622
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.100950118764846e-05,
+      "loss": 0.0374,
+      "step": 12623
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.100356294536817e-05,
+      "loss": 0.0339,
+      "step": 12624
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.0997624703087885e-05,
+      "loss": 0.0843,
+      "step": 12625
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.09916864608076e-05,
+      "loss": 0.0572,
+      "step": 12626
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.098574821852732e-05,
+      "loss": 0.0656,
+      "step": 12627
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.097980997624703e-05,
+      "loss": 0.0496,
+      "step": 12628
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.0973871733966747e-05,
+      "loss": 0.0269,
+      "step": 12629
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.096793349168646e-05,
+      "loss": 0.0369,
+      "step": 12630
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.096199524940618e-05,
+      "loss": 0.0541,
+      "step": 12631
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.0956057007125894e-05,
+      "loss": 0.0206,
+      "step": 12632
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.095011876484561e-05,
+      "loss": 0.0962,
+      "step": 12633
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.094418052256533e-05,
+      "loss": 0.0528,
+      "step": 12634
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.0938242280285035e-05,
+      "loss": 0.04,
+      "step": 12635
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.093230403800475e-05,
+      "loss": 0.021,
+      "step": 12636
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.092636579572447e-05,
+      "loss": 0.0747,
+      "step": 12637
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.092042755344418e-05,
+      "loss": 0.1064,
+      "step": 12638
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.0914489311163896e-05,
+      "loss": 0.09,
+      "step": 12639
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.090855106888361e-05,
+      "loss": 0.0509,
+      "step": 12640
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.090261282660333e-05,
+      "loss": 0.0588,
+      "step": 12641
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.0896674584323043e-05,
+      "loss": 0.0624,
+      "step": 12642
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.089073634204276e-05,
+      "loss": 0.0861,
+      "step": 12643
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.088479809976248e-05,
+      "loss": 0.0334,
+      "step": 12644
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.0878859857482184e-05,
+      "loss": 0.0403,
+      "step": 12645
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.08729216152019e-05,
+      "loss": 0.078,
+      "step": 12646
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.086698337292162e-05,
+      "loss": 0.0281,
+      "step": 12647
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.086104513064133e-05,
+      "loss": 0.0548,
+      "step": 12648
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.0855106888361045e-05,
+      "loss": 0.0659,
+      "step": 12649
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.084916864608076e-05,
+      "loss": 0.0377,
+      "step": 12650
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.084323040380048e-05,
+      "loss": 0.0885,
+      "step": 12651
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.083729216152019e-05,
+      "loss": 0.0522,
+      "step": 12652
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.0831353919239906e-05,
+      "loss": 0.0354,
+      "step": 12653
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.082541567695963e-05,
+      "loss": 0.0621,
+      "step": 12654
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.081947743467934e-05,
+      "loss": 0.0462,
+      "step": 12655
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.081353919239905e-05,
+      "loss": 0.033,
+      "step": 12656
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.080760095011877e-05,
+      "loss": 0.0508,
+      "step": 12657
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.080166270783848e-05,
+      "loss": 0.0579,
+      "step": 12658
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.0795724465558195e-05,
+      "loss": 0.0349,
+      "step": 12659
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.078978622327791e-05,
+      "loss": 0.0503,
+      "step": 12660
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.078384798099763e-05,
+      "loss": 0.0403,
+      "step": 12661
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.077790973871734e-05,
+      "loss": 0.0613,
+      "step": 12662
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.0771971496437056e-05,
+      "loss": 0.0619,
+      "step": 12663
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.0766033254156776e-05,
+      "loss": 0.0613,
+      "step": 12664
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.076009501187649e-05,
+      "loss": 0.0277,
+      "step": 12665
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.0754156769596196e-05,
+      "loss": 0.0351,
+      "step": 12666
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.074821852731592e-05,
+      "loss": 0.0499,
+      "step": 12667
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.074228028503563e-05,
+      "loss": 0.0362,
+      "step": 12668
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.0736342042755344e-05,
+      "loss": 0.0458,
+      "step": 12669
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.073040380047506e-05,
+      "loss": 0.0575,
+      "step": 12670
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.072446555819478e-05,
+      "loss": 0.0596,
+      "step": 12671
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.071852731591449e-05,
+      "loss": 0.0678,
+      "step": 12672
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.0712589073634205e-05,
+      "loss": 0.0452,
+      "step": 12673
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.0706650831353925e-05,
+      "loss": 0.0829,
+      "step": 12674
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.070071258907364e-05,
+      "loss": 0.0497,
+      "step": 12675
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.0694774346793346e-05,
+      "loss": 0.0521,
+      "step": 12676
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.0688836104513066e-05,
+      "loss": 0.0371,
+      "step": 12677
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.068289786223278e-05,
+      "loss": 0.0422,
+      "step": 12678
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.067695961995249e-05,
+      "loss": 0.0353,
+      "step": 12679
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.067102137767221e-05,
+      "loss": 0.0334,
+      "step": 12680
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.066508313539193e-05,
+      "loss": 0.0258,
+      "step": 12681
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.065914489311164e-05,
+      "loss": 0.0267,
+      "step": 12682
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.0653206650831354e-05,
+      "loss": 0.0483,
+      "step": 12683
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.0647268408551075e-05,
+      "loss": 0.0323,
+      "step": 12684
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.064133016627079e-05,
+      "loss": 0.0185,
+      "step": 12685
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.06353919239905e-05,
+      "loss": 0.0139,
+      "step": 12686
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.0629453681710216e-05,
+      "loss": 0.042,
+      "step": 12687
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.062351543942993e-05,
+      "loss": 0.0676,
+      "step": 12688
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.061757719714964e-05,
+      "loss": 0.0943,
+      "step": 12689
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.0611638954869356e-05,
+      "loss": 0.0725,
+      "step": 12690
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.060570071258908e-05,
+      "loss": 0.105,
+      "step": 12691
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.059976247030879e-05,
+      "loss": 0.0465,
+      "step": 12692
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.0593824228028504e-05,
+      "loss": 0.0533,
+      "step": 12693
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.0587885985748224e-05,
+      "loss": 0.0529,
+      "step": 12694
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.058194774346794e-05,
+      "loss": 0.0736,
+      "step": 12695
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.057600950118765e-05,
+      "loss": 0.0442,
+      "step": 12696
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.0570071258907365e-05,
+      "loss": 0.0915,
+      "step": 12697
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.056413301662708e-05,
+      "loss": 0.0591,
+      "step": 12698
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.055819477434679e-05,
+      "loss": 0.0405,
+      "step": 12699
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.0552256532066506e-05,
+      "loss": 0.071,
+      "step": 12700
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.0546318289786226e-05,
+      "loss": 0.0748,
+      "step": 12701
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.054038004750594e-05,
+      "loss": 0.0562,
+      "step": 12702
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.053444180522565e-05,
+      "loss": 0.0325,
+      "step": 12703
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.0528503562945373e-05,
+      "loss": 0.0446,
+      "step": 12704
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.052256532066509e-05,
+      "loss": 0.0377,
+      "step": 12705
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.05166270783848e-05,
+      "loss": 0.0411,
+      "step": 12706
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.051068883610451e-05,
+      "loss": 0.0672,
+      "step": 12707
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.0504750593824228e-05,
+      "loss": 0.05,
+      "step": 12708
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.049881235154394e-05,
+      "loss": 0.0392,
+      "step": 12709
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.049287410926366e-05,
+      "loss": 0.0458,
+      "step": 12710
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.0486935866983375e-05,
+      "loss": 0.0417,
+      "step": 12711
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.048099762470309e-05,
+      "loss": 0.0503,
+      "step": 12712
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.0475059382422806e-05,
+      "loss": 0.055,
+      "step": 12713
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.046912114014252e-05,
+      "loss": 0.0504,
+      "step": 12714
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.0463182897862236e-05,
+      "loss": 0.0577,
+      "step": 12715
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.045724465558195e-05,
+      "loss": 0.0758,
+      "step": 12716
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.0451306413301667e-05,
+      "loss": 0.0297,
+      "step": 12717
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.0445368171021377e-05,
+      "loss": 0.1518,
+      "step": 12718
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.043942992874109e-05,
+      "loss": 0.0331,
+      "step": 12719
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.0433491686460808e-05,
+      "loss": 0.0457,
+      "step": 12720
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.0427553444180525e-05,
+      "loss": 0.0391,
+      "step": 12721
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.0421615201900238e-05,
+      "loss": 0.0798,
+      "step": 12722
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.0415676959619955e-05,
+      "loss": 0.0215,
+      "step": 12723
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.040973871733967e-05,
+      "loss": 0.06,
+      "step": 12724
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.0403800475059386e-05,
+      "loss": 0.0265,
+      "step": 12725
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.03978622327791e-05,
+      "loss": 0.0911,
+      "step": 12726
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.0391923990498816e-05,
+      "loss": 0.028,
+      "step": 12727
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.0385985748218527e-05,
+      "loss": 0.0259,
+      "step": 12728
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.038004750593824e-05,
+      "loss": 0.0883,
+      "step": 12729
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.0374109263657957e-05,
+      "loss": 0.043,
+      "step": 12730
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.0368171021377674e-05,
+      "loss": 0.0204,
+      "step": 12731
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.0362232779097388e-05,
+      "loss": 0.0738,
+      "step": 12732
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.0356294536817105e-05,
+      "loss": 0.0456,
+      "step": 12733
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.0350356294536818e-05,
+      "loss": 0.0348,
+      "step": 12734
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.0344418052256535e-05,
+      "loss": 0.0317,
+      "step": 12735
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.033847980997625e-05,
+      "loss": 0.0129,
+      "step": 12736
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.0332541567695966e-05,
+      "loss": 0.0342,
+      "step": 12737
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.0326603325415683e-05,
+      "loss": 0.0106,
+      "step": 12738
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.032066508313539e-05,
+      "loss": 0.0687,
+      "step": 12739
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.0314726840855106e-05,
+      "loss": 0.0744,
+      "step": 12740
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.0308788598574823e-05,
+      "loss": 0.0645,
+      "step": 12741
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.0302850356294537e-05,
+      "loss": 0.0779,
+      "step": 12742
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.0296912114014254e-05,
+      "loss": 0.0981,
+      "step": 12743
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.0290973871733968e-05,
+      "loss": 0.0817,
+      "step": 12744
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.0285035629453685e-05,
+      "loss": 0.0534,
+      "step": 12745
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.0279097387173398e-05,
+      "loss": 0.0589,
+      "step": 12746
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.0273159144893115e-05,
+      "loss": 0.0321,
+      "step": 12747
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.0267220902612832e-05,
+      "loss": 0.0315,
+      "step": 12748
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.0261282660332542e-05,
+      "loss": 0.0507,
+      "step": 12749
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.0255344418052256e-05,
+      "loss": 0.0372,
+      "step": 12750
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.0249406175771973e-05,
+      "loss": 0.0633,
+      "step": 12751
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.0243467933491686e-05,
+      "loss": 0.0816,
+      "step": 12752
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.0237529691211403e-05,
+      "loss": 0.0413,
+      "step": 12753
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.0231591448931117e-05,
+      "loss": 0.032,
+      "step": 12754
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.0225653206650834e-05,
+      "loss": 0.0324,
+      "step": 12755
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.0219714964370547e-05,
+      "loss": 0.0434,
+      "step": 12756
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.0213776722090264e-05,
+      "loss": 0.041,
+      "step": 12757
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.020783847980998e-05,
+      "loss": 0.0506,
+      "step": 12758
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.020190023752969e-05,
+      "loss": 0.028,
+      "step": 12759
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.0195961995249405e-05,
+      "loss": 0.0492,
+      "step": 12760
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.0190023752969122e-05,
+      "loss": 0.0503,
+      "step": 12761
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.0184085510688836e-05,
+      "loss": 0.0618,
+      "step": 12762
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.0178147268408553e-05,
+      "loss": 0.049,
+      "step": 12763
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.0172209026128266e-05,
+      "loss": 0.0417,
+      "step": 12764
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.0166270783847983e-05,
+      "loss": 0.0604,
+      "step": 12765
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.0160332541567697e-05,
+      "loss": 0.0835,
+      "step": 12766
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.0154394299287414e-05,
+      "loss": 0.0263,
+      "step": 12767
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.014845605700713e-05,
+      "loss": 0.0561,
+      "step": 12768
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.0142517814726844e-05,
+      "loss": 0.0495,
+      "step": 12769
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.0136579572446555e-05,
+      "loss": 0.0393,
+      "step": 12770
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.013064133016627e-05,
+      "loss": 0.0298,
+      "step": 12771
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.0124703087885985e-05,
+      "loss": 0.035,
+      "step": 12772
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.0118764845605702e-05,
+      "loss": 0.0608,
+      "step": 12773
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.0112826603325416e-05,
+      "loss": 0.0733,
+      "step": 12774
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.0106888361045133e-05,
+      "loss": 0.0333,
+      "step": 12775
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.010095011876485e-05,
+      "loss": 0.0479,
+      "step": 12776
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.0095011876484563e-05,
+      "loss": 0.0358,
+      "step": 12777
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.008907363420428e-05,
+      "loss": 0.02,
+      "step": 12778
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.0083135391923994e-05,
+      "loss": 0.0321,
+      "step": 12779
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.0077197149643704e-05,
+      "loss": 0.0616,
+      "step": 12780
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.007125890736342e-05,
+      "loss": 0.0694,
+      "step": 12781
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.0065320665083134e-05,
+      "loss": 0.0632,
+      "step": 12782
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.005938242280285e-05,
+      "loss": 0.0355,
+      "step": 12783
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.005938242280285e-05,
+      "loss": 0.0832,
+      "step": 12784
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.0053444180522565e-05,
+      "loss": 0.0289,
+      "step": 12785
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.0047505938242282e-05,
+      "loss": 0.0344,
+      "step": 12786
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.0041567695962e-05,
+      "loss": 0.0574,
+      "step": 12787
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.0035629453681712e-05,
+      "loss": 0.0251,
+      "step": 12788
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.002969121140143e-05,
+      "loss": 0.0813,
+      "step": 12789
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.0023752969121143e-05,
+      "loss": 0.085,
+      "step": 12790
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.0017814726840853e-05,
+      "loss": 0.0326,
+      "step": 12791
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.001187648456057e-05,
+      "loss": 0.0612,
+      "step": 12792
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.0005938242280284e-05,
+      "loss": 0.0534,
+      "step": 12793
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3e-05,
+      "loss": 0.0544,
+      "step": 12794
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 2.9994061757719714e-05,
+      "loss": 0.0402,
+      "step": 12795
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 2.998812351543943e-05,
+      "loss": 0.0562,
+      "step": 12796
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 2.9982185273159148e-05,
+      "loss": 0.0643,
+      "step": 12797
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 2.9976247030878862e-05,
+      "loss": 0.0486,
+      "step": 12798
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 2.997030878859858e-05,
+      "loss": 0.0466,
+      "step": 12799
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 2.9964370546318292e-05,
+      "loss": 0.0427,
+      "step": 12800
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 2.995843230403801e-05,
+      "loss": 0.0347,
+      "step": 12801
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 2.995249406175772e-05,
+      "loss": 0.0632,
+      "step": 12802
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 2.9946555819477433e-05,
+      "loss": 0.0542,
+      "step": 12803
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 2.994061757719715e-05,
+      "loss": 0.0346,
+      "step": 12804
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 2.9934679334916864e-05,
+      "loss": 0.0747,
+      "step": 12805
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 2.992874109263658e-05,
+      "loss": 0.0537,
+      "step": 12806
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 2.9922802850356298e-05,
+      "loss": 0.0376,
+      "step": 12807
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 2.991686460807601e-05,
+      "loss": 0.0561,
+      "step": 12808
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 2.9910926365795728e-05,
+      "loss": 0.0603,
+      "step": 12809
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 2.9904988123515442e-05,
+      "loss": 0.052,
+      "step": 12810
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 2.989904988123516e-05,
+      "loss": 0.0483,
+      "step": 12811
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 2.989311163895487e-05,
+      "loss": 0.0439,
+      "step": 12812
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 2.9887173396674583e-05,
+      "loss": 0.0651,
+      "step": 12813
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 2.98812351543943e-05,
+      "loss": 0.075,
+      "step": 12814
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 2.9875296912114013e-05,
+      "loss": 0.055,
+      "step": 12815
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 2.986935866983373e-05,
+      "loss": 0.0183,
+      "step": 12816
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 2.9863420427553447e-05,
+      "loss": 0.1331,
+      "step": 12817
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 2.985748218527316e-05,
+      "loss": 0.0443,
+      "step": 12818
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 2.9851543942992878e-05,
+      "loss": 0.0337,
+      "step": 12819
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 2.984560570071259e-05,
+      "loss": 0.0474,
+      "step": 12820
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 2.9839667458432308e-05,
+      "loss": 0.0418,
+      "step": 12821
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 2.9833729216152018e-05,
+      "loss": 0.0663,
+      "step": 12822
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 2.9827790973871732e-05,
+      "loss": 0.0734,
+      "step": 12823
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 2.982185273159145e-05,
+      "loss": 0.0954,
+      "step": 12824
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 2.9815914489311166e-05,
+      "loss": 0.0396,
+      "step": 12825
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 2.980997624703088e-05,
+      "loss": 0.0328,
+      "step": 12826
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 2.9804038004750596e-05,
+      "loss": 0.0565,
+      "step": 12827
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 2.979809976247031e-05,
+      "loss": 0.0344,
+      "step": 12828
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 2.9792161520190027e-05,
+      "loss": 0.0424,
+      "step": 12829
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 2.978622327790974e-05,
+      "loss": 0.0372,
+      "step": 12830
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 2.9780285035629457e-05,
+      "loss": 0.0302,
+      "step": 12831
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 2.977434679334917e-05,
+      "loss": 0.0751,
+      "step": 12832
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 2.976840855106888e-05,
+      "loss": 0.0311,
+      "step": 12833
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 2.9762470308788598e-05,
+      "loss": 0.0478,
+      "step": 12834
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 2.9756532066508315e-05,
+      "loss": 0.0296,
+      "step": 12835
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 2.975059382422803e-05,
+      "loss": 0.041,
+      "step": 12836
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 2.9744655581947746e-05,
+      "loss": 0.0632,
+      "step": 12837
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 2.973871733966746e-05,
+      "loss": 0.0472,
+      "step": 12838
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 2.9732779097387176e-05,
+      "loss": 0.1118,
+      "step": 12839
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 2.972684085510689e-05,
+      "loss": 0.0652,
+      "step": 12840
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 2.9720902612826607e-05,
+      "loss": 0.0543,
+      "step": 12841
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 2.9714964370546324e-05,
+      "loss": 0.0669,
+      "step": 12842
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 2.970902612826603e-05,
+      "loss": 0.0929,
+      "step": 12843
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 2.9703087885985748e-05,
+      "loss": 0.0659,
+      "step": 12844
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 2.9697149643705465e-05,
+      "loss": 0.04,
+      "step": 12845
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 2.9691211401425178e-05,
+      "loss": 0.052,
+      "step": 12846
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 2.9685273159144895e-05,
+      "loss": 0.0685,
+      "step": 12847
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 2.967933491686461e-05,
+      "loss": 0.043,
+      "step": 12848
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 2.9673396674584326e-05,
+      "loss": 0.0504,
+      "step": 12849
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 2.966745843230404e-05,
+      "loss": 0.044,
+      "step": 12850
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 2.9661520190023756e-05,
+      "loss": 0.059,
+      "step": 12851
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 2.9655581947743473e-05,
+      "loss": 0.0641,
+      "step": 12852
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 2.9649643705463187e-05,
+      "loss": 0.042,
+      "step": 12853
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 2.9643705463182897e-05,
+      "loss": 0.07,
+      "step": 12854
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 2.9637767220902614e-05,
+      "loss": 0.4609,
+      "step": 12855
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 2.9631828978622327e-05,
+      "loss": 0.0781,
+      "step": 12856
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 2.9625890736342044e-05,
+      "loss": 0.0426,
+      "step": 12857
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 2.9619952494061758e-05,
+      "loss": 0.0308,
+      "step": 12858
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 2.9614014251781475e-05,
+      "loss": 0.0569,
+      "step": 12859
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 2.960807600950119e-05,
+      "loss": 0.0571,
+      "step": 12860
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 2.9602137767220906e-05,
+      "loss": 0.0255,
+      "step": 12861
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 2.9596199524940623e-05,
+      "loss": 0.0492,
+      "step": 12862
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 2.9590261282660336e-05,
+      "loss": 0.038,
+      "step": 12863
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 2.9584323040380046e-05,
+      "loss": 0.0316,
+      "step": 12864
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 2.9578384798099763e-05,
+      "loss": 0.0964,
+      "step": 12865
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 2.9572446555819477e-05,
+      "loss": 0.06,
+      "step": 12866
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 2.9566508313539194e-05,
+      "loss": 0.0419,
+      "step": 12867
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 2.9560570071258907e-05,
+      "loss": 0.0835,
+      "step": 12868
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 2.9554631828978624e-05,
+      "loss": 0.0761,
+      "step": 12869
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 2.9548693586698338e-05,
+      "loss": 0.0537,
+      "step": 12870
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 2.9542755344418055e-05,
+      "loss": 0.038,
+      "step": 12871
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 2.9536817102137772e-05,
+      "loss": 0.0297,
+      "step": 12872
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 2.9530878859857485e-05,
+      "loss": 0.0477,
+      "step": 12873
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 2.9524940617577196e-05,
+      "loss": 0.0361,
+      "step": 12874
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 2.9519002375296913e-05,
+      "loss": 0.0654,
+      "step": 12875
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 2.9513064133016626e-05,
+      "loss": 0.053,
+      "step": 12876
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 2.9507125890736343e-05,
+      "loss": 0.0372,
+      "step": 12877
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 2.9501187648456057e-05,
+      "loss": 0.031,
+      "step": 12878
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 2.9495249406175774e-05,
+      "loss": 0.0417,
+      "step": 12879
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 2.9489311163895487e-05,
+      "loss": 0.0377,
+      "step": 12880
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 2.9483372921615204e-05,
+      "loss": 0.0413,
+      "step": 12881
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 2.947743467933492e-05,
+      "loss": 0.0804,
+      "step": 12882
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 2.9471496437054635e-05,
+      "loss": 0.0429,
+      "step": 12883
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 2.9465558194774352e-05,
+      "loss": 0.0524,
+      "step": 12884
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 2.9459619952494062e-05,
+      "loss": 0.0206,
+      "step": 12885
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 2.9453681710213776e-05,
+      "loss": 0.0554,
+      "step": 12886
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 2.9447743467933493e-05,
+      "loss": 0.0972,
+      "step": 12887
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 2.9441805225653206e-05,
+      "loss": 0.0265,
+      "step": 12888
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 2.9435866983372923e-05,
+      "loss": 0.073,
+      "step": 12889
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 2.9429928741092637e-05,
+      "loss": 0.0977,
+      "step": 12890
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 2.9423990498812354e-05,
+      "loss": 0.0796,
+      "step": 12891
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 2.941805225653207e-05,
+      "loss": 0.0825,
+      "step": 12892
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 2.9412114014251784e-05,
+      "loss": 0.0808,
+      "step": 12893
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 2.94061757719715e-05,
+      "loss": 0.0769,
+      "step": 12894
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 2.940023752969121e-05,
+      "loss": 0.049,
+      "step": 12895
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 2.9394299287410925e-05,
+      "loss": 0.0721,
+      "step": 12896
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 2.9388361045130642e-05,
+      "loss": 0.1359,
+      "step": 12897
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 2.9382422802850355e-05,
+      "loss": 0.1003,
+      "step": 12898
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 2.9376484560570072e-05,
+      "loss": 0.0676,
+      "step": 12899
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 2.937054631828979e-05,
+      "loss": 0.0406,
+      "step": 12900
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 2.9364608076009503e-05,
+      "loss": 0.0469,
+      "step": 12901
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 2.935866983372922e-05,
+      "loss": 0.0558,
+      "step": 12902
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 2.9352731591448934e-05,
+      "loss": 0.0446,
+      "step": 12903
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 2.934679334916865e-05,
+      "loss": 0.0588,
+      "step": 12904
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 2.934085510688836e-05,
+      "loss": 0.0702,
+      "step": 12905
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 2.9334916864608074e-05,
+      "loss": 0.0432,
+      "step": 12906
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 2.932897862232779e-05,
+      "loss": 0.0551,
+      "step": 12907
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 2.9323040380047505e-05,
+      "loss": 0.0602,
+      "step": 12908
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 2.9317102137767222e-05,
+      "loss": 0.074,
+      "step": 12909
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 2.931116389548694e-05,
+      "loss": 0.0678,
+      "step": 12910
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 2.9305225653206652e-05,
+      "loss": 0.0296,
+      "step": 12911
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 2.929928741092637e-05,
+      "loss": 0.0684,
+      "step": 12912
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 2.9293349168646083e-05,
+      "loss": 0.0725,
+      "step": 12913
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 2.92874109263658e-05,
+      "loss": 0.0499,
+      "step": 12914
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 2.9281472684085513e-05,
+      "loss": 0.0556,
+      "step": 12915
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 2.9275534441805224e-05,
+      "loss": 0.0542,
+      "step": 12916
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 2.926959619952494e-05,
+      "loss": 0.0289,
+      "step": 12917
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 2.9263657957244654e-05,
+      "loss": 0.0777,
+      "step": 12918
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 2.925771971496437e-05,
+      "loss": 0.0515,
+      "step": 12919
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 2.9251781472684088e-05,
+      "loss": 0.0626,
+      "step": 12920
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 2.92458432304038e-05,
+      "loss": 0.0303,
+      "step": 12921
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 2.923990498812352e-05,
+      "loss": 0.028,
+      "step": 12922
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 2.9233966745843232e-05,
+      "loss": 0.042,
+      "step": 12923
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 2.922802850356295e-05,
+      "loss": 0.0455,
+      "step": 12924
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 2.9222090261282663e-05,
+      "loss": 0.1844,
+      "step": 12925
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 2.9216152019002373e-05,
+      "loss": 0.0316,
+      "step": 12926
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 2.921021377672209e-05,
+      "loss": 0.0418,
+      "step": 12927
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 2.9204275534441804e-05,
+      "loss": 0.0722,
+      "step": 12928
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 2.919833729216152e-05,
+      "loss": 0.0179,
+      "step": 12929
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 2.9192399049881237e-05,
+      "loss": 0.0261,
+      "step": 12930
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 2.918646080760095e-05,
+      "loss": 0.0214,
+      "step": 12931
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 2.9180522565320668e-05,
+      "loss": 0.0472,
+      "step": 12932
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 2.917458432304038e-05,
+      "loss": 0.0175,
+      "step": 12933
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 2.91686460807601e-05,
+      "loss": 0.0154,
+      "step": 12934
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 2.9162707838479812e-05,
+      "loss": 0.0314,
+      "step": 12935
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 2.915676959619953e-05,
+      "loss": 0.0398,
+      "step": 12936
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 2.915083135391924e-05,
+      "loss": 0.0471,
+      "step": 12937
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 2.9144893111638953e-05,
+      "loss": 0.0485,
+      "step": 12938
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 2.913895486935867e-05,
+      "loss": 0.0674,
+      "step": 12939
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 2.9133016627078387e-05,
+      "loss": 0.1205,
+      "step": 12940
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 2.91270783847981e-05,
+      "loss": 0.0435,
+      "step": 12941
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 2.9121140142517817e-05,
+      "loss": 0.0697,
+      "step": 12942
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 2.911520190023753e-05,
+      "loss": 0.0464,
+      "step": 12943
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 2.9109263657957248e-05,
+      "loss": 0.0294,
+      "step": 12944
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 2.910332541567696e-05,
+      "loss": 0.0572,
+      "step": 12945
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 2.909738717339668e-05,
+      "loss": 0.0567,
+      "step": 12946
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 2.909144893111639e-05,
+      "loss": 0.0404,
+      "step": 12947
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 2.9085510688836106e-05,
+      "loss": 0.063,
+      "step": 12948
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 2.907957244655582e-05,
+      "loss": 0.0508,
+      "step": 12949
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 2.9073634204275536e-05,
+      "loss": 0.0676,
+      "step": 12950
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 2.906769596199525e-05,
+      "loss": 0.0654,
+      "step": 12951
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 2.9061757719714967e-05,
+      "loss": 0.0677,
+      "step": 12952
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 2.905581947743468e-05,
+      "loss": 0.0457,
+      "step": 12953
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 2.9049881235154397e-05,
+      "loss": 0.0274,
+      "step": 12954
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 2.904394299287411e-05,
+      "loss": 0.056,
+      "step": 12955
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 2.9038004750593828e-05,
+      "loss": 0.0797,
+      "step": 12956
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 2.9032066508313538e-05,
+      "loss": 0.0606,
+      "step": 12957
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 2.9026128266033255e-05,
+      "loss": 0.0577,
+      "step": 12958
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 2.902019002375297e-05,
+      "loss": 0.0628,
+      "step": 12959
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 2.9014251781472686e-05,
+      "loss": 0.0675,
+      "step": 12960
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 2.90083135391924e-05,
+      "loss": 0.0495,
+      "step": 12961
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 2.9002375296912116e-05,
+      "loss": 0.0388,
+      "step": 12962
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 2.899643705463183e-05,
+      "loss": 0.0721,
+      "step": 12963
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 2.8990498812351547e-05,
+      "loss": 0.0222,
+      "step": 12964
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 2.898456057007126e-05,
+      "loss": 0.0298,
+      "step": 12965
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 2.8978622327790977e-05,
+      "loss": 0.0332,
+      "step": 12966
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 2.8972684085510694e-05,
+      "loss": 0.0289,
+      "step": 12967
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 2.8966745843230404e-05,
+      "loss": 0.0325,
+      "step": 12968
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 2.8960807600950118e-05,
+      "loss": 0.0755,
+      "step": 12969
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 2.8954869358669835e-05,
+      "loss": 0.0979,
+      "step": 12970
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 2.894893111638955e-05,
+      "loss": 0.0597,
+      "step": 12971
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 2.8942992874109265e-05,
+      "loss": 0.0844,
+      "step": 12972
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 2.893705463182898e-05,
+      "loss": 0.0558,
+      "step": 12973
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 2.8931116389548696e-05,
+      "loss": 0.0429,
+      "step": 12974
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 2.8925178147268413e-05,
+      "loss": 0.0572,
+      "step": 12975
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 2.8919239904988127e-05,
+      "loss": 0.0505,
+      "step": 12976
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 2.8913301662707844e-05,
+      "loss": 0.0284,
+      "step": 12977
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 2.8907363420427554e-05,
+      "loss": 0.0796,
+      "step": 12978
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 2.8901425178147267e-05,
+      "loss": 0.0636,
+      "step": 12979
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 2.8895486935866984e-05,
+      "loss": 0.0352,
+      "step": 12980
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 2.8889548693586698e-05,
+      "loss": 0.0403,
+      "step": 12981
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 2.8883610451306415e-05,
+      "loss": 0.0353,
+      "step": 12982
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 2.887767220902613e-05,
+      "loss": 0.0793,
+      "step": 12983
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 2.8871733966745845e-05,
+      "loss": 0.0188,
+      "step": 12984
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 2.8865795724465562e-05,
+      "loss": 0.0953,
+      "step": 12985
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 2.8859857482185276e-05,
+      "loss": 0.0197,
+      "step": 12986
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 2.8853919239904993e-05,
+      "loss": 0.0465,
+      "step": 12987
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 2.8847980997624703e-05,
+      "loss": 0.0915,
+      "step": 12988
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 2.8842042755344417e-05,
+      "loss": 0.0561,
+      "step": 12989
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 2.8836104513064134e-05,
+      "loss": 0.0978,
+      "step": 12990
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 2.8830166270783847e-05,
+      "loss": 0.057,
+      "step": 12991
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 2.8824228028503564e-05,
+      "loss": 0.0564,
+      "step": 12992
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 2.8818289786223278e-05,
+      "loss": 0.0444,
+      "step": 12993
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 2.8812351543942995e-05,
+      "loss": 0.0311,
+      "step": 12994
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 2.880641330166271e-05,
+      "loss": 0.0607,
+      "step": 12995
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 2.8800475059382425e-05,
+      "loss": 0.0562,
+      "step": 12996
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 2.8794536817102142e-05,
+      "loss": 0.0588,
+      "step": 12997
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 2.8788598574821856e-05,
+      "loss": 0.0427,
+      "step": 12998
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 2.8782660332541566e-05,
+      "loss": 0.0572,
+      "step": 12999
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 2.8776722090261283e-05,
+      "loss": 0.0511,
+      "step": 13000
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 2.8770783847980997e-05,
+      "loss": 0.0479,
+      "step": 13001
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 2.8764845605700714e-05,
+      "loss": 0.0609,
+      "step": 13002
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 2.8758907363420427e-05,
+      "loss": 0.0559,
+      "step": 13003
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 2.8752969121140144e-05,
+      "loss": 0.0597,
+      "step": 13004
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 2.874703087885986e-05,
+      "loss": 0.135,
+      "step": 13005
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 2.8741092636579575e-05,
+      "loss": 0.0632,
+      "step": 13006
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 2.873515439429929e-05,
+      "loss": 0.0448,
+      "step": 13007
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 2.8729216152019005e-05,
+      "loss": 0.031,
+      "step": 13008
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 2.8723277909738715e-05,
+      "loss": 0.0627,
+      "step": 13009
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 2.8717339667458432e-05,
+      "loss": 0.062,
+      "step": 13010
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 2.8711401425178146e-05,
+      "loss": 0.0581,
+      "step": 13011
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 2.8705463182897863e-05,
+      "loss": 0.0424,
+      "step": 13012
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 2.8699524940617576e-05,
+      "loss": 0.0399,
+      "step": 13013
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 2.8693586698337293e-05,
+      "loss": 0.0345,
+      "step": 13014
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 2.868764845605701e-05,
+      "loss": 0.0673,
+      "step": 13015
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 2.8681710213776724e-05,
+      "loss": 0.0394,
+      "step": 13016
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 2.867577197149644e-05,
+      "loss": 0.0356,
+      "step": 13017
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 2.8669833729216155e-05,
+      "loss": 0.048,
+      "step": 13018
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 2.866389548693587e-05,
+      "loss": 0.0282,
+      "step": 13019
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 2.8657957244655582e-05,
+      "loss": 0.0336,
+      "step": 13020
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 2.8652019002375295e-05,
+      "loss": 0.046,
+      "step": 13021
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 2.8646080760095012e-05,
+      "loss": 0.0389,
+      "step": 13022
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 2.864014251781473e-05,
+      "loss": 0.0218,
+      "step": 13023
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 2.8634204275534443e-05,
+      "loss": 0.0343,
+      "step": 13024
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 2.862826603325416e-05,
+      "loss": 0.0462,
+      "step": 13025
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 2.8622327790973873e-05,
+      "loss": 0.0421,
+      "step": 13026
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 2.861638954869359e-05,
+      "loss": 0.0369,
+      "step": 13027
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 2.8610451306413304e-05,
+      "loss": 0.1146,
+      "step": 13028
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 2.860451306413302e-05,
+      "loss": 0.0447,
+      "step": 13029
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 2.859857482185273e-05,
+      "loss": 0.0319,
+      "step": 13030
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 2.8592636579572445e-05,
+      "loss": 0.022,
+      "step": 13031
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 2.858669833729216e-05,
+      "loss": 0.076,
+      "step": 13032
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 2.858076009501188e-05,
+      "loss": 0.0739,
+      "step": 13033
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 2.8574821852731592e-05,
+      "loss": 0.0259,
+      "step": 13034
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 2.856888361045131e-05,
+      "loss": 0.0213,
+      "step": 13035
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 2.8562945368171023e-05,
+      "loss": 0.1086,
+      "step": 13036
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 2.855700712589074e-05,
+      "loss": 0.0963,
+      "step": 13037
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 2.8551068883610453e-05,
+      "loss": 0.0396,
+      "step": 13038
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 2.854513064133017e-05,
+      "loss": 0.0777,
+      "step": 13039
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 2.853919239904988e-05,
+      "loss": 0.0535,
+      "step": 13040
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 2.8533254156769594e-05,
+      "loss": 0.0506,
+      "step": 13041
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 2.852731591448931e-05,
+      "loss": 0.0671,
+      "step": 13042
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 2.8521377672209028e-05,
+      "loss": 0.0497,
+      "step": 13043
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 2.851543942992874e-05,
+      "loss": 0.0791,
+      "step": 13044
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 2.850950118764846e-05,
+      "loss": 0.0542,
+      "step": 13045
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 2.8503562945368172e-05,
+      "loss": 0.0384,
+      "step": 13046
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 2.849762470308789e-05,
+      "loss": 0.0549,
+      "step": 13047
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 2.8491686460807603e-05,
+      "loss": 0.0698,
+      "step": 13048
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 2.848574821852732e-05,
+      "loss": 0.0478,
+      "step": 13049
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 2.8479809976247037e-05,
+      "loss": 0.0839,
+      "step": 13050
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 2.8473871733966743e-05,
+      "loss": 0.0386,
+      "step": 13051
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 2.846793349168646e-05,
+      "loss": 0.0231,
+      "step": 13052
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 2.8461995249406177e-05,
+      "loss": 0.0418,
+      "step": 13053
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 2.845605700712589e-05,
+      "loss": 0.0672,
+      "step": 13054
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 2.8450118764845608e-05,
+      "loss": 0.0533,
+      "step": 13055
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 2.844418052256532e-05,
+      "loss": 0.0712,
+      "step": 13056
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 2.843824228028504e-05,
+      "loss": 0.0301,
+      "step": 13057
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 2.8432304038004752e-05,
+      "loss": 0.0234,
+      "step": 13058
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 2.842636579572447e-05,
+      "loss": 0.0448,
+      "step": 13059
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 2.8420427553444186e-05,
+      "loss": 0.0389,
+      "step": 13060
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 2.8414489311163893e-05,
+      "loss": 0.0612,
+      "step": 13061
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 2.840855106888361e-05,
+      "loss": 0.0529,
+      "step": 13062
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 2.8402612826603327e-05,
+      "loss": 0.0626,
+      "step": 13063
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 2.839667458432304e-05,
+      "loss": 0.0397,
+      "step": 13064
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 2.8390736342042757e-05,
+      "loss": 0.0317,
+      "step": 13065
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 2.838479809976247e-05,
+      "loss": 0.0602,
+      "step": 13066
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 2.8378859857482188e-05,
+      "loss": 0.0318,
+      "step": 13067
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 2.83729216152019e-05,
+      "loss": 0.0598,
+      "step": 13068
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 2.8366983372921618e-05,
+      "loss": 0.0547,
+      "step": 13069
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 2.8361045130641335e-05,
+      "loss": 0.0836,
+      "step": 13070
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 2.8355106888361045e-05,
+      "loss": 0.0509,
+      "step": 13071
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 2.834916864608076e-05,
+      "loss": 0.0308,
+      "step": 13072
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 2.8343230403800476e-05,
+      "loss": 0.0516,
+      "step": 13073
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 2.833729216152019e-05,
+      "loss": 0.0364,
+      "step": 13074
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 2.8331353919239907e-05,
+      "loss": 0.0762,
+      "step": 13075
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 2.832541567695962e-05,
+      "loss": 0.0389,
+      "step": 13076
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 2.8319477434679337e-05,
+      "loss": 0.0512,
+      "step": 13077
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 2.831353919239905e-05,
+      "loss": 0.033,
+      "step": 13078
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 2.8307600950118768e-05,
+      "loss": 0.034,
+      "step": 13079
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 2.8301662707838485e-05,
+      "loss": 0.0281,
+      "step": 13080
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 2.8295724465558198e-05,
+      "loss": 0.0418,
+      "step": 13081
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 2.828978622327791e-05,
+      "loss": 0.0254,
+      "step": 13082
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 2.8283847980997625e-05,
+      "loss": 0.0185,
+      "step": 13083
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 2.827790973871734e-05,
+      "loss": 0.0243,
+      "step": 13084
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 2.8271971496437056e-05,
+      "loss": 0.0312,
+      "step": 13085
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 2.826603325415677e-05,
+      "loss": 0.052,
+      "step": 13086
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 2.8260095011876486e-05,
+      "loss": 0.0307,
+      "step": 13087
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 2.82541567695962e-05,
+      "loss": 0.0458,
+      "step": 13088
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 2.8248218527315917e-05,
+      "loss": 0.0397,
+      "step": 13089
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 2.8242280285035634e-05,
+      "loss": 0.0643,
+      "step": 13090
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 2.8236342042755348e-05,
+      "loss": 0.0698,
+      "step": 13091
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 2.8230403800475058e-05,
+      "loss": 0.044,
+      "step": 13092
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 2.8224465558194775e-05,
+      "loss": 0.0605,
+      "step": 13093
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 2.821852731591449e-05,
+      "loss": 0.0508,
+      "step": 13094
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 2.8212589073634205e-05,
+      "loss": 0.0791,
+      "step": 13095
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 2.820665083135392e-05,
+      "loss": 0.0598,
+      "step": 13096
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 2.8200712589073636e-05,
+      "loss": 0.0499,
+      "step": 13097
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 2.8194774346793353e-05,
+      "loss": 0.0449,
+      "step": 13098
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 2.8188836104513066e-05,
+      "loss": 0.0765,
+      "step": 13099
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 2.8182897862232783e-05,
+      "loss": 0.0725,
+      "step": 13100
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 2.8176959619952497e-05,
+      "loss": 0.0477,
+      "step": 13101
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 2.8171021377672207e-05,
+      "loss": 0.0724,
+      "step": 13102
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 2.8165083135391924e-05,
+      "loss": 0.0625,
+      "step": 13103
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 2.8159144893111638e-05,
+      "loss": 0.0421,
+      "step": 13104
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 2.8153206650831355e-05,
+      "loss": 0.0429,
+      "step": 13105
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 2.8147268408551068e-05,
+      "loss": 0.0319,
+      "step": 13106
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 2.8141330166270785e-05,
+      "loss": 0.0401,
+      "step": 13107
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 2.8135391923990502e-05,
+      "loss": 0.0443,
+      "step": 13108
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 2.8129453681710216e-05,
+      "loss": 0.0458,
+      "step": 13109
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 2.8123515439429933e-05,
+      "loss": 0.028,
+      "step": 13110
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 2.8117577197149646e-05,
+      "loss": 0.0567,
+      "step": 13111
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 2.8111638954869363e-05,
+      "loss": 0.0413,
+      "step": 13112
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 2.8105700712589073e-05,
+      "loss": 0.0596,
+      "step": 13113
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 2.8099762470308787e-05,
+      "loss": 0.0827,
+      "step": 13114
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 2.8093824228028504e-05,
+      "loss": 0.0554,
+      "step": 13115
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 2.8087885985748218e-05,
+      "loss": 0.0302,
+      "step": 13116
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 2.8081947743467935e-05,
+      "loss": 0.0461,
+      "step": 13117
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 2.807600950118765e-05,
+      "loss": 0.0459,
+      "step": 13118
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 2.8070071258907365e-05,
+      "loss": 0.0786,
+      "step": 13119
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 2.8064133016627082e-05,
+      "loss": 0.0335,
+      "step": 13120
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 2.8058194774346796e-05,
+      "loss": 0.0353,
+      "step": 13121
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 2.8052256532066513e-05,
+      "loss": 0.0385,
+      "step": 13122
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 2.8046318289786223e-05,
+      "loss": 0.0654,
+      "step": 13123
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 2.8040380047505936e-05,
+      "loss": 0.0634,
+      "step": 13124
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 2.8034441805225653e-05,
+      "loss": 0.058,
+      "step": 13125
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 2.8028503562945367e-05,
+      "loss": 0.034,
+      "step": 13126
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 2.8022565320665084e-05,
+      "loss": 0.0435,
+      "step": 13127
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 2.80166270783848e-05,
+      "loss": 0.0512,
+      "step": 13128
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 2.8010688836104514e-05,
+      "loss": 0.0337,
+      "step": 13129
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 2.800475059382423e-05,
+      "loss": 0.0247,
+      "step": 13130
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 2.7998812351543945e-05,
+      "loss": 0.0334,
+      "step": 13131
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 2.7992874109263662e-05,
+      "loss": 0.0176,
+      "step": 13132
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 2.7986935866983376e-05,
+      "loss": 0.0295,
+      "step": 13133
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 2.7980997624703086e-05,
+      "loss": 0.0279,
+      "step": 13134
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 2.7975059382422803e-05,
+      "loss": 0.2284,
+      "step": 13135
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 2.7969121140142516e-05,
+      "loss": 0.0389,
+      "step": 13136
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 2.7963182897862233e-05,
+      "loss": 0.0057,
+      "step": 13137
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 2.795724465558195e-05,
+      "loss": 0.0603,
+      "step": 13138
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 2.7951306413301664e-05,
+      "loss": 0.0734,
+      "step": 13139
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 2.794536817102138e-05,
+      "loss": 0.0513,
+      "step": 13140
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 2.7939429928741094e-05,
+      "loss": 0.0647,
+      "step": 13141
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 2.793349168646081e-05,
+      "loss": 0.0456,
+      "step": 13142
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 2.7927553444180525e-05,
+      "loss": 0.0338,
+      "step": 13143
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 2.7921615201900235e-05,
+      "loss": 0.0352,
+      "step": 13144
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 2.7915676959619952e-05,
+      "loss": 0.0756,
+      "step": 13145
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 2.790973871733967e-05,
+      "loss": 0.081,
+      "step": 13146
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 2.7903800475059383e-05,
+      "loss": 0.0534,
+      "step": 13147
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 2.78978622327791e-05,
+      "loss": 0.0427,
+      "step": 13148
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 2.7891923990498813e-05,
+      "loss": 0.0653,
+      "step": 13149
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 2.788598574821853e-05,
+      "loss": 0.0398,
+      "step": 13150
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 2.7880047505938244e-05,
+      "loss": 0.0364,
+      "step": 13151
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 2.787410926365796e-05,
+      "loss": 0.037,
+      "step": 13152
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 2.7868171021377674e-05,
+      "loss": 0.0429,
+      "step": 13153
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 2.7862232779097384e-05,
+      "loss": 0.0303,
+      "step": 13154
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 2.78562945368171e-05,
+      "loss": 0.0357,
+      "step": 13155
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 2.785035629453682e-05,
+      "loss": 0.0681,
+      "step": 13156
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 2.7844418052256532e-05,
+      "loss": 0.0381,
+      "step": 13157
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 2.783847980997625e-05,
+      "loss": 0.0238,
+      "step": 13158
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 2.7832541567695963e-05,
+      "loss": 0.0405,
+      "step": 13159
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 2.782660332541568e-05,
+      "loss": 0.0439,
+      "step": 13160
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 2.7820665083135393e-05,
+      "loss": 0.0459,
+      "step": 13161
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 2.781472684085511e-05,
+      "loss": 0.0362,
+      "step": 13162
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 2.7808788598574824e-05,
+      "loss": 0.0545,
+      "step": 13163
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 2.780285035629454e-05,
+      "loss": 0.0473,
+      "step": 13164
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 2.779691211401425e-05,
+      "loss": 0.0314,
+      "step": 13165
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 2.7790973871733968e-05,
+      "loss": 0.038,
+      "step": 13166
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 2.778503562945368e-05,
+      "loss": 0.0459,
+      "step": 13167
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 2.77790973871734e-05,
+      "loss": 0.0511,
+      "step": 13168
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 2.7773159144893112e-05,
+      "loss": 0.0611,
+      "step": 13169
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 2.776722090261283e-05,
+      "loss": 0.0434,
+      "step": 13170
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 2.7761282660332542e-05,
+      "loss": 0.0371,
+      "step": 13171
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 2.775534441805226e-05,
+      "loss": 0.0373,
+      "step": 13172
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 2.7749406175771976e-05,
+      "loss": 0.061,
+      "step": 13173
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 2.774346793349169e-05,
+      "loss": 0.019,
+      "step": 13174
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 2.77375296912114e-05,
+      "loss": 0.02,
+      "step": 13175
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 2.7731591448931117e-05,
+      "loss": 0.0707,
+      "step": 13176
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 2.772565320665083e-05,
+      "loss": 0.0401,
+      "step": 13177
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 2.7719714964370548e-05,
+      "loss": 0.0564,
+      "step": 13178
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 2.771377672209026e-05,
+      "loss": 0.0398,
+      "step": 13179
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 2.7707838479809978e-05,
+      "loss": 0.0654,
+      "step": 13180
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 2.7701900237529692e-05,
+      "loss": 0.0491,
+      "step": 13181
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 2.769596199524941e-05,
+      "loss": 0.0434,
+      "step": 13182
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 2.7690023752969126e-05,
+      "loss": 0.0302,
+      "step": 13183
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 2.768408551068884e-05,
+      "loss": 0.0246,
+      "step": 13184
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 2.767814726840855e-05,
+      "loss": 0.0181,
+      "step": 13185
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 2.7672209026128266e-05,
+      "loss": 0.048,
+      "step": 13186
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 2.766627078384798e-05,
+      "loss": 0.0364,
+      "step": 13187
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 2.7660332541567697e-05,
+      "loss": 0.0265,
+      "step": 13188
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 2.765439429928741e-05,
+      "loss": 0.0678,
+      "step": 13189
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 2.7648456057007128e-05,
+      "loss": 0.0404,
+      "step": 13190
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 2.764251781472684e-05,
+      "loss": 0.0473,
+      "step": 13191
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 2.7636579572446558e-05,
+      "loss": 0.0557,
+      "step": 13192
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 2.7630641330166275e-05,
+      "loss": 0.0445,
+      "step": 13193
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 2.762470308788599e-05,
+      "loss": 0.0435,
+      "step": 13194
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 2.7618764845605706e-05,
+      "loss": 0.0394,
+      "step": 13195
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 2.7612826603325416e-05,
+      "loss": 0.1577,
+      "step": 13196
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 2.760688836104513e-05,
+      "loss": 0.0513,
+      "step": 13197
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 2.7600950118764846e-05,
+      "loss": 0.059,
+      "step": 13198
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 2.759501187648456e-05,
+      "loss": 0.0666,
+      "step": 13199
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 2.7589073634204277e-05,
+      "loss": 0.0981,
+      "step": 13200
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 2.758313539192399e-05,
+      "loss": 0.069,
+      "step": 13201
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 2.7577197149643707e-05,
+      "loss": 0.0353,
+      "step": 13202
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 2.7571258907363424e-05,
+      "loss": 0.0349,
+      "step": 13203
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 2.7565320665083138e-05,
+      "loss": 0.0649,
+      "step": 13204
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 2.7559382422802855e-05,
+      "loss": 0.0828,
+      "step": 13205
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 2.7553444180522565e-05,
+      "loss": 0.0404,
+      "step": 13206
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 2.754750593824228e-05,
+      "loss": 0.0524,
+      "step": 13207
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 2.7541567695961996e-05,
+      "loss": 0.0567,
+      "step": 13208
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 2.753562945368171e-05,
+      "loss": 0.0693,
+      "step": 13209
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 2.7529691211401426e-05,
+      "loss": 0.0466,
+      "step": 13210
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 2.752375296912114e-05,
+      "loss": 0.1716,
+      "step": 13211
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 2.7517814726840857e-05,
+      "loss": 0.018,
+      "step": 13212
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 2.7511876484560574e-05,
+      "loss": 0.0431,
+      "step": 13213
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 2.7505938242280287e-05,
+      "loss": 0.0599,
+      "step": 13214
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 0.0471,
+      "step": 13215
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 2.7494061757719718e-05,
+      "loss": 0.0279,
+      "step": 13216
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 2.7488123515439428e-05,
+      "loss": 0.0924,
+      "step": 13217
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 2.7482185273159145e-05,
+      "loss": 0.039,
+      "step": 13218
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 2.747624703087886e-05,
+      "loss": 0.029,
+      "step": 13219
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 2.7470308788598576e-05,
+      "loss": 0.0553,
+      "step": 13220
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 2.7464370546318293e-05,
+      "loss": 0.0571,
+      "step": 13221
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 2.7458432304038006e-05,
+      "loss": 0.0581,
+      "step": 13222
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 2.7452494061757723e-05,
+      "loss": 0.0556,
+      "step": 13223
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 2.7446555819477437e-05,
+      "loss": 0.0421,
+      "step": 13224
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 2.7440617577197154e-05,
+      "loss": 0.0774,
+      "step": 13225
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 2.7434679334916867e-05,
+      "loss": 0.0535,
+      "step": 13226
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 2.7428741092636577e-05,
+      "loss": 0.0524,
+      "step": 13227
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 2.7422802850356294e-05,
+      "loss": 0.0471,
+      "step": 13228
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 2.7416864608076008e-05,
+      "loss": 0.0665,
+      "step": 13229
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 2.7410926365795725e-05,
+      "loss": 0.0374,
+      "step": 13230
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 2.7404988123515442e-05,
+      "loss": 0.0436,
+      "step": 13231
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 2.7399049881235156e-05,
+      "loss": 0.0442,
+      "step": 13232
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 2.7393111638954873e-05,
+      "loss": 0.0604,
+      "step": 13233
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 2.7387173396674586e-05,
+      "loss": 0.0497,
+      "step": 13234
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 2.7381235154394303e-05,
+      "loss": 0.0559,
+      "step": 13235
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 2.7375296912114017e-05,
+      "loss": 0.0714,
+      "step": 13236
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 2.7369358669833727e-05,
+      "loss": 0.0707,
+      "step": 13237
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 2.7363420427553444e-05,
+      "loss": 0.0793,
+      "step": 13238
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 2.7357482185273157e-05,
+      "loss": 0.0953,
+      "step": 13239
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 2.7351543942992874e-05,
+      "loss": 0.0616,
+      "step": 13240
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 2.734560570071259e-05,
+      "loss": 0.0645,
+      "step": 13241
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 2.7339667458432305e-05,
+      "loss": 0.0617,
+      "step": 13242
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 2.7333729216152022e-05,
+      "loss": 0.0385,
+      "step": 13243
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 2.7327790973871735e-05,
+      "loss": 0.0592,
+      "step": 13244
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 2.7321852731591452e-05,
+      "loss": 0.0453,
+      "step": 13245
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 2.7315914489311166e-05,
+      "loss": 0.0636,
+      "step": 13246
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 2.7309976247030883e-05,
+      "loss": 0.0407,
+      "step": 13247
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 2.7304038004750593e-05,
+      "loss": 0.0634,
+      "step": 13248
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 2.7298099762470307e-05,
+      "loss": 0.0604,
+      "step": 13249
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 2.7292161520190024e-05,
+      "loss": 0.0433,
+      "step": 13250
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 2.728622327790974e-05,
+      "loss": 0.0289,
+      "step": 13251
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 2.7280285035629454e-05,
+      "loss": 0.0737,
+      "step": 13252
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 2.727434679334917e-05,
+      "loss": 0.0708,
+      "step": 13253
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 2.7268408551068885e-05,
+      "loss": 0.0534,
+      "step": 13254
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 2.7262470308788602e-05,
+      "loss": 0.0561,
+      "step": 13255
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 2.7256532066508315e-05,
+      "loss": 0.0575,
+      "step": 13256
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 2.7250593824228032e-05,
+      "loss": 0.047,
+      "step": 13257
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 2.7244655581947743e-05,
+      "loss": 0.0567,
+      "step": 13258
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 2.7238717339667456e-05,
+      "loss": 0.0439,
+      "step": 13259
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 2.7232779097387173e-05,
+      "loss": 0.0561,
+      "step": 13260
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 2.722684085510689e-05,
+      "loss": 0.3379,
+      "step": 13261
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 2.7220902612826604e-05,
+      "loss": 0.0459,
+      "step": 13262
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 2.721496437054632e-05,
+      "loss": 0.0448,
+      "step": 13263
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 2.7209026128266034e-05,
+      "loss": 0.0834,
+      "step": 13264
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 2.720308788598575e-05,
+      "loss": 0.0466,
+      "step": 13265
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 2.7197149643705465e-05,
+      "loss": 0.04,
+      "step": 13266
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 2.7191211401425182e-05,
+      "loss": 0.0798,
+      "step": 13267
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 2.7185273159144892e-05,
+      "loss": 0.0506,
+      "step": 13268
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 2.717933491686461e-05,
+      "loss": 0.0442,
+      "step": 13269
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 2.7173396674584322e-05,
+      "loss": 0.0498,
+      "step": 13270
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 2.716745843230404e-05,
+      "loss": 0.0573,
+      "step": 13271
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 2.7161520190023753e-05,
+      "loss": 0.0483,
+      "step": 13272
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 2.715558194774347e-05,
+      "loss": 0.0451,
+      "step": 13273
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 2.7149643705463184e-05,
+      "loss": 0.052,
+      "step": 13274
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 2.71437054631829e-05,
+      "loss": 0.0645,
+      "step": 13275
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 2.7137767220902614e-05,
+      "loss": 0.0589,
+      "step": 13276
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 2.713182897862233e-05,
+      "loss": 0.0419,
+      "step": 13277
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 2.7125890736342048e-05,
+      "loss": 0.0307,
+      "step": 13278
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 2.7119952494061758e-05,
+      "loss": 0.0504,
+      "step": 13279
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 2.7114014251781472e-05,
+      "loss": 0.0402,
+      "step": 13280
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 2.710807600950119e-05,
+      "loss": 0.0342,
+      "step": 13281
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 2.7102137767220902e-05,
+      "loss": 0.0575,
+      "step": 13282
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 2.709619952494062e-05,
+      "loss": 0.0282,
+      "step": 13283
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 2.7090261282660333e-05,
+      "loss": 0.0342,
+      "step": 13284
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 2.708432304038005e-05,
+      "loss": 0.044,
+      "step": 13285
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 2.7078384798099763e-05,
+      "loss": 0.0418,
+      "step": 13286
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 2.707244655581948e-05,
+      "loss": 0.0202,
+      "step": 13287
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 2.7066508313539197e-05,
+      "loss": 0.0574,
+      "step": 13288
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 2.7060570071258908e-05,
+      "loss": 0.0846,
+      "step": 13289
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 2.705463182897862e-05,
+      "loss": 0.0805,
+      "step": 13290
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 2.7048693586698338e-05,
+      "loss": 0.0687,
+      "step": 13291
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 2.7042755344418052e-05,
+      "loss": 0.0685,
+      "step": 13292
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 2.703681710213777e-05,
+      "loss": 0.0784,
+      "step": 13293
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 2.7030878859857482e-05,
+      "loss": 0.0277,
+      "step": 13294
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 2.70249406175772e-05,
+      "loss": 0.0305,
+      "step": 13295
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 2.7019002375296916e-05,
+      "loss": 0.0485,
+      "step": 13296
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 2.701306413301663e-05,
+      "loss": 0.0715,
+      "step": 13297
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 2.7007125890736347e-05,
+      "loss": 0.1055,
+      "step": 13298
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 2.7001187648456057e-05,
+      "loss": 0.092,
+      "step": 13299
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 2.699524940617577e-05,
+      "loss": 0.0502,
+      "step": 13300
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 2.6989311163895488e-05,
+      "loss": 0.0642,
+      "step": 13301
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 2.69833729216152e-05,
+      "loss": 0.0515,
+      "step": 13302
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 2.6977434679334918e-05,
+      "loss": 0.039,
+      "step": 13303
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 2.697149643705463e-05,
+      "loss": 0.0404,
+      "step": 13304
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 2.696555819477435e-05,
+      "loss": 0.0463,
+      "step": 13305
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 2.6959619952494066e-05,
+      "loss": 0.0678,
+      "step": 13306
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 2.695368171021378e-05,
+      "loss": 0.0459,
+      "step": 13307
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 2.6947743467933496e-05,
+      "loss": 0.0508,
+      "step": 13308
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 2.694180522565321e-05,
+      "loss": 0.0494,
+      "step": 13309
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 2.693586698337292e-05,
+      "loss": 0.069,
+      "step": 13310
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 2.6929928741092637e-05,
+      "loss": 0.0518,
+      "step": 13311
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 2.692399049881235e-05,
+      "loss": 0.0449,
+      "step": 13312
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 2.6918052256532067e-05,
+      "loss": 0.0687,
+      "step": 13313
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 2.691211401425178e-05,
+      "loss": 0.0462,
+      "step": 13314
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 2.6906175771971498e-05,
+      "loss": 0.1058,
+      "step": 13315
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 2.6900237529691215e-05,
+      "loss": 0.0576,
+      "step": 13316
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 2.689429928741093e-05,
+      "loss": 0.0321,
+      "step": 13317
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 2.6888361045130645e-05,
+      "loss": 0.0555,
+      "step": 13318
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 2.688242280285036e-05,
+      "loss": 0.0481,
+      "step": 13319
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 2.687648456057007e-05,
+      "loss": 0.033,
+      "step": 13320
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 2.6870546318289786e-05,
+      "loss": 0.0694,
+      "step": 13321
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 2.68646080760095e-05,
+      "loss": 0.027,
+      "step": 13322
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 2.6858669833729217e-05,
+      "loss": 0.0353,
+      "step": 13323
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 2.685273159144893e-05,
+      "loss": 0.0367,
+      "step": 13324
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 2.6846793349168647e-05,
+      "loss": 0.0389,
+      "step": 13325
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 2.6840855106888364e-05,
+      "loss": 0.0413,
+      "step": 13326
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 2.6834916864608078e-05,
+      "loss": 0.0336,
+      "step": 13327
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 2.6828978622327795e-05,
+      "loss": 0.0347,
+      "step": 13328
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 2.682304038004751e-05,
+      "loss": 0.0465,
+      "step": 13329
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 2.6817102137767225e-05,
+      "loss": 0.0579,
+      "step": 13330
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 2.6811163895486936e-05,
+      "loss": 0.0275,
+      "step": 13331
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 2.680522565320665e-05,
+      "loss": 0.0201,
+      "step": 13332
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 2.6799287410926366e-05,
+      "loss": 0.0391,
+      "step": 13333
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 2.679334916864608e-05,
+      "loss": 0.0274,
+      "step": 13334
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 2.6787410926365797e-05,
+      "loss": 0.0156,
+      "step": 13335
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 2.6781472684085514e-05,
+      "loss": 0.0394,
+      "step": 13336
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 2.6775534441805227e-05,
+      "loss": 0.019,
+      "step": 13337
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 2.6769596199524944e-05,
+      "loss": 0.0146,
+      "step": 13338
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 2.6763657957244658e-05,
+      "loss": 0.0654,
+      "step": 13339
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 2.6757719714964375e-05,
+      "loss": 0.0529,
+      "step": 13340
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 2.6751781472684085e-05,
+      "loss": 0.0668,
+      "step": 13341
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 2.67458432304038e-05,
+      "loss": 0.0832,
+      "step": 13342
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 2.6739904988123515e-05,
+      "loss": 0.0564,
+      "step": 13343
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 2.6733966745843232e-05,
+      "loss": 0.0644,
+      "step": 13344
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 2.6728028503562946e-05,
+      "loss": 0.0587,
+      "step": 13345
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 2.6722090261282663e-05,
+      "loss": 0.0508,
+      "step": 13346
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 2.6716152019002377e-05,
+      "loss": 0.0546,
+      "step": 13347
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 2.6710213776722094e-05,
+      "loss": 0.0626,
+      "step": 13348
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 2.6704275534441807e-05,
+      "loss": 0.0486,
+      "step": 13349
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 2.6698337292161524e-05,
+      "loss": 0.205,
+      "step": 13350
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 2.6692399049881234e-05,
+      "loss": 0.0577,
+      "step": 13351
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 2.6686460807600948e-05,
+      "loss": 0.0753,
+      "step": 13352
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 2.6680522565320665e-05,
+      "loss": 0.0981,
+      "step": 13353
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 2.6674584323040382e-05,
+      "loss": 0.048,
+      "step": 13354
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 2.6668646080760095e-05,
+      "loss": 0.0452,
+      "step": 13355
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 2.6662707838479812e-05,
+      "loss": 0.0305,
+      "step": 13356
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 2.6656769596199526e-05,
+      "loss": 0.0389,
+      "step": 13357
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 2.6650831353919243e-05,
+      "loss": 0.0683,
+      "step": 13358
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 2.6644893111638956e-05,
+      "loss": 0.0531,
+      "step": 13359
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 2.6638954869358673e-05,
+      "loss": 0.0492,
+      "step": 13360
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 2.6633016627078387e-05,
+      "loss": 0.0526,
+      "step": 13361
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 2.6627078384798097e-05,
+      "loss": 0.0357,
+      "step": 13362
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 2.6621140142517814e-05,
+      "loss": 0.0907,
+      "step": 13363
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 2.661520190023753e-05,
+      "loss": 0.0345,
+      "step": 13364
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 2.6609263657957245e-05,
+      "loss": 0.0279,
+      "step": 13365
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 2.6603325415676962e-05,
+      "loss": 0.0298,
+      "step": 13366
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 2.6597387173396675e-05,
+      "loss": 0.0454,
+      "step": 13367
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 2.6591448931116392e-05,
+      "loss": 0.0338,
+      "step": 13368
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 2.6585510688836106e-05,
+      "loss": 0.0396,
+      "step": 13369
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 2.6579572446555823e-05,
+      "loss": 0.0551,
+      "step": 13370
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 2.657363420427554e-05,
+      "loss": 0.0376,
+      "step": 13371
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 2.6567695961995247e-05,
+      "loss": 0.0699,
+      "step": 13372
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 2.6561757719714964e-05,
+      "loss": 0.0516,
+      "step": 13373
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 2.655581947743468e-05,
+      "loss": 0.1019,
+      "step": 13374
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 2.6549881235154394e-05,
+      "loss": 0.0356,
+      "step": 13375
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 2.654394299287411e-05,
+      "loss": 0.055,
+      "step": 13376
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 2.6538004750593825e-05,
+      "loss": 0.0304,
+      "step": 13377
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 2.653206650831354e-05,
+      "loss": 0.0649,
+      "step": 13378
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 2.6526128266033255e-05,
+      "loss": 0.0786,
+      "step": 13379
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 2.6520190023752972e-05,
+      "loss": 0.0224,
+      "step": 13380
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 2.651425178147269e-05,
+      "loss": 0.0918,
+      "step": 13381
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 2.6508313539192396e-05,
+      "loss": 0.0436,
+      "step": 13382
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 2.6502375296912113e-05,
+      "loss": 0.057,
+      "step": 13383
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 2.649643705463183e-05,
+      "loss": 0.0667,
+      "step": 13384
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 2.6490498812351543e-05,
+      "loss": 0.124,
+      "step": 13385
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 2.648456057007126e-05,
+      "loss": 0.0346,
+      "step": 13386
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 2.6478622327790974e-05,
+      "loss": 0.0325,
+      "step": 13387
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 2.647268408551069e-05,
+      "loss": 0.0367,
+      "step": 13388
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 2.6466745843230405e-05,
+      "loss": 0.0488,
+      "step": 13389
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 2.646080760095012e-05,
+      "loss": 0.0373,
+      "step": 13390
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 2.645486935866984e-05,
+      "loss": 0.0662,
+      "step": 13391
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 2.6448931116389552e-05,
+      "loss": 0.0456,
+      "step": 13392
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 2.6442992874109262e-05,
+      "loss": 0.0579,
+      "step": 13393
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 2.643705463182898e-05,
+      "loss": 0.0478,
+      "step": 13394
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 2.6431116389548693e-05,
+      "loss": 0.0809,
+      "step": 13395
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 2.642517814726841e-05,
+      "loss": 0.0359,
+      "step": 13396
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 2.6419239904988123e-05,
+      "loss": 0.0564,
+      "step": 13397
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 2.641330166270784e-05,
+      "loss": 0.0397,
+      "step": 13398
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 2.6407363420427554e-05,
+      "loss": 0.0372,
+      "step": 13399
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 2.640142517814727e-05,
+      "loss": 0.0311,
+      "step": 13400
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 2.6395486935866988e-05,
+      "loss": 0.0509,
+      "step": 13401
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 2.63895486935867e-05,
+      "loss": 0.0507,
+      "step": 13402
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 2.638361045130641e-05,
+      "loss": 0.0514,
+      "step": 13403
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 2.637767220902613e-05,
+      "loss": 0.0483,
+      "step": 13404
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 2.6371733966745842e-05,
+      "loss": 0.0535,
+      "step": 13405
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 2.636579572446556e-05,
+      "loss": 0.0549,
+      "step": 13406
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 2.6359857482185273e-05,
+      "loss": 0.0437,
+      "step": 13407
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 2.635391923990499e-05,
+      "loss": 0.0295,
+      "step": 13408
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 2.6347980997624703e-05,
+      "loss": 0.0458,
+      "step": 13409
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 2.634204275534442e-05,
+      "loss": 0.042,
+      "step": 13410
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 2.6336104513064137e-05,
+      "loss": 0.0727,
+      "step": 13411
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 2.633016627078385e-05,
+      "loss": 0.0489,
+      "step": 13412
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 2.6324228028503568e-05,
+      "loss": 0.0512,
+      "step": 13413
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 2.6318289786223278e-05,
+      "loss": 0.0684,
+      "step": 13414
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 2.631235154394299e-05,
+      "loss": 0.0291,
+      "step": 13415
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 2.630641330166271e-05,
+      "loss": 0.0283,
+      "step": 13416
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 2.6300475059382422e-05,
+      "loss": 0.0269,
+      "step": 13417
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 2.629453681710214e-05,
+      "loss": 0.0282,
+      "step": 13418
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 2.6288598574821856e-05,
+      "loss": 0.0249,
+      "step": 13419
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 2.628266033254157e-05,
+      "loss": 0.0467,
+      "step": 13420
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 2.6276722090261287e-05,
+      "loss": 0.0541,
+      "step": 13421
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 2.6270783847981e-05,
+      "loss": 0.0574,
+      "step": 13422
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 2.6264845605700717e-05,
+      "loss": 0.0689,
+      "step": 13423
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 2.6258907363420427e-05,
+      "loss": 0.033,
+      "step": 13424
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 2.625296912114014e-05,
+      "loss": 0.0181,
+      "step": 13425
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 2.6247030878859858e-05,
+      "loss": 0.0407,
+      "step": 13426
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 2.624109263657957e-05,
+      "loss": 0.0823,
+      "step": 13427
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 2.623515439429929e-05,
+      "loss": 0.0232,
+      "step": 13428
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 2.6229216152019005e-05,
+      "loss": 0.0447,
+      "step": 13429
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 2.622327790973872e-05,
+      "loss": 0.0684,
+      "step": 13430
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 2.6217339667458436e-05,
+      "loss": 0.0657,
+      "step": 13431
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 2.621140142517815e-05,
+      "loss": 0.0709,
+      "step": 13432
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 2.6205463182897866e-05,
+      "loss": 0.0538,
+      "step": 13433
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 2.6199524940617577e-05,
+      "loss": 0.0455,
+      "step": 13434
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 2.619358669833729e-05,
+      "loss": 0.0449,
+      "step": 13435
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 2.6187648456057007e-05,
+      "loss": 0.0369,
+      "step": 13436
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 2.618171021377672e-05,
+      "loss": 0.0427,
+      "step": 13437
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 2.6175771971496438e-05,
+      "loss": 0.0425,
+      "step": 13438
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 2.6169833729216155e-05,
+      "loss": 0.0563,
+      "step": 13439
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 2.616389548693587e-05,
+      "loss": 0.0442,
+      "step": 13440
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 2.6157957244655585e-05,
+      "loss": 0.0482,
+      "step": 13441
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 2.61520190023753e-05,
+      "loss": 0.068,
+      "step": 13442
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 2.6146080760095016e-05,
+      "loss": 0.0483,
+      "step": 13443
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 2.614014251781473e-05,
+      "loss": 0.0478,
+      "step": 13444
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 2.613420427553444e-05,
+      "loss": 0.048,
+      "step": 13445
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 2.6128266033254157e-05,
+      "loss": 0.0477,
+      "step": 13446
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 2.612232779097387e-05,
+      "loss": 0.0553,
+      "step": 13447
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 2.6116389548693587e-05,
+      "loss": 0.051,
+      "step": 13448
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 2.6110451306413304e-05,
+      "loss": 0.0358,
+      "step": 13449
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 2.6104513064133018e-05,
+      "loss": 0.0493,
+      "step": 13450
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 2.6098574821852735e-05,
+      "loss": 0.0474,
+      "step": 13451
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 2.6092636579572448e-05,
+      "loss": 0.0263,
+      "step": 13452
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 2.6086698337292165e-05,
+      "loss": 0.0324,
+      "step": 13453
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 2.608076009501188e-05,
+      "loss": 0.0479,
+      "step": 13454
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 2.607482185273159e-05,
+      "loss": 0.0206,
+      "step": 13455
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 2.6068883610451306e-05,
+      "loss": 0.0298,
+      "step": 13456
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 2.606294536817102e-05,
+      "loss": 0.059,
+      "step": 13457
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 2.6057007125890737e-05,
+      "loss": 0.0458,
+      "step": 13458
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 2.6051068883610453e-05,
+      "loss": 0.0194,
+      "step": 13459
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 2.6045130641330167e-05,
+      "loss": 0.079,
+      "step": 13460
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 2.6039192399049884e-05,
+      "loss": 0.0652,
+      "step": 13461
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 2.6033254156769598e-05,
+      "loss": 0.069,
+      "step": 13462
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 2.6027315914489315e-05,
+      "loss": 0.0335,
+      "step": 13463
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 2.6021377672209028e-05,
+      "loss": 0.0606,
+      "step": 13464
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 2.601543942992874e-05,
+      "loss": 0.0331,
+      "step": 13465
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 2.6009501187648455e-05,
+      "loss": 0.0185,
+      "step": 13466
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 2.6003562945368172e-05,
+      "loss": 0.0301,
+      "step": 13467
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 2.5997624703087886e-05,
+      "loss": 0.0773,
+      "step": 13468
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 2.5991686460807603e-05,
+      "loss": 0.0637,
+      "step": 13469
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 2.5985748218527316e-05,
+      "loss": 0.0493,
+      "step": 13470
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 2.5979809976247033e-05,
+      "loss": 0.0499,
+      "step": 13471
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 2.5973871733966747e-05,
+      "loss": 0.0429,
+      "step": 13472
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 2.5967933491686464e-05,
+      "loss": 0.0401,
+      "step": 13473
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 2.5961995249406178e-05,
+      "loss": 0.0417,
+      "step": 13474
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 2.5956057007125894e-05,
+      "loss": 0.0101,
+      "step": 13475
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 2.5950118764845605e-05,
+      "loss": 0.0159,
+      "step": 13476
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 2.594418052256532e-05,
+      "loss": 0.1439,
+      "step": 13477
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 2.5938242280285035e-05,
+      "loss": 0.0203,
+      "step": 13478
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 2.5932304038004752e-05,
+      "loss": 0.019,
+      "step": 13479
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 2.5926365795724466e-05,
+      "loss": 0.0346,
+      "step": 13480
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 2.5920427553444183e-05,
+      "loss": 0.0778,
+      "step": 13481
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 2.5914489311163896e-05,
+      "loss": 0.0406,
+      "step": 13482
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 2.5908551068883613e-05,
+      "loss": 0.0594,
+      "step": 13483
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 2.5902612826603327e-05,
+      "loss": 0.0825,
+      "step": 13484
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 2.5896674584323044e-05,
+      "loss": 0.0469,
+      "step": 13485
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 2.5890736342042754e-05,
+      "loss": 0.0426,
+      "step": 13486
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 2.588479809976247e-05,
+      "loss": 0.0324,
+      "step": 13487
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 2.5878859857482185e-05,
+      "loss": 0.0343,
+      "step": 13488
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 2.58729216152019e-05,
+      "loss": 0.0491,
+      "step": 13489
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 2.5866983372921615e-05,
+      "loss": 0.0414,
+      "step": 13490
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 2.5861045130641332e-05,
+      "loss": 0.0789,
+      "step": 13491
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 2.5855106888361046e-05,
+      "loss": 0.0651,
+      "step": 13492
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 2.5849168646080763e-05,
+      "loss": 0.0181,
+      "step": 13493
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 2.584323040380048e-05,
+      "loss": 0.0305,
+      "step": 13494
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 2.5837292161520193e-05,
+      "loss": 0.0414,
+      "step": 13495
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 2.583135391923991e-05,
+      "loss": 0.0373,
+      "step": 13496
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 2.582541567695962e-05,
+      "loss": 0.0542,
+      "step": 13497
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 2.5819477434679334e-05,
+      "loss": 0.0351,
+      "step": 13498
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 2.581353919239905e-05,
+      "loss": 0.1151,
+      "step": 13499
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 2.5807600950118764e-05,
+      "loss": 0.0549,
+      "step": 13500
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 2.580166270783848e-05,
+      "loss": 0.0205,
+      "step": 13501
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 2.5795724465558195e-05,
+      "loss": 0.0275,
+      "step": 13502
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 2.5789786223277912e-05,
+      "loss": 0.0343,
+      "step": 13503
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 2.578384798099763e-05,
+      "loss": 0.0162,
+      "step": 13504
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 2.5777909738717343e-05,
+      "loss": 0.066,
+      "step": 13505
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 2.577197149643706e-05,
+      "loss": 0.0347,
+      "step": 13506
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 2.576603325415677e-05,
+      "loss": 0.0468,
+      "step": 13507
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 2.5760095011876483e-05,
+      "loss": 0.079,
+      "step": 13508
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 2.57541567695962e-05,
+      "loss": 0.0435,
+      "step": 13509
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 2.5748218527315914e-05,
+      "loss": 0.0336,
+      "step": 13510
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 2.574228028503563e-05,
+      "loss": 0.043,
+      "step": 13511
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 2.5736342042755344e-05,
+      "loss": 0.0264,
+      "step": 13512
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 2.573040380047506e-05,
+      "loss": 0.028,
+      "step": 13513
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 2.572446555819478e-05,
+      "loss": 0.0563,
+      "step": 13514
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 2.5718527315914492e-05,
+      "loss": 0.0516,
+      "step": 13515
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 2.571258907363421e-05,
+      "loss": 0.0255,
+      "step": 13516
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 2.570665083135392e-05,
+      "loss": 0.0169,
+      "step": 13517
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 2.5700712589073633e-05,
+      "loss": 0.0296,
+      "step": 13518
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 2.569477434679335e-05,
+      "loss": 0.0602,
+      "step": 13519
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 2.5688836104513063e-05,
+      "loss": 0.0687,
+      "step": 13520
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 2.568289786223278e-05,
+      "loss": 0.0401,
+      "step": 13521
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 2.5676959619952494e-05,
+      "loss": 0.0183,
+      "step": 13522
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 2.567102137767221e-05,
+      "loss": 0.0145,
+      "step": 13523
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 2.5665083135391928e-05,
+      "loss": 0.0167,
+      "step": 13524
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 2.565914489311164e-05,
+      "loss": 0.0257,
+      "step": 13525
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 2.5653206650831358e-05,
+      "loss": 0.0115,
+      "step": 13526
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 2.5647268408551072e-05,
+      "loss": 0.0362,
+      "step": 13527
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 2.5641330166270782e-05,
+      "loss": 0.0256,
+      "step": 13528
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 2.56353919239905e-05,
+      "loss": 0.0124,
+      "step": 13529
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 2.5629453681710213e-05,
+      "loss": 0.0519,
+      "step": 13530
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 2.562351543942993e-05,
+      "loss": 0.0774,
+      "step": 13531
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 2.5617577197149643e-05,
+      "loss": 0.0561,
+      "step": 13532
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 2.561163895486936e-05,
+      "loss": 0.0757,
+      "step": 13533
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 2.5605700712589077e-05,
+      "loss": 0.0741,
+      "step": 13534
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 2.559976247030879e-05,
+      "loss": 0.0745,
+      "step": 13535
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 2.5593824228028508e-05,
+      "loss": 0.0728,
+      "step": 13536
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 2.558788598574822e-05,
+      "loss": 0.0393,
+      "step": 13537
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 2.558194774346793e-05,
+      "loss": 0.0344,
+      "step": 13538
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 2.557600950118765e-05,
+      "loss": 0.0386,
+      "step": 13539
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 2.5570071258907362e-05,
+      "loss": 0.0631,
+      "step": 13540
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 2.556413301662708e-05,
+      "loss": 0.0566,
+      "step": 13541
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 2.5558194774346796e-05,
+      "loss": 0.0515,
+      "step": 13542
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 2.555225653206651e-05,
+      "loss": 0.0476,
+      "step": 13543
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 2.5546318289786226e-05,
+      "loss": 0.041,
+      "step": 13544
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 2.554038004750594e-05,
+      "loss": 0.0221,
+      "step": 13545
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 2.5534441805225657e-05,
+      "loss": 0.0261,
+      "step": 13546
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 2.552850356294537e-05,
+      "loss": 0.0339,
+      "step": 13547
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 2.552256532066508e-05,
+      "loss": 0.0918,
+      "step": 13548
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 2.5516627078384798e-05,
+      "loss": 0.0372,
+      "step": 13549
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 2.551068883610451e-05,
+      "loss": 0.0391,
+      "step": 13550
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 2.5504750593824228e-05,
+      "loss": 0.0376,
+      "step": 13551
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 2.5498812351543945e-05,
+      "loss": 0.0457,
+      "step": 13552
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 2.549287410926366e-05,
+      "loss": 0.0493,
+      "step": 13553
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 2.5486935866983376e-05,
+      "loss": 0.0407,
+      "step": 13554
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 2.548099762470309e-05,
+      "loss": 0.0577,
+      "step": 13555
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 2.5475059382422806e-05,
+      "loss": 0.0401,
+      "step": 13556
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 2.546912114014252e-05,
+      "loss": 0.053,
+      "step": 13557
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 2.5463182897862237e-05,
+      "loss": 0.0349,
+      "step": 13558
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 2.5457244655581947e-05,
+      "loss": 0.0277,
+      "step": 13559
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 2.545130641330166e-05,
+      "loss": 0.0518,
+      "step": 13560
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 2.5445368171021378e-05,
+      "loss": 0.0361,
+      "step": 13561
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 2.5439429928741095e-05,
+      "loss": 0.0557,
+      "step": 13562
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 2.5433491686460808e-05,
+      "loss": 0.0369,
+      "step": 13563
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 2.5427553444180525e-05,
+      "loss": 0.039,
+      "step": 13564
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 2.542161520190024e-05,
+      "loss": 0.0395,
+      "step": 13565
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 2.5415676959619956e-05,
+      "loss": 0.0371,
+      "step": 13566
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 2.540973871733967e-05,
+      "loss": 0.0311,
+      "step": 13567
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 2.5403800475059386e-05,
+      "loss": 0.0477,
+      "step": 13568
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 2.5397862232779096e-05,
+      "loss": 0.0222,
+      "step": 13569
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 2.539192399049881e-05,
+      "loss": 0.0521,
+      "step": 13570
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 2.5385985748218527e-05,
+      "loss": 0.025,
+      "step": 13571
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 2.5380047505938244e-05,
+      "loss": 0.0532,
+      "step": 13572
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 2.5374109263657958e-05,
+      "loss": 0.0633,
+      "step": 13573
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 2.5368171021377674e-05,
+      "loss": 0.0196,
+      "step": 13574
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 2.5362232779097388e-05,
+      "loss": 0.0156,
+      "step": 13575
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 2.5356294536817105e-05,
+      "loss": 0.017,
+      "step": 13576
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 2.535035629453682e-05,
+      "loss": 0.0311,
+      "step": 13577
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 2.5344418052256536e-05,
+      "loss": 0.0288,
+      "step": 13578
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 2.5338479809976246e-05,
+      "loss": 0.0189,
+      "step": 13579
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 2.533254156769596e-05,
+      "loss": 0.0419,
+      "step": 13580
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 2.5326603325415676e-05,
+      "loss": 0.0487,
+      "step": 13581
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 2.5320665083135393e-05,
+      "loss": 0.0958,
+      "step": 13582
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 2.5314726840855107e-05,
+      "loss": 0.0345,
+      "step": 13583
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 2.5308788598574824e-05,
+      "loss": 0.0497,
+      "step": 13584
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 2.5302850356294537e-05,
+      "loss": 0.04,
+      "step": 13585
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 2.5296912114014254e-05,
+      "loss": 0.0396,
+      "step": 13586
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 2.5290973871733968e-05,
+      "loss": 0.0426,
+      "step": 13587
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 2.5285035629453685e-05,
+      "loss": 0.0434,
+      "step": 13588
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 2.5279097387173402e-05,
+      "loss": 0.0547,
+      "step": 13589
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 2.5273159144893112e-05,
+      "loss": 0.0343,
+      "step": 13590
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 2.5267220902612826e-05,
+      "loss": 0.0301,
+      "step": 13591
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 2.5261282660332543e-05,
+      "loss": 0.0317,
+      "step": 13592
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 2.5255344418052256e-05,
+      "loss": 0.0341,
+      "step": 13593
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 2.5249406175771973e-05,
+      "loss": 0.0327,
+      "step": 13594
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 2.5243467933491687e-05,
+      "loss": 0.0676,
+      "step": 13595
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 2.5237529691211404e-05,
+      "loss": 0.0328,
+      "step": 13596
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 2.5231591448931117e-05,
+      "loss": 0.0407,
+      "step": 13597
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 2.5225653206650834e-05,
+      "loss": 0.0392,
+      "step": 13598
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 2.521971496437055e-05,
+      "loss": 0.0638,
+      "step": 13599
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 2.521377672209026e-05,
+      "loss": 0.045,
+      "step": 13600
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 2.5207838479809975e-05,
+      "loss": 0.0369,
+      "step": 13601
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 2.5201900237529692e-05,
+      "loss": 0.0387,
+      "step": 13602
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 2.5195961995249406e-05,
+      "loss": 0.019,
+      "step": 13603
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 2.5190023752969123e-05,
+      "loss": 0.0507,
+      "step": 13604
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 2.5184085510688836e-05,
+      "loss": 0.0289,
+      "step": 13605
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 2.5178147268408553e-05,
+      "loss": 0.0375,
+      "step": 13606
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 2.5172209026128267e-05,
+      "loss": 0.0387,
+      "step": 13607
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 2.5166270783847984e-05,
+      "loss": 0.0351,
+      "step": 13608
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 2.51603325415677e-05,
+      "loss": 0.0275,
+      "step": 13609
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 2.5154394299287414e-05,
+      "loss": 0.04,
+      "step": 13610
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 2.5148456057007124e-05,
+      "loss": 0.0262,
+      "step": 13611
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 2.514251781472684e-05,
+      "loss": 0.0325,
+      "step": 13612
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 2.5136579572446555e-05,
+      "loss": 0.0345,
+      "step": 13613
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 2.5130641330166272e-05,
+      "loss": 0.0585,
+      "step": 13614
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 2.5124703087885986e-05,
+      "loss": 0.0497,
+      "step": 13615
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 2.5118764845605702e-05,
+      "loss": 0.0307,
+      "step": 13616
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 2.511282660332542e-05,
+      "loss": 0.0347,
+      "step": 13617
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 2.5106888361045133e-05,
+      "loss": 0.0365,
+      "step": 13618
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 2.510095011876485e-05,
+      "loss": 0.0421,
+      "step": 13619
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 2.5095011876484564e-05,
+      "loss": 0.029,
+      "step": 13620
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 2.5089073634204274e-05,
+      "loss": 0.0171,
+      "step": 13621
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 2.508313539192399e-05,
+      "loss": 0.0549,
+      "step": 13622
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 2.5077197149643704e-05,
+      "loss": 0.0301,
+      "step": 13623
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 2.507125890736342e-05,
+      "loss": 0.0288,
+      "step": 13624
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 2.5065320665083135e-05,
+      "loss": 0.048,
+      "step": 13625
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 2.5059382422802852e-05,
+      "loss": 0.0417,
+      "step": 13626
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 2.505344418052257e-05,
+      "loss": 0.0132,
+      "step": 13627
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 2.5047505938242282e-05,
+      "loss": 0.0142,
+      "step": 13628
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 2.5041567695962e-05,
+      "loss": 0.0189,
+      "step": 13629
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 2.5035629453681713e-05,
+      "loss": 0.0855,
+      "step": 13630
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 2.5029691211401423e-05,
+      "loss": 0.112,
+      "step": 13631
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 2.502375296912114e-05,
+      "loss": 0.0442,
+      "step": 13632
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 2.5017814726840854e-05,
+      "loss": 0.0494,
+      "step": 13633
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 2.501187648456057e-05,
+      "loss": 0.0292,
+      "step": 13634
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 2.5005938242280284e-05,
+      "loss": 0.1012,
+      "step": 13635
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0596,
+      "step": 13636
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 2.4994061757719718e-05,
+      "loss": 0.0402,
+      "step": 13637
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 2.4988123515439432e-05,
+      "loss": 0.0518,
+      "step": 13638
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 2.4982185273159145e-05,
+      "loss": 0.0433,
+      "step": 13639
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 2.497624703087886e-05,
+      "loss": 0.0425,
+      "step": 13640
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 2.4970308788598576e-05,
+      "loss": 0.0846,
+      "step": 13641
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 2.4964370546318293e-05,
+      "loss": 0.0402,
+      "step": 13642
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 2.4958432304038006e-05,
+      "loss": 0.0823,
+      "step": 13643
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 2.495249406175772e-05,
+      "loss": 0.066,
+      "step": 13644
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 2.4946555819477434e-05,
+      "loss": 0.0607,
+      "step": 13645
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 2.494061757719715e-05,
+      "loss": 0.0391,
+      "step": 13646
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 2.4934679334916868e-05,
+      "loss": 0.0386,
+      "step": 13647
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 2.492874109263658e-05,
+      "loss": 0.0461,
+      "step": 13648
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 2.4922802850356295e-05,
+      "loss": 0.0641,
+      "step": 13649
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 2.4916864608076008e-05,
+      "loss": 0.0542,
+      "step": 13650
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 2.4910926365795725e-05,
+      "loss": 0.0536,
+      "step": 13651
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 2.4904988123515442e-05,
+      "loss": 0.0498,
+      "step": 13652
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 2.4899049881235156e-05,
+      "loss": 0.0449,
+      "step": 13653
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 2.4893111638954873e-05,
+      "loss": 0.0399,
+      "step": 13654
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 2.4887173396674583e-05,
+      "loss": 0.0317,
+      "step": 13655
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 2.48812351543943e-05,
+      "loss": 0.0421,
+      "step": 13656
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 2.4875296912114017e-05,
+      "loss": 0.0395,
+      "step": 13657
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 2.486935866983373e-05,
+      "loss": 0.0316,
+      "step": 13658
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 2.4863420427553447e-05,
+      "loss": 0.0317,
+      "step": 13659
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 2.4857482185273158e-05,
+      "loss": 0.0407,
+      "step": 13660
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 2.4851543942992875e-05,
+      "loss": 0.0392,
+      "step": 13661
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 2.484560570071259e-05,
+      "loss": 0.0271,
+      "step": 13662
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 2.4839667458432305e-05,
+      "loss": 0.02,
+      "step": 13663
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 2.4833729216152022e-05,
+      "loss": 0.0544,
+      "step": 13664
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 2.4827790973871736e-05,
+      "loss": 0.0355,
+      "step": 13665
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 2.482185273159145e-05,
+      "loss": 0.0296,
+      "step": 13666
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 2.4815914489311166e-05,
+      "loss": 0.0543,
+      "step": 13667
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 2.480997624703088e-05,
+      "loss": 0.026,
+      "step": 13668
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 2.4804038004750597e-05,
+      "loss": 0.0249,
+      "step": 13669
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 2.479809976247031e-05,
+      "loss": 0.0525,
+      "step": 13670
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 2.4792161520190024e-05,
+      "loss": 0.0524,
+      "step": 13671
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 2.478622327790974e-05,
+      "loss": 0.0449,
+      "step": 13672
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 2.4780285035629455e-05,
+      "loss": 0.0239,
+      "step": 13673
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 2.477434679334917e-05,
+      "loss": 0.0446,
+      "step": 13674
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 2.4768408551068885e-05,
+      "loss": 0.0406,
+      "step": 13675
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 2.47624703087886e-05,
+      "loss": 0.029,
+      "step": 13676
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 2.4756532066508316e-05,
+      "loss": 0.0113,
+      "step": 13677
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 2.475059382422803e-05,
+      "loss": 0.0365,
+      "step": 13678
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 2.4744655581947746e-05,
+      "loss": 0.0095,
+      "step": 13679
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 2.473871733966746e-05,
+      "loss": 0.0199,
+      "step": 13680
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 2.4732779097387173e-05,
+      "loss": 0.0683,
+      "step": 13681
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 2.472684085510689e-05,
+      "loss": 0.0604,
+      "step": 13682
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 2.4720902612826604e-05,
+      "loss": 0.0598,
+      "step": 13683
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 2.471496437054632e-05,
+      "loss": 0.0495,
+      "step": 13684
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 2.4709026128266034e-05,
+      "loss": 0.0845,
+      "step": 13685
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 2.4703087885985748e-05,
+      "loss": 0.0369,
+      "step": 13686
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 2.4697149643705465e-05,
+      "loss": 0.0683,
+      "step": 13687
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 2.469121140142518e-05,
+      "loss": 0.0736,
+      "step": 13688
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 2.4685273159144896e-05,
+      "loss": 0.0385,
+      "step": 13689
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 2.467933491686461e-05,
+      "loss": 0.0358,
+      "step": 13690
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 2.4673396674584323e-05,
+      "loss": 0.0621,
+      "step": 13691
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 2.466745843230404e-05,
+      "loss": 0.0432,
+      "step": 13692
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 2.4661520190023753e-05,
+      "loss": 0.039,
+      "step": 13693
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 2.465558194774347e-05,
+      "loss": 0.0459,
+      "step": 13694
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 2.4649643705463184e-05,
+      "loss": 0.0812,
+      "step": 13695
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 2.4643705463182897e-05,
+      "loss": 0.0434,
+      "step": 13696
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 2.4637767220902614e-05,
+      "loss": 0.0216,
+      "step": 13697
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 2.4631828978622328e-05,
+      "loss": 0.0529,
+      "step": 13698
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 2.4625890736342045e-05,
+      "loss": 0.0258,
+      "step": 13699
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 2.461995249406176e-05,
+      "loss": 0.0438,
+      "step": 13700
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 2.4614014251781472e-05,
+      "loss": 0.0302,
+      "step": 13701
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 2.460807600950119e-05,
+      "loss": 0.0418,
+      "step": 13702
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 2.4602137767220903e-05,
+      "loss": 0.0263,
+      "step": 13703
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 2.459619952494062e-05,
+      "loss": 0.0394,
+      "step": 13704
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 2.4590261282660333e-05,
+      "loss": 0.0369,
+      "step": 13705
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 2.4584323040380047e-05,
+      "loss": 0.0267,
+      "step": 13706
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 2.4578384798099764e-05,
+      "loss": 0.0604,
+      "step": 13707
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 2.4572446555819477e-05,
+      "loss": 0.0347,
+      "step": 13708
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 2.4566508313539194e-05,
+      "loss": 0.0358,
+      "step": 13709
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 2.4560570071258908e-05,
+      "loss": 0.0316,
+      "step": 13710
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 2.4554631828978625e-05,
+      "loss": 0.0415,
+      "step": 13711
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 2.454869358669834e-05,
+      "loss": 0.0324,
+      "step": 13712
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 2.4542755344418052e-05,
+      "loss": 0.0384,
+      "step": 13713
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 2.453681710213777e-05,
+      "loss": 0.0318,
+      "step": 13714
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 2.4530878859857482e-05,
+      "loss": 0.0483,
+      "step": 13715
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 2.45249406175772e-05,
+      "loss": 0.0408,
+      "step": 13716
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 2.4519002375296913e-05,
+      "loss": 0.0353,
+      "step": 13717
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 2.4513064133016627e-05,
+      "loss": 0.0432,
+      "step": 13718
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 2.4507125890736344e-05,
+      "loss": 0.0134,
+      "step": 13719
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 2.4501187648456057e-05,
+      "loss": 0.0514,
+      "step": 13720
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 2.4495249406175774e-05,
+      "loss": 0.1123,
+      "step": 13721
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 2.4489311163895488e-05,
+      "loss": 0.0321,
+      "step": 13722
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 2.44833729216152e-05,
+      "loss": 0.0419,
+      "step": 13723
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 2.4477434679334918e-05,
+      "loss": 0.0306,
+      "step": 13724
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 2.4471496437054632e-05,
+      "loss": 0.0461,
+      "step": 13725
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 2.446555819477435e-05,
+      "loss": 0.0313,
+      "step": 13726
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 2.4459619952494062e-05,
+      "loss": 0.0685,
+      "step": 13727
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 2.4453681710213776e-05,
+      "loss": 0.1113,
+      "step": 13728
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 2.4447743467933493e-05,
+      "loss": 0.0105,
+      "step": 13729
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 2.4441805225653207e-05,
+      "loss": 0.0366,
+      "step": 13730
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 2.4435866983372923e-05,
+      "loss": 0.0662,
+      "step": 13731
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 2.4429928741092637e-05,
+      "loss": 0.0541,
+      "step": 13732
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 2.442399049881235e-05,
+      "loss": 0.062,
+      "step": 13733
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 2.4418052256532068e-05,
+      "loss": 0.0332,
+      "step": 13734
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 2.4412114014251785e-05,
+      "loss": 0.0513,
+      "step": 13735
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 2.4406175771971498e-05,
+      "loss": 0.0379,
+      "step": 13736
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 2.4400237529691215e-05,
+      "loss": 0.0418,
+      "step": 13737
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 2.4394299287410925e-05,
+      "loss": 0.0407,
+      "step": 13738
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 2.4388361045130642e-05,
+      "loss": 0.1511,
+      "step": 13739
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 2.438242280285036e-05,
+      "loss": 0.0285,
+      "step": 13740
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 2.4376484560570073e-05,
+      "loss": 0.0369,
+      "step": 13741
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 2.437054631828979e-05,
+      "loss": 0.0449,
+      "step": 13742
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 2.43646080760095e-05,
+      "loss": 0.0706,
+      "step": 13743
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 2.4358669833729217e-05,
+      "loss": 0.0463,
+      "step": 13744
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 2.4352731591448934e-05,
+      "loss": 0.0414,
+      "step": 13745
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 2.4346793349168648e-05,
+      "loss": 0.077,
+      "step": 13746
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 2.4340855106888365e-05,
+      "loss": 0.0736,
+      "step": 13747
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 2.4334916864608075e-05,
+      "loss": 0.0409,
+      "step": 13748
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 2.432897862232779e-05,
+      "loss": 0.0474,
+      "step": 13749
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 2.432304038004751e-05,
+      "loss": 0.0581,
+      "step": 13750
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 2.4317102137767222e-05,
+      "loss": 0.059,
+      "step": 13751
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 2.431116389548694e-05,
+      "loss": 0.0384,
+      "step": 13752
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 2.430522565320665e-05,
+      "loss": 0.0336,
+      "step": 13753
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 2.4299287410926366e-05,
+      "loss": 0.0387,
+      "step": 13754
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 2.4293349168646083e-05,
+      "loss": 0.025,
+      "step": 13755
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 2.4287410926365797e-05,
+      "loss": 0.0498,
+      "step": 13756
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 2.4281472684085514e-05,
+      "loss": 0.0359,
+      "step": 13757
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 2.4275534441805224e-05,
+      "loss": 0.0274,
+      "step": 13758
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 2.426959619952494e-05,
+      "loss": 0.0397,
+      "step": 13759
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 2.4263657957244658e-05,
+      "loss": 0.0537,
+      "step": 13760
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 2.425771971496437e-05,
+      "loss": 0.0493,
+      "step": 13761
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 2.425178147268409e-05,
+      "loss": 0.0321,
+      "step": 13762
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 2.4245843230403802e-05,
+      "loss": 0.0524,
+      "step": 13763
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 2.4239904988123516e-05,
+      "loss": 0.0313,
+      "step": 13764
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 2.4233966745843233e-05,
+      "loss": 0.038,
+      "step": 13765
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 2.4228028503562946e-05,
+      "loss": 0.0382,
+      "step": 13766
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 2.4222090261282663e-05,
+      "loss": 0.0569,
+      "step": 13767
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 2.4216152019002377e-05,
+      "loss": 0.0233,
+      "step": 13768
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 2.421021377672209e-05,
+      "loss": 0.0269,
+      "step": 13769
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 2.4204275534441807e-05,
+      "loss": 0.0249,
+      "step": 13770
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 2.419833729216152e-05,
+      "loss": 0.0169,
+      "step": 13771
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 2.4192399049881238e-05,
+      "loss": 0.0396,
+      "step": 13772
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 2.418646080760095e-05,
+      "loss": 0.0155,
+      "step": 13773
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 2.4180522565320665e-05,
+      "loss": 0.0308,
+      "step": 13774
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 2.4174584323040382e-05,
+      "loss": 0.0191,
+      "step": 13775
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 2.4168646080760096e-05,
+      "loss": 0.029,
+      "step": 13776
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 2.4162707838479813e-05,
+      "loss": 0.0089,
+      "step": 13777
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 2.4156769596199526e-05,
+      "loss": 0.0366,
+      "step": 13778
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 2.415083135391924e-05,
+      "loss": 0.0603,
+      "step": 13779
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 2.4144893111638957e-05,
+      "loss": 0.0337,
+      "step": 13780
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 2.413895486935867e-05,
+      "loss": 0.0281,
+      "step": 13781
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 2.4133016627078387e-05,
+      "loss": 0.088,
+      "step": 13782
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 2.41270783847981e-05,
+      "loss": 0.0343,
+      "step": 13783
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 2.4121140142517814e-05,
+      "loss": 0.0723,
+      "step": 13784
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 2.411520190023753e-05,
+      "loss": 0.0967,
+      "step": 13785
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 2.4109263657957245e-05,
+      "loss": 0.0759,
+      "step": 13786
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 2.4103325415676962e-05,
+      "loss": 0.0397,
+      "step": 13787
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 2.4097387173396676e-05,
+      "loss": 0.0411,
+      "step": 13788
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 2.409144893111639e-05,
+      "loss": 0.0629,
+      "step": 13789
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 2.4085510688836106e-05,
+      "loss": 0.0455,
+      "step": 13790
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 2.407957244655582e-05,
+      "loss": 0.0545,
+      "step": 13791
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 2.4073634204275537e-05,
+      "loss": 0.0509,
+      "step": 13792
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 2.406769596199525e-05,
+      "loss": 0.062,
+      "step": 13793
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 2.4061757719714967e-05,
+      "loss": 0.0427,
+      "step": 13794
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 2.405581947743468e-05,
+      "loss": 0.0409,
+      "step": 13795
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 2.4049881235154394e-05,
+      "loss": 0.0543,
+      "step": 13796
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 2.404394299287411e-05,
+      "loss": 0.0264,
+      "step": 13797
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 2.4038004750593825e-05,
+      "loss": 0.0618,
+      "step": 13798
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 2.4032066508313542e-05,
+      "loss": 0.049,
+      "step": 13799
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 2.4026128266033255e-05,
+      "loss": 0.0695,
+      "step": 13800
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 2.402019002375297e-05,
+      "loss": 0.0379,
+      "step": 13801
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 2.4014251781472686e-05,
+      "loss": 0.0239,
+      "step": 13802
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 2.40083135391924e-05,
+      "loss": 0.0352,
+      "step": 13803
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 2.4002375296912117e-05,
+      "loss": 0.0408,
+      "step": 13804
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 2.399643705463183e-05,
+      "loss": 0.0241,
+      "step": 13805
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 2.3990498812351544e-05,
+      "loss": 0.0518,
+      "step": 13806
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 2.398456057007126e-05,
+      "loss": 0.0306,
+      "step": 13807
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 2.3978622327790974e-05,
+      "loss": 0.0777,
+      "step": 13808
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 2.397268408551069e-05,
+      "loss": 0.0901,
+      "step": 13809
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 2.3966745843230405e-05,
+      "loss": 0.0296,
+      "step": 13810
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 2.396080760095012e-05,
+      "loss": 0.0459,
+      "step": 13811
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 2.3954869358669835e-05,
+      "loss": 0.0172,
+      "step": 13812
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 2.394893111638955e-05,
+      "loss": 0.0756,
+      "step": 13813
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 2.3942992874109266e-05,
+      "loss": 0.0645,
+      "step": 13814
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 2.393705463182898e-05,
+      "loss": 0.0313,
+      "step": 13815
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 2.3931116389548693e-05,
+      "loss": 0.0234,
+      "step": 13816
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 2.392517814726841e-05,
+      "loss": 0.0303,
+      "step": 13817
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 2.3919239904988124e-05,
+      "loss": 0.0408,
+      "step": 13818
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 2.391330166270784e-05,
+      "loss": 0.0521,
+      "step": 13819
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 2.3907363420427558e-05,
+      "loss": 0.0194,
+      "step": 13820
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 2.3901425178147268e-05,
+      "loss": 0.0349,
+      "step": 13821
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 2.3895486935866985e-05,
+      "loss": 0.0297,
+      "step": 13822
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 2.3889548693586698e-05,
+      "loss": 0.0261,
+      "step": 13823
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 2.3883610451306415e-05,
+      "loss": 0.0332,
+      "step": 13824
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 2.3877672209026132e-05,
+      "loss": 0.0429,
+      "step": 13825
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 2.3871733966745842e-05,
+      "loss": 0.0084,
+      "step": 13826
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 2.386579572446556e-05,
+      "loss": 0.0194,
+      "step": 13827
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 2.3859857482185273e-05,
+      "loss": 0.0224,
+      "step": 13828
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 2.385391923990499e-05,
+      "loss": 0.0173,
+      "step": 13829
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 2.3847980997624707e-05,
+      "loss": 0.0201,
+      "step": 13830
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 2.3842042755344417e-05,
+      "loss": 0.049,
+      "step": 13831
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 2.3836104513064134e-05,
+      "loss": 0.0662,
+      "step": 13832
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 2.3830166270783848e-05,
+      "loss": 0.0523,
+      "step": 13833
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 2.3824228028503565e-05,
+      "loss": 0.0761,
+      "step": 13834
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 2.381828978622328e-05,
+      "loss": 0.0764,
+      "step": 13835
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 2.3812351543942992e-05,
+      "loss": 0.0681,
+      "step": 13836
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 2.380641330166271e-05,
+      "loss": 0.0266,
+      "step": 13837
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 2.3800475059382422e-05,
+      "loss": 0.0344,
+      "step": 13838
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 2.379453681710214e-05,
+      "loss": 0.0675,
+      "step": 13839
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 2.3788598574821856e-05,
+      "loss": 0.0664,
+      "step": 13840
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 2.3782660332541566e-05,
+      "loss": 0.0369,
+      "step": 13841
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 2.3776722090261283e-05,
+      "loss": 0.1794,
+      "step": 13842
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 2.3770783847980997e-05,
+      "loss": 0.0413,
+      "step": 13843
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 2.3764845605700714e-05,
+      "loss": 0.0289,
+      "step": 13844
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 2.375890736342043e-05,
+      "loss": 0.061,
+      "step": 13845
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 2.375296912114014e-05,
+      "loss": 0.0436,
+      "step": 13846
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 2.3747030878859858e-05,
+      "loss": 0.0464,
+      "step": 13847
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 2.374109263657957e-05,
+      "loss": 0.029,
+      "step": 13848
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 2.373515439429929e-05,
+      "loss": 0.0388,
+      "step": 13849
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 2.3729216152019006e-05,
+      "loss": 0.025,
+      "step": 13850
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 2.372327790973872e-05,
+      "loss": 0.0387,
+      "step": 13851
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 2.3717339667458433e-05,
+      "loss": 0.0347,
+      "step": 13852
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 2.3711401425178146e-05,
+      "loss": 0.0449,
+      "step": 13853
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 2.3705463182897863e-05,
+      "loss": 0.0591,
+      "step": 13854
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 2.369952494061758e-05,
+      "loss": 0.0334,
+      "step": 13855
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 2.3693586698337294e-05,
+      "loss": 0.0346,
+      "step": 13856
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 2.3687648456057007e-05,
+      "loss": 0.0482,
+      "step": 13857
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 2.368171021377672e-05,
+      "loss": 0.0294,
+      "step": 13858
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 2.3675771971496438e-05,
+      "loss": 0.0314,
+      "step": 13859
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 2.3669833729216155e-05,
+      "loss": 0.0328,
+      "step": 13860
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 2.366389548693587e-05,
+      "loss": 0.0467,
+      "step": 13861
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 2.3657957244655582e-05,
+      "loss": 0.0603,
+      "step": 13862
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 2.36520190023753e-05,
+      "loss": 0.0342,
+      "step": 13863
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 2.3646080760095013e-05,
+      "loss": 0.0252,
+      "step": 13864
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 2.364014251781473e-05,
+      "loss": 0.0274,
+      "step": 13865
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 2.3634204275534443e-05,
+      "loss": 0.0243,
+      "step": 13866
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 2.3628266033254157e-05,
+      "loss": 0.04,
+      "step": 13867
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 2.3622327790973874e-05,
+      "loss": 0.0319,
+      "step": 13868
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 2.3616389548693587e-05,
+      "loss": 0.0314,
+      "step": 13869
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 2.3610451306413304e-05,
+      "loss": 0.0309,
+      "step": 13870
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 2.3604513064133018e-05,
+      "loss": 0.039,
+      "step": 13871
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 2.359857482185273e-05,
+      "loss": 0.0534,
+      "step": 13872
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 2.359263657957245e-05,
+      "loss": 0.0205,
+      "step": 13873
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 2.3586698337292162e-05,
+      "loss": 0.021,
+      "step": 13874
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 2.358076009501188e-05,
+      "loss": 0.0277,
+      "step": 13875
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 2.3574821852731593e-05,
+      "loss": 0.036,
+      "step": 13876
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 2.356888361045131e-05,
+      "loss": 0.0496,
+      "step": 13877
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 2.3562945368171023e-05,
+      "loss": 0.0132,
+      "step": 13878
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 2.3557007125890737e-05,
+      "loss": 0.0719,
+      "step": 13879
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 2.3551068883610454e-05,
+      "loss": 0.0563,
+      "step": 13880
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 2.3545130641330167e-05,
+      "loss": 0.072,
+      "step": 13881
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 2.3539192399049884e-05,
+      "loss": 0.075,
+      "step": 13882
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 2.3533254156769598e-05,
+      "loss": 0.0493,
+      "step": 13883
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 2.352731591448931e-05,
+      "loss": 0.0778,
+      "step": 13884
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 2.352137767220903e-05,
+      "loss": 0.051,
+      "step": 13885
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 2.3515439429928742e-05,
+      "loss": 0.0415,
+      "step": 13886
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 2.350950118764846e-05,
+      "loss": 0.0475,
+      "step": 13887
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 2.3503562945368173e-05,
+      "loss": 0.0298,
+      "step": 13888
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 2.3497624703087886e-05,
+      "loss": 0.0445,
+      "step": 13889
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 2.3491686460807603e-05,
+      "loss": 0.0683,
+      "step": 13890
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 2.3485748218527317e-05,
+      "loss": 0.0393,
+      "step": 13891
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 2.3479809976247034e-05,
+      "loss": 0.0536,
+      "step": 13892
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 2.3473871733966747e-05,
+      "loss": 0.0581,
+      "step": 13893
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 2.346793349168646e-05,
+      "loss": 0.0217,
+      "step": 13894
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 2.3461995249406178e-05,
+      "loss": 0.0567,
+      "step": 13895
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 2.345605700712589e-05,
+      "loss": 0.0257,
+      "step": 13896
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 2.3450118764845608e-05,
+      "loss": 0.0592,
+      "step": 13897
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 2.3444180522565322e-05,
+      "loss": 0.0506,
+      "step": 13898
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 2.3438242280285035e-05,
+      "loss": 0.0478,
+      "step": 13899
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 2.3432304038004752e-05,
+      "loss": 0.0351,
+      "step": 13900
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 2.3426365795724466e-05,
+      "loss": 0.0518,
+      "step": 13901
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 2.3420427553444183e-05,
+      "loss": 0.0349,
+      "step": 13902
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 2.3414489311163897e-05,
+      "loss": 0.0433,
+      "step": 13903
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 2.340855106888361e-05,
+      "loss": 0.047,
+      "step": 13904
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 2.3402612826603327e-05,
+      "loss": 0.0476,
+      "step": 13905
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 2.339667458432304e-05,
+      "loss": 0.0299,
+      "step": 13906
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 2.3390736342042758e-05,
+      "loss": 0.0352,
+      "step": 13907
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 2.338479809976247e-05,
+      "loss": 0.0293,
+      "step": 13908
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 2.3378859857482185e-05,
+      "loss": 0.034,
+      "step": 13909
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 2.3372921615201902e-05,
+      "loss": 0.0479,
+      "step": 13910
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 2.3366983372921615e-05,
+      "loss": 0.027,
+      "step": 13911
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 2.3361045130641332e-05,
+      "loss": 0.044,
+      "step": 13912
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 2.3355106888361046e-05,
+      "loss": 0.0345,
+      "step": 13913
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 2.334916864608076e-05,
+      "loss": 0.0524,
+      "step": 13914
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 2.3343230403800476e-05,
+      "loss": 0.0471,
+      "step": 13915
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 2.333729216152019e-05,
+      "loss": 0.038,
+      "step": 13916
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 2.3331353919239907e-05,
+      "loss": 0.0274,
+      "step": 13917
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 2.332541567695962e-05,
+      "loss": 0.0561,
+      "step": 13918
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 2.3319477434679334e-05,
+      "loss": 0.0358,
+      "step": 13919
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 2.331353919239905e-05,
+      "loss": 0.0407,
+      "step": 13920
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 2.3307600950118765e-05,
+      "loss": 0.045,
+      "step": 13921
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 2.330166270783848e-05,
+      "loss": 0.024,
+      "step": 13922
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 2.3295724465558195e-05,
+      "loss": 0.0362,
+      "step": 13923
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 2.328978622327791e-05,
+      "loss": 0.0259,
+      "step": 13924
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 2.3283847980997626e-05,
+      "loss": 0.0167,
+      "step": 13925
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 2.327790973871734e-05,
+      "loss": 0.0133,
+      "step": 13926
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 2.3271971496437056e-05,
+      "loss": 0.0659,
+      "step": 13927
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 2.326603325415677e-05,
+      "loss": 0.0434,
+      "step": 13928
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 2.3260095011876484e-05,
+      "loss": 0.0282,
+      "step": 13929
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 2.32541567695962e-05,
+      "loss": 0.0288,
+      "step": 13930
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 2.3248218527315914e-05,
+      "loss": 0.031,
+      "step": 13931
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 2.324228028503563e-05,
+      "loss": 0.0885,
+      "step": 13932
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 2.3236342042755348e-05,
+      "loss": 0.0224,
+      "step": 13933
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 2.323040380047506e-05,
+      "loss": 0.0493,
+      "step": 13934
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 2.3224465558194775e-05,
+      "loss": 0.0476,
+      "step": 13935
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 2.321852731591449e-05,
+      "loss": 0.0615,
+      "step": 13936
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 2.3212589073634206e-05,
+      "loss": 0.046,
+      "step": 13937
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 2.3206650831353923e-05,
+      "loss": 0.0349,
+      "step": 13938
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 2.3200712589073636e-05,
+      "loss": 0.0574,
+      "step": 13939
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 2.319477434679335e-05,
+      "loss": 0.0727,
+      "step": 13940
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 2.3188836104513063e-05,
+      "loss": 0.0316,
+      "step": 13941
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 2.318289786223278e-05,
+      "loss": 0.0486,
+      "step": 13942
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 2.3176959619952497e-05,
+      "loss": 0.0351,
+      "step": 13943
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 2.317102137767221e-05,
+      "loss": 0.0374,
+      "step": 13944
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 2.3165083135391925e-05,
+      "loss": 0.0412,
+      "step": 13945
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 2.3159144893111638e-05,
+      "loss": 0.0438,
+      "step": 13946
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 2.3153206650831355e-05,
+      "loss": 0.1007,
+      "step": 13947
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 2.3147268408551072e-05,
+      "loss": 0.038,
+      "step": 13948
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 2.3141330166270786e-05,
+      "loss": 0.0604,
+      "step": 13949
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 2.31353919239905e-05,
+      "loss": 0.0335,
+      "step": 13950
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 2.3129453681710213e-05,
+      "loss": 0.0396,
+      "step": 13951
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 2.312351543942993e-05,
+      "loss": 0.0396,
+      "step": 13952
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 2.3117577197149647e-05,
+      "loss": 0.0198,
+      "step": 13953
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 2.311163895486936e-05,
+      "loss": 0.0297,
+      "step": 13954
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 2.3105700712589074e-05,
+      "loss": 0.0409,
+      "step": 13955
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 2.3099762470308787e-05,
+      "loss": 0.0529,
+      "step": 13956
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 2.3093824228028504e-05,
+      "loss": 0.0427,
+      "step": 13957
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 2.308788598574822e-05,
+      "loss": 0.0528,
+      "step": 13958
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 2.3081947743467935e-05,
+      "loss": 0.0274,
+      "step": 13959
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 2.3076009501187652e-05,
+      "loss": 0.0342,
+      "step": 13960
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 2.3070071258907362e-05,
+      "loss": 0.0283,
+      "step": 13961
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 2.306413301662708e-05,
+      "loss": 0.0437,
+      "step": 13962
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 2.3058194774346796e-05,
+      "loss": 0.0241,
+      "step": 13963
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 2.305225653206651e-05,
+      "loss": 0.0421,
+      "step": 13964
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 2.3046318289786227e-05,
+      "loss": 0.0725,
+      "step": 13965
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 2.3040380047505937e-05,
+      "loss": 0.0452,
+      "step": 13966
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 2.3034441805225654e-05,
+      "loss": 0.048,
+      "step": 13967
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 2.302850356294537e-05,
+      "loss": 0.0628,
+      "step": 13968
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 2.3022565320665084e-05,
+      "loss": 0.0235,
+      "step": 13969
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 2.30166270783848e-05,
+      "loss": 0.0261,
+      "step": 13970
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 2.301068883610451e-05,
+      "loss": 0.0324,
+      "step": 13971
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 2.300475059382423e-05,
+      "loss": 0.0399,
+      "step": 13972
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 2.2998812351543945e-05,
+      "loss": 0.0204,
+      "step": 13973
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 2.299287410926366e-05,
+      "loss": 0.0555,
+      "step": 13974
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 2.2986935866983376e-05,
+      "loss": 0.0385,
+      "step": 13975
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 2.2980997624703086e-05,
+      "loss": 0.0386,
+      "step": 13976
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 2.2975059382422803e-05,
+      "loss": 0.031,
+      "step": 13977
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 2.296912114014252e-05,
+      "loss": 0.1035,
+      "step": 13978
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 2.2963182897862234e-05,
+      "loss": 0.0103,
+      "step": 13979
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 2.295724465558195e-05,
+      "loss": 0.0407,
+      "step": 13980
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 2.295130641330166e-05,
+      "loss": 0.0659,
+      "step": 13981
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 2.2945368171021378e-05,
+      "loss": 0.0484,
+      "step": 13982
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 2.2939429928741095e-05,
+      "loss": 0.0709,
+      "step": 13983
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 2.293349168646081e-05,
+      "loss": 0.0517,
+      "step": 13984
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 2.2927553444180525e-05,
+      "loss": 0.0381,
+      "step": 13985
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 2.292161520190024e-05,
+      "loss": 0.0423,
+      "step": 13986
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 2.2915676959619953e-05,
+      "loss": 0.0702,
+      "step": 13987
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 2.290973871733967e-05,
+      "loss": 0.1714,
+      "step": 13988
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 2.2903800475059383e-05,
+      "loss": 0.0504,
+      "step": 13989
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 2.28978622327791e-05,
+      "loss": 0.0326,
+      "step": 13990
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 2.2891923990498814e-05,
+      "loss": 0.0303,
+      "step": 13991
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 2.2885985748218527e-05,
+      "loss": 0.0297,
+      "step": 13992
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 2.2880047505938244e-05,
+      "loss": 0.0368,
+      "step": 13993
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 2.2874109263657958e-05,
+      "loss": 0.0601,
+      "step": 13994
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 2.2868171021377675e-05,
+      "loss": 0.0339,
+      "step": 13995
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 2.2862232779097388e-05,
+      "loss": 0.0565,
+      "step": 13996
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 2.2856294536817102e-05,
+      "loss": 0.0341,
+      "step": 13997
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 2.285035629453682e-05,
+      "loss": 0.0472,
+      "step": 13998
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 2.2844418052256532e-05,
+      "loss": 0.0449,
+      "step": 13999
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 2.283847980997625e-05,
+      "loss": 0.0489,
+      "step": 14000
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 2.2832541567695963e-05,
+      "loss": 0.0481,
+      "step": 14001
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 2.2826603325415677e-05,
+      "loss": 0.0461,
+      "step": 14002
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 2.2820665083135394e-05,
+      "loss": 0.0372,
+      "step": 14003
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 2.2814726840855107e-05,
+      "loss": 0.0302,
+      "step": 14004
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 2.2808788598574824e-05,
+      "loss": 0.0286,
+      "step": 14005
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 2.2802850356294538e-05,
+      "loss": 0.0386,
+      "step": 14006
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 2.279691211401425e-05,
+      "loss": 0.0518,
+      "step": 14007
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 2.2790973871733968e-05,
+      "loss": 0.0658,
+      "step": 14008
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 2.2785035629453682e-05,
+      "loss": 0.0224,
+      "step": 14009
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 2.27790973871734e-05,
+      "loss": 0.0385,
+      "step": 14010
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 2.2773159144893112e-05,
+      "loss": 0.0325,
+      "step": 14011
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 2.2767220902612826e-05,
+      "loss": 0.0355,
+      "step": 14012
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 2.2761282660332543e-05,
+      "loss": 0.0555,
+      "step": 14013
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 2.2755344418052256e-05,
+      "loss": 0.0376,
+      "step": 14014
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 2.2749406175771973e-05,
+      "loss": 0.0384,
+      "step": 14015
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 2.2743467933491687e-05,
+      "loss": 0.052,
+      "step": 14016
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 2.2737529691211404e-05,
+      "loss": 0.0411,
+      "step": 14017
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 2.2731591448931118e-05,
+      "loss": 0.0287,
+      "step": 14018
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 2.272565320665083e-05,
+      "loss": 0.0528,
+      "step": 14019
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 2.2719714964370548e-05,
+      "loss": 0.0428,
+      "step": 14020
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 2.271377672209026e-05,
+      "loss": 0.039,
+      "step": 14021
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 2.270783847980998e-05,
+      "loss": 0.0298,
+      "step": 14022
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 2.2701900237529692e-05,
+      "loss": 0.0177,
+      "step": 14023
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 2.2695961995249406e-05,
+      "loss": 0.027,
+      "step": 14024
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 2.2690023752969123e-05,
+      "loss": 0.0318,
+      "step": 14025
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 2.2684085510688836e-05,
+      "loss": 0.0597,
+      "step": 14026
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 2.2678147268408553e-05,
+      "loss": 0.0221,
+      "step": 14027
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 2.2672209026128267e-05,
+      "loss": 0.029,
+      "step": 14028
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 2.266627078384798e-05,
+      "loss": 0.0202,
+      "step": 14029
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 2.2660332541567697e-05,
+      "loss": 0.0158,
+      "step": 14030
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 2.265439429928741e-05,
+      "loss": 0.0594,
+      "step": 14031
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 2.2648456057007128e-05,
+      "loss": 0.0572,
+      "step": 14032
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 2.264251781472684e-05,
+      "loss": 0.0644,
+      "step": 14033
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 2.2636579572446555e-05,
+      "loss": 0.0328,
+      "step": 14034
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 2.2630641330166272e-05,
+      "loss": 0.0832,
+      "step": 14035
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 2.2624703087885986e-05,
+      "loss": 0.0294,
+      "step": 14036
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 2.2618764845605703e-05,
+      "loss": 0.0436,
+      "step": 14037
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 2.2612826603325416e-05,
+      "loss": 0.0499,
+      "step": 14038
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 2.260688836104513e-05,
+      "loss": 0.0404,
+      "step": 14039
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 2.2600950118764847e-05,
+      "loss": 0.0344,
+      "step": 14040
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 2.259501187648456e-05,
+      "loss": 0.0451,
+      "step": 14041
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 2.2589073634204277e-05,
+      "loss": 0.0542,
+      "step": 14042
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 2.2583135391923994e-05,
+      "loss": 0.0888,
+      "step": 14043
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 2.2577197149643705e-05,
+      "loss": 0.0587,
+      "step": 14044
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 2.257125890736342e-05,
+      "loss": 0.0785,
+      "step": 14045
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 2.2565320665083135e-05,
+      "loss": 0.0333,
+      "step": 14046
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 2.2559382422802852e-05,
+      "loss": 0.0356,
+      "step": 14047
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 2.255344418052257e-05,
+      "loss": 0.0264,
+      "step": 14048
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 2.254750593824228e-05,
+      "loss": 0.0972,
+      "step": 14049
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 2.2541567695961996e-05,
+      "loss": 0.0442,
+      "step": 14050
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 2.253562945368171e-05,
+      "loss": 0.0489,
+      "step": 14051
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 2.2529691211401427e-05,
+      "loss": 0.0476,
+      "step": 14052
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 2.2523752969121144e-05,
+      "loss": 0.0318,
+      "step": 14053
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 2.2517814726840854e-05,
+      "loss": 0.0393,
+      "step": 14054
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 2.251187648456057e-05,
+      "loss": 0.0311,
+      "step": 14055
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 2.2505938242280284e-05,
+      "loss": 0.0456,
+      "step": 14056
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 2.25e-05,
+      "loss": 0.0567,
+      "step": 14057
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 2.249406175771972e-05,
+      "loss": 0.0384,
+      "step": 14058
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 2.248812351543943e-05,
+      "loss": 0.0304,
+      "step": 14059
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 2.2482185273159146e-05,
+      "loss": 0.0405,
+      "step": 14060
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 2.2476247030878863e-05,
+      "loss": 0.0348,
+      "step": 14061
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 2.2470308788598576e-05,
+      "loss": 0.0254,
+      "step": 14062
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 2.2464370546318293e-05,
+      "loss": 0.05,
+      "step": 14063
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 2.2458432304038003e-05,
+      "loss": 0.0448,
+      "step": 14064
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 2.245249406175772e-05,
+      "loss": 0.0158,
+      "step": 14065
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 2.2446555819477437e-05,
+      "loss": 0.0479,
+      "step": 14066
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 2.244061757719715e-05,
+      "loss": 0.0222,
+      "step": 14067
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 2.2434679334916868e-05,
+      "loss": 0.037,
+      "step": 14068
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 2.2428741092636578e-05,
+      "loss": 0.1659,
+      "step": 14069
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 2.2422802850356295e-05,
+      "loss": 0.0446,
+      "step": 14070
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 2.2416864608076012e-05,
+      "loss": 0.0332,
+      "step": 14071
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 2.2410926365795725e-05,
+      "loss": 0.0165,
+      "step": 14072
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 2.2404988123515442e-05,
+      "loss": 0.0133,
+      "step": 14073
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 2.2399049881235156e-05,
+      "loss": 0.0261,
+      "step": 14074
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 2.239311163895487e-05,
+      "loss": 0.0418,
+      "step": 14075
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 2.2387173396674587e-05,
+      "loss": 0.0541,
+      "step": 14076
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 2.23812351543943e-05,
+      "loss": 0.0269,
+      "step": 14077
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 2.2375296912114017e-05,
+      "loss": 0.0279,
+      "step": 14078
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 2.236935866983373e-05,
+      "loss": 0.027,
+      "step": 14079
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 2.2363420427553444e-05,
+      "loss": 0.0585,
+      "step": 14080
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 2.235748218527316e-05,
+      "loss": 0.1129,
+      "step": 14081
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 2.2351543942992875e-05,
+      "loss": 0.0388,
+      "step": 14082
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 2.2345605700712592e-05,
+      "loss": 0.0434,
+      "step": 14083
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 2.2339667458432305e-05,
+      "loss": 0.0348,
+      "step": 14084
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 2.233372921615202e-05,
+      "loss": 0.0458,
+      "step": 14085
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 2.2327790973871736e-05,
+      "loss": 0.0392,
+      "step": 14086
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 2.232185273159145e-05,
+      "loss": 0.0413,
+      "step": 14087
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 2.2315914489311166e-05,
+      "loss": 0.0417,
+      "step": 14088
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 2.230997624703088e-05,
+      "loss": 0.0408,
+      "step": 14089
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 2.2304038004750594e-05,
+      "loss": 0.0572,
+      "step": 14090
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 2.229809976247031e-05,
+      "loss": 0.0522,
+      "step": 14091
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 2.2292161520190024e-05,
+      "loss": 0.0439,
+      "step": 14092
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 2.228622327790974e-05,
+      "loss": 0.0665,
+      "step": 14093
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 2.2280285035629455e-05,
+      "loss": 0.0342,
+      "step": 14094
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 2.2274346793349168e-05,
+      "loss": 0.0518,
+      "step": 14095
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 2.2268408551068885e-05,
+      "loss": 0.0178,
+      "step": 14096
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 2.22624703087886e-05,
+      "loss": 0.0912,
+      "step": 14097
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 2.2256532066508316e-05,
+      "loss": 0.0538,
+      "step": 14098
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 2.225059382422803e-05,
+      "loss": 0.0195,
+      "step": 14099
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 2.2244655581947746e-05,
+      "loss": 0.0368,
+      "step": 14100
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 2.223871733966746e-05,
+      "loss": 0.0777,
+      "step": 14101
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 2.2232779097387174e-05,
+      "loss": 0.0318,
+      "step": 14102
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 2.222684085510689e-05,
+      "loss": 0.0495,
+      "step": 14103
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 2.2220902612826604e-05,
+      "loss": 0.0345,
+      "step": 14104
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 2.221496437054632e-05,
+      "loss": 0.0329,
+      "step": 14105
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 2.2209026128266035e-05,
+      "loss": 0.03,
+      "step": 14106
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 2.2203087885985748e-05,
+      "loss": 0.0507,
+      "step": 14107
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 2.2197149643705465e-05,
+      "loss": 0.0496,
+      "step": 14108
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 2.219121140142518e-05,
+      "loss": 0.044,
+      "step": 14109
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 2.2185273159144896e-05,
+      "loss": 0.0406,
+      "step": 14110
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 2.217933491686461e-05,
+      "loss": 0.0333,
+      "step": 14111
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 2.2173396674584323e-05,
+      "loss": 0.0469,
+      "step": 14112
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 2.216745843230404e-05,
+      "loss": 0.0306,
+      "step": 14113
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 2.2161520190023753e-05,
+      "loss": 0.0474,
+      "step": 14114
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 2.215558194774347e-05,
+      "loss": 0.0515,
+      "step": 14115
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 2.2149643705463184e-05,
+      "loss": 0.0363,
+      "step": 14116
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 2.2143705463182898e-05,
+      "loss": 0.0453,
+      "step": 14117
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 2.2137767220902615e-05,
+      "loss": 0.0251,
+      "step": 14118
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 2.2131828978622328e-05,
+      "loss": 0.0229,
+      "step": 14119
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 2.2125890736342045e-05,
+      "loss": 0.0564,
+      "step": 14120
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 2.211995249406176e-05,
+      "loss": 0.0263,
+      "step": 14121
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 2.2114014251781472e-05,
+      "loss": 0.0264,
+      "step": 14122
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 2.210807600950119e-05,
+      "loss": 0.048,
+      "step": 14123
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 2.2102137767220903e-05,
+      "loss": 0.0301,
+      "step": 14124
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 2.209619952494062e-05,
+      "loss": 0.0206,
+      "step": 14125
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 2.2090261282660333e-05,
+      "loss": 0.0258,
+      "step": 14126
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 2.2084323040380047e-05,
+      "loss": 0.0819,
+      "step": 14127
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 2.2078384798099764e-05,
+      "loss": 0.0231,
+      "step": 14128
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 2.2072446555819477e-05,
+      "loss": 0.0123,
+      "step": 14129
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 2.2066508313539194e-05,
+      "loss": 0.0591,
+      "step": 14130
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 2.206057007125891e-05,
+      "loss": 0.0458,
+      "step": 14131
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 2.205463182897862e-05,
+      "loss": 0.0424,
+      "step": 14132
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 2.204869358669834e-05,
+      "loss": 0.0372,
+      "step": 14133
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 2.2042755344418052e-05,
+      "loss": 0.0306,
+      "step": 14134
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 2.203681710213777e-05,
+      "loss": 0.0477,
+      "step": 14135
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 2.2030878859857486e-05,
+      "loss": 0.0413,
+      "step": 14136
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 2.2024940617577196e-05,
+      "loss": 0.0558,
+      "step": 14137
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 2.2019002375296913e-05,
+      "loss": 0.0735,
+      "step": 14138
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 2.2013064133016627e-05,
+      "loss": 0.0249,
+      "step": 14139
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 2.2007125890736344e-05,
+      "loss": 0.0484,
+      "step": 14140
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 2.200118764845606e-05,
+      "loss": 0.0483,
+      "step": 14141
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 2.199524940617577e-05,
+      "loss": 0.0399,
+      "step": 14142
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 2.1989311163895488e-05,
+      "loss": 0.065,
+      "step": 14143
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 2.19833729216152e-05,
+      "loss": 0.0324,
+      "step": 14144
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 2.197743467933492e-05,
+      "loss": 0.0424,
+      "step": 14145
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 2.1971496437054635e-05,
+      "loss": 0.0405,
+      "step": 14146
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 2.1965558194774346e-05,
+      "loss": 0.0535,
+      "step": 14147
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 2.1959619952494063e-05,
+      "loss": 0.0359,
+      "step": 14148
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 2.1953681710213776e-05,
+      "loss": 0.0355,
+      "step": 14149
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 2.1947743467933493e-05,
+      "loss": 0.1599,
+      "step": 14150
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 2.194180522565321e-05,
+      "loss": 0.0411,
+      "step": 14151
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 2.193586698337292e-05,
+      "loss": 0.0214,
+      "step": 14152
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 2.1929928741092637e-05,
+      "loss": 0.06,
+      "step": 14153
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 2.192399049881235e-05,
+      "loss": 0.0386,
+      "step": 14154
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 2.1918052256532068e-05,
+      "loss": 0.039,
+      "step": 14155
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 2.1912114014251785e-05,
+      "loss": 0.0611,
+      "step": 14156
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 2.19061757719715e-05,
+      "loss": 0.0389,
+      "step": 14157
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 2.1900237529691212e-05,
+      "loss": 0.0641,
+      "step": 14158
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 2.1894299287410926e-05,
+      "loss": 0.0568,
+      "step": 14159
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 2.1888361045130643e-05,
+      "loss": 0.0444,
+      "step": 14160
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 2.188242280285036e-05,
+      "loss": 0.0477,
+      "step": 14161
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 2.1876484560570073e-05,
+      "loss": 0.0418,
+      "step": 14162
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 2.1870546318289787e-05,
+      "loss": 0.0313,
+      "step": 14163
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 2.18646080760095e-05,
+      "loss": 0.0385,
+      "step": 14164
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 2.1858669833729217e-05,
+      "loss": 0.0204,
+      "step": 14165
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 2.1852731591448934e-05,
+      "loss": 0.0154,
+      "step": 14166
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 2.1846793349168648e-05,
+      "loss": 0.0473,
+      "step": 14167
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 2.184085510688836e-05,
+      "loss": 0.0258,
+      "step": 14168
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 2.1834916864608075e-05,
+      "loss": 0.044,
+      "step": 14169
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 2.1828978622327792e-05,
+      "loss": 0.0395,
+      "step": 14170
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 2.182304038004751e-05,
+      "loss": 0.0217,
+      "step": 14171
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 2.1817102137767222e-05,
+      "loss": 0.0307,
+      "step": 14172
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 2.1811163895486936e-05,
+      "loss": 0.0385,
+      "step": 14173
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 2.180522565320665e-05,
+      "loss": 0.0275,
+      "step": 14174
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 2.1799287410926367e-05,
+      "loss": 0.0309,
+      "step": 14175
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 2.1793349168646084e-05,
+      "loss": 0.0385,
+      "step": 14176
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 2.1787410926365797e-05,
+      "loss": 0.0335,
+      "step": 14177
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 2.178147268408551e-05,
+      "loss": 0.0302,
+      "step": 14178
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 2.1775534441805224e-05,
+      "loss": 0.0041,
+      "step": 14179
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 2.176959619952494e-05,
+      "loss": 0.0183,
+      "step": 14180
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 2.1763657957244658e-05,
+      "loss": 0.0514,
+      "step": 14181
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 2.1757719714964372e-05,
+      "loss": 0.0603,
+      "step": 14182
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 2.1751781472684085e-05,
+      "loss": 0.0974,
+      "step": 14183
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 2.1745843230403802e-05,
+      "loss": 0.0537,
+      "step": 14184
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 2.1739904988123516e-05,
+      "loss": 0.0418,
+      "step": 14185
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 2.1733966745843233e-05,
+      "loss": 0.044,
+      "step": 14186
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 2.1728028503562946e-05,
+      "loss": 0.0378,
+      "step": 14187
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 2.1722090261282663e-05,
+      "loss": 0.045,
+      "step": 14188
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 2.1716152019002377e-05,
+      "loss": 0.0378,
+      "step": 14189
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 2.171021377672209e-05,
+      "loss": 0.0453,
+      "step": 14190
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 2.1704275534441808e-05,
+      "loss": 0.0417,
+      "step": 14191
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 2.169833729216152e-05,
+      "loss": 0.0269,
+      "step": 14192
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 2.1692399049881238e-05,
+      "loss": 0.0428,
+      "step": 14193
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 2.1686460807600952e-05,
+      "loss": 0.0399,
+      "step": 14194
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 2.1680522565320665e-05,
+      "loss": 0.0697,
+      "step": 14195
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 2.1674584323040382e-05,
+      "loss": 0.0258,
+      "step": 14196
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 2.1668646080760096e-05,
+      "loss": 0.0536,
+      "step": 14197
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 2.1662707838479813e-05,
+      "loss": 0.0425,
+      "step": 14198
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 2.1656769596199526e-05,
+      "loss": 0.034,
+      "step": 14199
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 2.165083135391924e-05,
+      "loss": 0.0354,
+      "step": 14200
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 2.1644893111638957e-05,
+      "loss": 0.0412,
+      "step": 14201
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 2.163895486935867e-05,
+      "loss": 0.0446,
+      "step": 14202
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 2.1633016627078387e-05,
+      "loss": 0.0487,
+      "step": 14203
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 2.16270783847981e-05,
+      "loss": 0.0444,
+      "step": 14204
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 2.1621140142517815e-05,
+      "loss": 0.0598,
+      "step": 14205
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 2.161520190023753e-05,
+      "loss": 0.0378,
+      "step": 14206
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 2.1609263657957245e-05,
+      "loss": 0.0311,
+      "step": 14207
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 2.1603325415676962e-05,
+      "loss": 0.0356,
+      "step": 14208
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 2.1597387173396676e-05,
+      "loss": 0.0309,
+      "step": 14209
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 2.159144893111639e-05,
+      "loss": 0.0276,
+      "step": 14210
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 2.1585510688836106e-05,
+      "loss": 0.0408,
+      "step": 14211
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 2.157957244655582e-05,
+      "loss": 0.0596,
+      "step": 14212
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 2.1573634204275537e-05,
+      "loss": 0.0365,
+      "step": 14213
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 2.156769596199525e-05,
+      "loss": 0.0811,
+      "step": 14214
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 2.1561757719714964e-05,
+      "loss": 0.0371,
+      "step": 14215
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 2.155581947743468e-05,
+      "loss": 0.0599,
+      "step": 14216
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 2.1549881235154395e-05,
+      "loss": 0.0271,
+      "step": 14217
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 2.154394299287411e-05,
+      "loss": 0.0348,
+      "step": 14218
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 2.1538004750593825e-05,
+      "loss": 0.0345,
+      "step": 14219
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 2.153206650831354e-05,
+      "loss": 0.043,
+      "step": 14220
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 2.1526128266033256e-05,
+      "loss": 0.0367,
+      "step": 14221
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 2.152019002375297e-05,
+      "loss": 0.0332,
+      "step": 14222
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 2.1514251781472686e-05,
+      "loss": 0.0315,
+      "step": 14223
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 2.15083135391924e-05,
+      "loss": 0.0313,
+      "step": 14224
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 2.1502375296912113e-05,
+      "loss": 0.1018,
+      "step": 14225
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 2.149643705463183e-05,
+      "loss": 0.0332,
+      "step": 14226
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 2.1490498812351544e-05,
+      "loss": 0.0136,
+      "step": 14227
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 2.148456057007126e-05,
+      "loss": 0.0182,
+      "step": 14228
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 2.1478622327790974e-05,
+      "loss": 0.076,
+      "step": 14229
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 2.1472684085510688e-05,
+      "loss": 0.0898,
+      "step": 14230
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 2.1466745843230405e-05,
+      "loss": 0.0496,
+      "step": 14231
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 2.146080760095012e-05,
+      "loss": 0.0688,
+      "step": 14232
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 2.1454869358669836e-05,
+      "loss": 0.0495,
+      "step": 14233
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 2.144893111638955e-05,
+      "loss": 0.0502,
+      "step": 14234
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 2.1442992874109263e-05,
+      "loss": 0.0227,
+      "step": 14235
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 2.143705463182898e-05,
+      "loss": 0.0402,
+      "step": 14236
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 2.1431116389548693e-05,
+      "loss": 0.0347,
+      "step": 14237
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 2.142517814726841e-05,
+      "loss": 0.0365,
+      "step": 14238
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 2.1419239904988124e-05,
+      "loss": 0.0446,
+      "step": 14239
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 2.141330166270784e-05,
+      "loss": 0.0441,
+      "step": 14240
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 2.1407363420427554e-05,
+      "loss": 0.0519,
+      "step": 14241
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 2.1401425178147268e-05,
+      "loss": 0.0616,
+      "step": 14242
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 2.1395486935866985e-05,
+      "loss": 0.0402,
+      "step": 14243
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 2.13895486935867e-05,
+      "loss": 0.045,
+      "step": 14244
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 2.1383610451306415e-05,
+      "loss": 0.0349,
+      "step": 14245
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 2.137767220902613e-05,
+      "loss": 0.044,
+      "step": 14246
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 2.1371733966745843e-05,
+      "loss": 0.0439,
+      "step": 14247
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 2.136579572446556e-05,
+      "loss": 0.0387,
+      "step": 14248
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 2.1359857482185273e-05,
+      "loss": 0.0418,
+      "step": 14249
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 2.135391923990499e-05,
+      "loss": 0.0433,
+      "step": 14250
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 2.1347980997624704e-05,
+      "loss": 0.0471,
+      "step": 14251
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 2.1342042755344417e-05,
+      "loss": 0.0489,
+      "step": 14252
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 2.1336104513064134e-05,
+      "loss": 0.0411,
+      "step": 14253
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 2.1330166270783848e-05,
+      "loss": 0.0276,
+      "step": 14254
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 2.1324228028503565e-05,
+      "loss": 0.0303,
+      "step": 14255
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 2.131828978622328e-05,
+      "loss": 0.0394,
+      "step": 14256
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 2.1312351543942992e-05,
+      "loss": 0.0344,
+      "step": 14257
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 2.130641330166271e-05,
+      "loss": 0.0268,
+      "step": 14258
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 2.1300475059382426e-05,
+      "loss": 0.0776,
+      "step": 14259
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 2.129453681710214e-05,
+      "loss": 0.0288,
+      "step": 14260
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 2.1288598574821853e-05,
+      "loss": 0.0217,
+      "step": 14261
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 2.1282660332541567e-05,
+      "loss": 0.027,
+      "step": 14262
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 2.1276722090261284e-05,
+      "loss": 0.0312,
+      "step": 14263
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 2.1270783847981e-05,
+      "loss": 0.0213,
+      "step": 14264
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 2.1264845605700714e-05,
+      "loss": 0.0496,
+      "step": 14265
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 2.1258907363420428e-05,
+      "loss": 0.0605,
+      "step": 14266
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 2.125296912114014e-05,
+      "loss": 0.0403,
+      "step": 14267
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 2.124703087885986e-05,
+      "loss": 0.0345,
+      "step": 14268
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 2.1241092636579575e-05,
+      "loss": 0.0258,
+      "step": 14269
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 2.123515439429929e-05,
+      "loss": 0.0295,
+      "step": 14270
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 2.1229216152019006e-05,
+      "loss": 0.0254,
+      "step": 14271
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 2.1223277909738716e-05,
+      "loss": 0.054,
+      "step": 14272
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 2.1217339667458433e-05,
+      "loss": 0.0488,
+      "step": 14273
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 2.121140142517815e-05,
+      "loss": 0.069,
+      "step": 14274
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 2.1205463182897864e-05,
+      "loss": 0.031,
+      "step": 14275
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 2.119952494061758e-05,
+      "loss": 0.0719,
+      "step": 14276
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 2.119358669833729e-05,
+      "loss": 0.066,
+      "step": 14277
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 2.1187648456057008e-05,
+      "loss": 0.0516,
+      "step": 14278
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 2.1181710213776725e-05,
+      "loss": 0.0351,
+      "step": 14279
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 2.1175771971496438e-05,
+      "loss": 0.0298,
+      "step": 14280
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 2.1169833729216155e-05,
+      "loss": 0.0618,
+      "step": 14281
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 2.1163895486935865e-05,
+      "loss": 0.0352,
+      "step": 14282
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 2.1157957244655582e-05,
+      "loss": 0.0625,
+      "step": 14283
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 2.11520190023753e-05,
+      "loss": 0.0444,
+      "step": 14284
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 2.1146080760095013e-05,
+      "loss": 0.0826,
+      "step": 14285
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 2.114014251781473e-05,
+      "loss": 0.0461,
+      "step": 14286
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 2.113420427553444e-05,
+      "loss": 0.0229,
+      "step": 14287
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 2.1128266033254157e-05,
+      "loss": 0.0677,
+      "step": 14288
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 2.1122327790973874e-05,
+      "loss": 0.036,
+      "step": 14289
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 2.1116389548693588e-05,
+      "loss": 0.0413,
+      "step": 14290
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 2.1110451306413305e-05,
+      "loss": 0.0284,
+      "step": 14291
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 2.1104513064133015e-05,
+      "loss": 0.0408,
+      "step": 14292
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 2.1098574821852732e-05,
+      "loss": 0.048,
+      "step": 14293
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 2.109263657957245e-05,
+      "loss": 0.0225,
+      "step": 14294
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 2.1086698337292162e-05,
+      "loss": 0.0282,
+      "step": 14295
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 2.108076009501188e-05,
+      "loss": 0.0612,
+      "step": 14296
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 2.1074821852731593e-05,
+      "loss": 0.0326,
+      "step": 14297
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 2.1068883610451306e-05,
+      "loss": 0.0392,
+      "step": 14298
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 2.1062945368171023e-05,
+      "loss": 0.0226,
+      "step": 14299
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 2.1057007125890737e-05,
+      "loss": 0.0304,
+      "step": 14300
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 2.1051068883610454e-05,
+      "loss": 0.0247,
+      "step": 14301
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 2.1045130641330167e-05,
+      "loss": 0.0391,
+      "step": 14302
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 2.103919239904988e-05,
+      "loss": 0.0439,
+      "step": 14303
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 2.1033254156769598e-05,
+      "loss": 0.0393,
+      "step": 14304
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 2.102731591448931e-05,
+      "loss": 0.0269,
+      "step": 14305
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 2.102137767220903e-05,
+      "loss": 0.0241,
+      "step": 14306
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 2.1015439429928742e-05,
+      "loss": 0.0483,
+      "step": 14307
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 2.1009501187648456e-05,
+      "loss": 0.056,
+      "step": 14308
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 2.1003562945368173e-05,
+      "loss": 0.0357,
+      "step": 14309
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 2.0997624703087886e-05,
+      "loss": 0.0339,
+      "step": 14310
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 2.0991686460807603e-05,
+      "loss": 0.0354,
+      "step": 14311
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 2.0985748218527317e-05,
+      "loss": 0.0283,
+      "step": 14312
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 2.097980997624703e-05,
+      "loss": 0.0462,
+      "step": 14313
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 2.0973871733966747e-05,
+      "loss": 0.0217,
+      "step": 14314
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 2.096793349168646e-05,
+      "loss": 0.0282,
+      "step": 14315
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 2.0961995249406178e-05,
+      "loss": 0.0411,
+      "step": 14316
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 2.095605700712589e-05,
+      "loss": 0.0299,
+      "step": 14317
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 2.0950118764845605e-05,
+      "loss": 0.0219,
+      "step": 14318
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 2.0944180522565322e-05,
+      "loss": 0.0093,
+      "step": 14319
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 2.0938242280285036e-05,
+      "loss": 0.0274,
+      "step": 14320
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 2.0932304038004753e-05,
+      "loss": 0.0373,
+      "step": 14321
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 2.0926365795724466e-05,
+      "loss": 0.0669,
+      "step": 14322
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 2.092042755344418e-05,
+      "loss": 0.0393,
+      "step": 14323
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 2.0914489311163897e-05,
+      "loss": 0.0534,
+      "step": 14324
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 2.090855106888361e-05,
+      "loss": 0.0379,
+      "step": 14325
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 2.0902612826603327e-05,
+      "loss": 0.0418,
+      "step": 14326
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 2.089667458432304e-05,
+      "loss": 0.0436,
+      "step": 14327
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 2.0890736342042758e-05,
+      "loss": 0.0404,
+      "step": 14328
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 2.088479809976247e-05,
+      "loss": 0.04,
+      "step": 14329
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 2.0878859857482185e-05,
+      "loss": 0.0431,
+      "step": 14330
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 2.0872921615201902e-05,
+      "loss": 0.054,
+      "step": 14331
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 2.0866983372921616e-05,
+      "loss": 0.0266,
+      "step": 14332
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 2.0861045130641333e-05,
+      "loss": 0.0261,
+      "step": 14333
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 2.0855106888361046e-05,
+      "loss": 0.0469,
+      "step": 14334
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 2.084916864608076e-05,
+      "loss": 0.0299,
+      "step": 14335
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 2.0843230403800477e-05,
+      "loss": 0.0484,
+      "step": 14336
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 2.083729216152019e-05,
+      "loss": 0.023,
+      "step": 14337
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 2.0831353919239907e-05,
+      "loss": 0.0307,
+      "step": 14338
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 2.082541567695962e-05,
+      "loss": 0.0332,
+      "step": 14339
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 2.0819477434679334e-05,
+      "loss": 0.0746,
+      "step": 14340
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 2.081353919239905e-05,
+      "loss": 0.0213,
+      "step": 14341
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 2.0807600950118765e-05,
+      "loss": 0.0499,
+      "step": 14342
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 2.0801662707838482e-05,
+      "loss": 0.047,
+      "step": 14343
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 2.0795724465558195e-05,
+      "loss": 0.0293,
+      "step": 14344
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 2.078978622327791e-05,
+      "loss": 0.0392,
+      "step": 14345
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 2.0783847980997626e-05,
+      "loss": 0.0262,
+      "step": 14346
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 2.077790973871734e-05,
+      "loss": 0.0393,
+      "step": 14347
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 2.0771971496437057e-05,
+      "loss": 0.0393,
+      "step": 14348
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 2.076603325415677e-05,
+      "loss": 0.0699,
+      "step": 14349
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 2.0760095011876484e-05,
+      "loss": 0.0287,
+      "step": 14350
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 2.07541567695962e-05,
+      "loss": 0.018,
+      "step": 14351
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 2.0748218527315914e-05,
+      "loss": 0.0557,
+      "step": 14352
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 2.074228028503563e-05,
+      "loss": 0.0491,
+      "step": 14353
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 2.0736342042755348e-05,
+      "loss": 0.0337,
+      "step": 14354
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 2.073040380047506e-05,
+      "loss": 0.0333,
+      "step": 14355
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 2.0724465558194775e-05,
+      "loss": 0.0219,
+      "step": 14356
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 2.071852731591449e-05,
+      "loss": 0.042,
+      "step": 14357
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 2.0712589073634206e-05,
+      "loss": 0.0383,
+      "step": 14358
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 2.0706650831353923e-05,
+      "loss": 0.0267,
+      "step": 14359
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 2.0700712589073633e-05,
+      "loss": 0.0248,
+      "step": 14360
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 2.069477434679335e-05,
+      "loss": 0.0349,
+      "step": 14361
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 2.0688836104513064e-05,
+      "loss": 0.0544,
+      "step": 14362
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 2.068289786223278e-05,
+      "loss": 0.0473,
+      "step": 14363
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 2.0676959619952498e-05,
+      "loss": 0.0357,
+      "step": 14364
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 2.0671021377672208e-05,
+      "loss": 0.0343,
+      "step": 14365
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 2.0665083135391925e-05,
+      "loss": 0.0251,
+      "step": 14366
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 2.065914489311164e-05,
+      "loss": 0.0223,
+      "step": 14367
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 2.0653206650831355e-05,
+      "loss": 0.0195,
+      "step": 14368
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 2.0647268408551072e-05,
+      "loss": 0.0437,
+      "step": 14369
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 2.0641330166270782e-05,
+      "loss": 0.011,
+      "step": 14370
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 2.06353919239905e-05,
+      "loss": 0.0343,
+      "step": 14371
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 2.0629453681710213e-05,
+      "loss": 0.0612,
+      "step": 14372
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 2.062351543942993e-05,
+      "loss": 0.0655,
+      "step": 14373
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 2.0617577197149647e-05,
+      "loss": 0.0663,
+      "step": 14374
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 2.0611638954869357e-05,
+      "loss": 0.0268,
+      "step": 14375
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 2.0605700712589074e-05,
+      "loss": 0.0335,
+      "step": 14376
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 2.0599762470308788e-05,
+      "loss": 0.0573,
+      "step": 14377
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 2.0593824228028505e-05,
+      "loss": 0.0346,
+      "step": 14378
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 2.058788598574822e-05,
+      "loss": 0.0349,
+      "step": 14379
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 2.0581947743467935e-05,
+      "loss": 0.0352,
+      "step": 14380
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 2.057600950118765e-05,
+      "loss": 0.0583,
+      "step": 14381
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 2.0570071258907366e-05,
+      "loss": 0.0331,
+      "step": 14382
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 2.056413301662708e-05,
+      "loss": 0.0265,
+      "step": 14383
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 2.0558194774346796e-05,
+      "loss": 0.0263,
+      "step": 14384
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 2.055225653206651e-05,
+      "loss": 0.0312,
+      "step": 14385
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 2.0546318289786223e-05,
+      "loss": 0.0365,
+      "step": 14386
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 2.054038004750594e-05,
+      "loss": 0.038,
+      "step": 14387
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 2.0534441805225654e-05,
+      "loss": 0.0226,
+      "step": 14388
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 2.052850356294537e-05,
+      "loss": 0.0553,
+      "step": 14389
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 2.0522565320665085e-05,
+      "loss": 0.0538,
+      "step": 14390
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 2.0516627078384798e-05,
+      "loss": 0.0291,
+      "step": 14391
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 2.0510688836104515e-05,
+      "loss": 0.0423,
+      "step": 14392
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 2.050475059382423e-05,
+      "loss": 0.041,
+      "step": 14393
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 2.0498812351543946e-05,
+      "loss": 0.0349,
+      "step": 14394
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 2.049287410926366e-05,
+      "loss": 0.0443,
+      "step": 14395
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 2.0486935866983373e-05,
+      "loss": 0.0477,
+      "step": 14396
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 2.048099762470309e-05,
+      "loss": 0.0393,
+      "step": 14397
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 2.0475059382422803e-05,
+      "loss": 0.0254,
+      "step": 14398
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 2.046912114014252e-05,
+      "loss": 0.0384,
+      "step": 14399
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 2.0463182897862234e-05,
+      "loss": 0.0263,
+      "step": 14400
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 2.0457244655581948e-05,
+      "loss": 0.0242,
+      "step": 14401
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 2.0451306413301664e-05,
+      "loss": 0.0149,
+      "step": 14402
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 2.0445368171021378e-05,
+      "loss": 0.0283,
+      "step": 14403
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 2.0439429928741095e-05,
+      "loss": 0.0299,
+      "step": 14404
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 2.043349168646081e-05,
+      "loss": 0.0648,
+      "step": 14405
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 2.0427553444180522e-05,
+      "loss": 0.0348,
+      "step": 14406
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 2.042161520190024e-05,
+      "loss": 0.0333,
+      "step": 14407
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 2.0415676959619953e-05,
+      "loss": 0.032,
+      "step": 14408
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 2.040973871733967e-05,
+      "loss": 0.0405,
+      "step": 14409
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 2.0403800475059383e-05,
+      "loss": 0.0351,
+      "step": 14410
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 2.03978622327791e-05,
+      "loss": 0.0401,
+      "step": 14411
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 2.0391923990498814e-05,
+      "loss": 0.0189,
+      "step": 14412
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 2.0385985748218527e-05,
+      "loss": 0.0605,
+      "step": 14413
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 2.0380047505938244e-05,
+      "loss": 0.0975,
+      "step": 14414
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 2.0374109263657958e-05,
+      "loss": 0.0194,
+      "step": 14415
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 2.0368171021377675e-05,
+      "loss": 0.0256,
+      "step": 14416
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 2.036223277909739e-05,
+      "loss": 0.0378,
+      "step": 14417
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 2.0356294536817102e-05,
+      "loss": 0.0304,
+      "step": 14418
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 2.035035629453682e-05,
+      "loss": 0.0123,
+      "step": 14419
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 2.0344418052256533e-05,
+      "loss": 0.02,
+      "step": 14420
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 2.033847980997625e-05,
+      "loss": 0.0165,
+      "step": 14421
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 2.0332541567695963e-05,
+      "loss": 0.0744,
+      "step": 14422
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 2.0326603325415677e-05,
+      "loss": 0.0309,
+      "step": 14423
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 2.0320665083135394e-05,
+      "loss": 0.0497,
+      "step": 14424
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 2.0314726840855107e-05,
+      "loss": 0.0355,
+      "step": 14425
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 2.0308788598574824e-05,
+      "loss": 0.0455,
+      "step": 14426
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 2.0302850356294538e-05,
+      "loss": 0.0493,
+      "step": 14427
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 2.029691211401425e-05,
+      "loss": 0.0484,
+      "step": 14428
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 2.029097387173397e-05,
+      "loss": 0.0589,
+      "step": 14429
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 2.0285035629453682e-05,
+      "loss": 0.0496,
+      "step": 14430
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 2.02790973871734e-05,
+      "loss": 0.0307,
+      "step": 14431
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 2.0273159144893113e-05,
+      "loss": 0.0577,
+      "step": 14432
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 2.0267220902612826e-05,
+      "loss": 0.0383,
+      "step": 14433
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 2.0261282660332543e-05,
+      "loss": 0.0434,
+      "step": 14434
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 2.0255344418052257e-05,
+      "loss": 0.0222,
+      "step": 14435
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 2.0249406175771974e-05,
+      "loss": 0.0366,
+      "step": 14436
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 2.0243467933491687e-05,
+      "loss": 0.0333,
+      "step": 14437
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 2.02375296912114e-05,
+      "loss": 0.0185,
+      "step": 14438
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 2.0231591448931118e-05,
+      "loss": 0.042,
+      "step": 14439
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 2.022565320665083e-05,
+      "loss": 0.0547,
+      "step": 14440
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 2.021971496437055e-05,
+      "loss": 0.0396,
+      "step": 14441
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 2.0213776722090262e-05,
+      "loss": 0.0415,
+      "step": 14442
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 2.0207838479809975e-05,
+      "loss": 0.0256,
+      "step": 14443
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 2.0201900237529692e-05,
+      "loss": 0.0495,
+      "step": 14444
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 2.0195961995249406e-05,
+      "loss": 0.0586,
+      "step": 14445
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 2.0190023752969123e-05,
+      "loss": 0.0424,
+      "step": 14446
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 2.0184085510688837e-05,
+      "loss": 0.0407,
+      "step": 14447
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 2.017814726840855e-05,
+      "loss": 0.0279,
+      "step": 14448
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 2.0172209026128267e-05,
+      "loss": 0.0413,
+      "step": 14449
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 2.016627078384798e-05,
+      "loss": 0.0263,
+      "step": 14450
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 2.0160332541567698e-05,
+      "loss": 0.0289,
+      "step": 14451
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 2.015439429928741e-05,
+      "loss": 0.0238,
+      "step": 14452
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 2.0148456057007125e-05,
+      "loss": 0.0368,
+      "step": 14453
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 2.0142517814726842e-05,
+      "loss": 0.0307,
+      "step": 14454
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 2.0136579572446555e-05,
+      "loss": 0.0295,
+      "step": 14455
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 2.0130641330166272e-05,
+      "loss": 0.0408,
+      "step": 14456
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 2.012470308788599e-05,
+      "loss": 0.0577,
+      "step": 14457
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 2.01187648456057e-05,
+      "loss": 0.0299,
+      "step": 14458
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 2.0112826603325416e-05,
+      "loss": 0.0342,
+      "step": 14459
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 2.010688836104513e-05,
+      "loss": 0.038,
+      "step": 14460
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 2.0100950118764847e-05,
+      "loss": 0.0248,
+      "step": 14461
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 2.0095011876484564e-05,
+      "loss": 0.0341,
+      "step": 14462
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 2.0089073634204274e-05,
+      "loss": 0.0294,
+      "step": 14463
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 2.008313539192399e-05,
+      "loss": 0.0139,
+      "step": 14464
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 2.0077197149643705e-05,
+      "loss": 0.0465,
+      "step": 14465
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 2.0071258907363422e-05,
+      "loss": 0.0418,
+      "step": 14466
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 2.006532066508314e-05,
+      "loss": 0.0518,
+      "step": 14467
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 2.0059382422802852e-05,
+      "loss": 0.0466,
+      "step": 14468
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 2.0053444180522566e-05,
+      "loss": 0.0052,
+      "step": 14469
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 2.004750593824228e-05,
+      "loss": 0.0665,
+      "step": 14470
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 2.0041567695961996e-05,
+      "loss": 0.0103,
+      "step": 14471
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 2.0035629453681713e-05,
+      "loss": 0.0203,
+      "step": 14472
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 2.0029691211401427e-05,
+      "loss": 0.0616,
+      "step": 14473
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 2.002375296912114e-05,
+      "loss": 0.0359,
+      "step": 14474
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 2.0017814726840854e-05,
+      "loss": 0.0435,
+      "step": 14475
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 2.001187648456057e-05,
+      "loss": 0.0395,
+      "step": 14476
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 2.0005938242280288e-05,
+      "loss": 0.029,
+      "step": 14477
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0587,
+      "step": 14478
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 1.9994061757719715e-05,
+      "loss": 0.0517,
+      "step": 14479
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 1.998812351543943e-05,
+      "loss": 0.0553,
+      "step": 14480
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 1.9982185273159146e-05,
+      "loss": 0.0364,
+      "step": 14481
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 1.9976247030878863e-05,
+      "loss": 0.0277,
+      "step": 14482
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 1.9970308788598576e-05,
+      "loss": 0.0512,
+      "step": 14483
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 1.996437054631829e-05,
+      "loss": 0.0671,
+      "step": 14484
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 1.9958432304038003e-05,
+      "loss": 0.0551,
+      "step": 14485
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 1.995249406175772e-05,
+      "loss": 0.048,
+      "step": 14486
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 1.9946555819477437e-05,
+      "loss": 0.0467,
+      "step": 14487
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 1.994061757719715e-05,
+      "loss": 0.0263,
+      "step": 14488
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 1.9934679334916865e-05,
+      "loss": 0.0243,
+      "step": 14489
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 1.9928741092636578e-05,
+      "loss": 0.0238,
+      "step": 14490
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 1.9922802850356295e-05,
+      "loss": 0.029,
+      "step": 14491
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 1.9916864608076012e-05,
+      "loss": 0.0334,
+      "step": 14492
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 1.9910926365795726e-05,
+      "loss": 0.0431,
+      "step": 14493
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 1.9904988123515443e-05,
+      "loss": 0.0354,
+      "step": 14494
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 1.9899049881235153e-05,
+      "loss": 0.0313,
+      "step": 14495
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 1.989311163895487e-05,
+      "loss": 0.0556,
+      "step": 14496
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 1.9887173396674587e-05,
+      "loss": 0.0384,
+      "step": 14497
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 1.98812351543943e-05,
+      "loss": 0.0243,
+      "step": 14498
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 1.9875296912114017e-05,
+      "loss": 0.0321,
+      "step": 14499
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 1.9869358669833728e-05,
+      "loss": 0.0277,
+      "step": 14500
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 1.9863420427553444e-05,
+      "loss": 0.0363,
+      "step": 14501
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 1.985748218527316e-05,
+      "loss": 0.0366,
+      "step": 14502
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 1.9851543942992875e-05,
+      "loss": 0.0371,
+      "step": 14503
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 1.9845605700712592e-05,
+      "loss": 0.0286,
+      "step": 14504
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 1.9839667458432306e-05,
+      "loss": 0.0504,
+      "step": 14505
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 1.983372921615202e-05,
+      "loss": 0.0288,
+      "step": 14506
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 1.9827790973871736e-05,
+      "loss": 0.0349,
+      "step": 14507
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 1.982185273159145e-05,
+      "loss": 0.017,
+      "step": 14508
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 1.9815914489311167e-05,
+      "loss": 0.0186,
+      "step": 14509
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 1.980997624703088e-05,
+      "loss": 0.0641,
+      "step": 14510
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 1.9804038004750594e-05,
+      "loss": 0.0691,
+      "step": 14511
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 1.979809976247031e-05,
+      "loss": 0.0189,
+      "step": 14512
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 1.9792161520190024e-05,
+      "loss": 0.0313,
+      "step": 14513
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 1.978622327790974e-05,
+      "loss": 0.0128,
+      "step": 14514
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 1.9780285035629455e-05,
+      "loss": 0.035,
+      "step": 14515
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 1.977434679334917e-05,
+      "loss": 0.0277,
+      "step": 14516
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 1.9768408551068885e-05,
+      "loss": 0.3718,
+      "step": 14517
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 1.97624703087886e-05,
+      "loss": 0.0685,
+      "step": 14518
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 1.9756532066508316e-05,
+      "loss": 0.0228,
+      "step": 14519
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 1.975059382422803e-05,
+      "loss": 0.0703,
+      "step": 14520
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 1.9744655581947743e-05,
+      "loss": 0.2371,
+      "step": 14521
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 1.973871733966746e-05,
+      "loss": 0.0499,
+      "step": 14522
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 1.9732779097387174e-05,
+      "loss": 0.0627,
+      "step": 14523
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 1.972684085510689e-05,
+      "loss": 0.0673,
+      "step": 14524
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 1.9720902612826604e-05,
+      "loss": 0.0529,
+      "step": 14525
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 1.9714964370546318e-05,
+      "loss": 0.0533,
+      "step": 14526
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 1.9709026128266035e-05,
+      "loss": 0.0315,
+      "step": 14527
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 1.970308788598575e-05,
+      "loss": 0.0264,
+      "step": 14528
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 1.9697149643705465e-05,
+      "loss": 0.0508,
+      "step": 14529
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 1.969121140142518e-05,
+      "loss": 0.0586,
+      "step": 14530
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 1.9685273159144893e-05,
+      "loss": 0.0604,
+      "step": 14531
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 1.967933491686461e-05,
+      "loss": 0.0551,
+      "step": 14532
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 1.9673396674584323e-05,
+      "loss": 0.0459,
+      "step": 14533
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 1.966745843230404e-05,
+      "loss": 0.0406,
+      "step": 14534
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 1.9661520190023754e-05,
+      "loss": 0.0347,
+      "step": 14535
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 1.9655581947743467e-05,
+      "loss": 0.0437,
+      "step": 14536
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 1.9649643705463184e-05,
+      "loss": 0.0273,
+      "step": 14537
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 1.9643705463182898e-05,
+      "loss": 0.0355,
+      "step": 14538
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 1.9637767220902615e-05,
+      "loss": 0.0772,
+      "step": 14539
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 1.963182897862233e-05,
+      "loss": 0.0346,
+      "step": 14540
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 1.9625890736342042e-05,
+      "loss": 0.0327,
+      "step": 14541
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 1.961995249406176e-05,
+      "loss": 0.0532,
+      "step": 14542
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 1.9614014251781472e-05,
+      "loss": 0.0669,
+      "step": 14543
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 1.960807600950119e-05,
+      "loss": 0.0504,
+      "step": 14544
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 1.9602137767220903e-05,
+      "loss": 0.0498,
+      "step": 14545
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 1.9596199524940617e-05,
+      "loss": 0.0347,
+      "step": 14546
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 1.9590261282660334e-05,
+      "loss": 0.0344,
+      "step": 14547
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 1.9584323040380047e-05,
+      "loss": 0.0316,
+      "step": 14548
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 1.9578384798099764e-05,
+      "loss": 0.0317,
+      "step": 14549
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 1.9572446555819478e-05,
+      "loss": 0.0289,
+      "step": 14550
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 1.9566508313539195e-05,
+      "loss": 0.0672,
+      "step": 14551
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 1.9560570071258908e-05,
+      "loss": 0.0279,
+      "step": 14552
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 1.9554631828978622e-05,
+      "loss": 0.0152,
+      "step": 14553
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 1.954869358669834e-05,
+      "loss": 0.0652,
+      "step": 14554
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 1.9542755344418052e-05,
+      "loss": 0.0552,
+      "step": 14555
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 1.953681710213777e-05,
+      "loss": 0.0498,
+      "step": 14556
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 1.9530878859857483e-05,
+      "loss": 0.0351,
+      "step": 14557
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 1.9524940617577197e-05,
+      "loss": 0.0156,
+      "step": 14558
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 1.9519002375296913e-05,
+      "loss": 0.0495,
+      "step": 14559
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 1.9513064133016627e-05,
+      "loss": 0.0286,
+      "step": 14560
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 1.9507125890736344e-05,
+      "loss": 0.0439,
+      "step": 14561
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 1.9501187648456058e-05,
+      "loss": 0.0418,
+      "step": 14562
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 1.949524940617577e-05,
+      "loss": 0.0381,
+      "step": 14563
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 1.9489311163895488e-05,
+      "loss": 0.0715,
+      "step": 14564
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 1.9483372921615202e-05,
+      "loss": 0.038,
+      "step": 14565
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 1.947743467933492e-05,
+      "loss": 0.029,
+      "step": 14566
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 1.9471496437054632e-05,
+      "loss": 0.0288,
+      "step": 14567
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 1.9465558194774346e-05,
+      "loss": 0.0373,
+      "step": 14568
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 1.9459619952494063e-05,
+      "loss": 0.0601,
+      "step": 14569
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 1.9453681710213776e-05,
+      "loss": 0.0105,
+      "step": 14570
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 1.9447743467933493e-05,
+      "loss": 0.0123,
+      "step": 14571
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 1.9441805225653207e-05,
+      "loss": 0.0467,
+      "step": 14572
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 1.943586698337292e-05,
+      "loss": 0.0371,
+      "step": 14573
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 1.9429928741092638e-05,
+      "loss": 0.0799,
+      "step": 14574
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 1.942399049881235e-05,
+      "loss": 0.0621,
+      "step": 14575
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 1.9418052256532068e-05,
+      "loss": 0.044,
+      "step": 14576
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 1.9412114014251785e-05,
+      "loss": 0.0326,
+      "step": 14577
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 1.9406175771971495e-05,
+      "loss": 0.0532,
+      "step": 14578
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 1.9400237529691212e-05,
+      "loss": 0.0352,
+      "step": 14579
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 1.939429928741093e-05,
+      "loss": 0.0201,
+      "step": 14580
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 1.9388361045130643e-05,
+      "loss": 0.0657,
+      "step": 14581
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 1.938242280285036e-05,
+      "loss": 0.0721,
+      "step": 14582
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 1.937648456057007e-05,
+      "loss": 0.0355,
+      "step": 14583
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 1.9370546318289787e-05,
+      "loss": 0.0397,
+      "step": 14584
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 1.9364608076009504e-05,
+      "loss": 0.0651,
+      "step": 14585
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 1.9358669833729217e-05,
+      "loss": 0.0222,
+      "step": 14586
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 1.9352731591448934e-05,
+      "loss": 0.0378,
+      "step": 14587
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 1.9346793349168645e-05,
+      "loss": 0.0585,
+      "step": 14588
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 1.934085510688836e-05,
+      "loss": 0.0476,
+      "step": 14589
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 1.933491686460808e-05,
+      "loss": 0.0452,
+      "step": 14590
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 1.9328978622327792e-05,
+      "loss": 0.027,
+      "step": 14591
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 1.932304038004751e-05,
+      "loss": 0.0283,
+      "step": 14592
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 1.931710213776722e-05,
+      "loss": 0.0465,
+      "step": 14593
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 1.9311163895486936e-05,
+      "loss": 0.0213,
+      "step": 14594
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 1.9305225653206653e-05,
+      "loss": 0.0266,
+      "step": 14595
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 1.9299287410926367e-05,
+      "loss": 0.0335,
+      "step": 14596
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 1.9293349168646084e-05,
+      "loss": 0.0318,
+      "step": 14597
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 1.9287410926365794e-05,
+      "loss": 0.0272,
+      "step": 14598
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 1.928147268408551e-05,
+      "loss": 0.0208,
+      "step": 14599
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 1.9275534441805228e-05,
+      "loss": 0.0383,
+      "step": 14600
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 1.926959619952494e-05,
+      "loss": 0.0257,
+      "step": 14601
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 1.926365795724466e-05,
+      "loss": 0.0299,
+      "step": 14602
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 1.925771971496437e-05,
+      "loss": 0.0275,
+      "step": 14603
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 1.9251781472684086e-05,
+      "loss": 0.0273,
+      "step": 14604
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 1.9245843230403803e-05,
+      "loss": 0.042,
+      "step": 14605
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 1.9239904988123516e-05,
+      "loss": 0.0211,
+      "step": 14606
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 1.9233966745843233e-05,
+      "loss": 0.0385,
+      "step": 14607
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 1.9228028503562947e-05,
+      "loss": 0.0428,
+      "step": 14608
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 1.922209026128266e-05,
+      "loss": 0.0386,
+      "step": 14609
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 1.9216152019002377e-05,
+      "loss": 0.0247,
+      "step": 14610
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 1.921021377672209e-05,
+      "loss": 0.0729,
+      "step": 14611
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 1.9204275534441808e-05,
+      "loss": 0.0425,
+      "step": 14612
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 1.919833729216152e-05,
+      "loss": 0.0431,
+      "step": 14613
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 1.9192399049881235e-05,
+      "loss": 0.0441,
+      "step": 14614
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 1.9186460807600952e-05,
+      "loss": 0.031,
+      "step": 14615
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 1.9180522565320666e-05,
+      "loss": 0.0264,
+      "step": 14616
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 1.9174584323040382e-05,
+      "loss": 0.0296,
+      "step": 14617
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 1.9168646080760096e-05,
+      "loss": 0.0178,
+      "step": 14618
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 1.916270783847981e-05,
+      "loss": 0.0127,
+      "step": 14619
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 1.9156769596199527e-05,
+      "loss": 0.0307,
+      "step": 14620
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 1.915083135391924e-05,
+      "loss": 0.0996,
+      "step": 14621
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 1.9144893111638957e-05,
+      "loss": 0.1372,
+      "step": 14622
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 1.913895486935867e-05,
+      "loss": 0.0702,
+      "step": 14623
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 1.9133016627078384e-05,
+      "loss": 0.0442,
+      "step": 14624
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 1.91270783847981e-05,
+      "loss": 0.0535,
+      "step": 14625
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 1.9121140142517815e-05,
+      "loss": 0.0512,
+      "step": 14626
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 1.9115201900237532e-05,
+      "loss": 0.0298,
+      "step": 14627
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 1.9109263657957245e-05,
+      "loss": 0.035,
+      "step": 14628
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 1.910332541567696e-05,
+      "loss": 0.0443,
+      "step": 14629
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 1.9097387173396676e-05,
+      "loss": 0.0392,
+      "step": 14630
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 1.909144893111639e-05,
+      "loss": 0.0391,
+      "step": 14631
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 1.9085510688836107e-05,
+      "loss": 0.0312,
+      "step": 14632
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 1.907957244655582e-05,
+      "loss": 0.0365,
+      "step": 14633
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 1.9073634204275537e-05,
+      "loss": 0.0284,
+      "step": 14634
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 1.906769596199525e-05,
+      "loss": 0.0333,
+      "step": 14635
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 1.9061757719714964e-05,
+      "loss": 0.0551,
+      "step": 14636
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 1.905581947743468e-05,
+      "loss": 0.0514,
+      "step": 14637
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 1.9049881235154395e-05,
+      "loss": 0.0376,
+      "step": 14638
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 1.9043942992874112e-05,
+      "loss": 0.0263,
+      "step": 14639
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 1.9038004750593825e-05,
+      "loss": 0.0484,
+      "step": 14640
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 1.903206650831354e-05,
+      "loss": 0.0391,
+      "step": 14641
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 1.9026128266033256e-05,
+      "loss": 0.032,
+      "step": 14642
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 1.902019002375297e-05,
+      "loss": 0.0464,
+      "step": 14643
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 1.9014251781472686e-05,
+      "loss": 0.0432,
+      "step": 14644
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 1.90083135391924e-05,
+      "loss": 0.0485,
+      "step": 14645
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 1.9002375296912114e-05,
+      "loss": 0.0239,
+      "step": 14646
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 1.899643705463183e-05,
+      "loss": 0.0608,
+      "step": 14647
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 1.8990498812351544e-05,
+      "loss": 0.1005,
+      "step": 14648
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 1.898456057007126e-05,
+      "loss": 0.0266,
+      "step": 14649
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 1.8978622327790975e-05,
+      "loss": 0.0615,
+      "step": 14650
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 1.8972684085510688e-05,
+      "loss": 0.0384,
+      "step": 14651
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 1.8966745843230405e-05,
+      "loss": 0.0291,
+      "step": 14652
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 1.896080760095012e-05,
+      "loss": 0.0209,
+      "step": 14653
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 1.8954869358669836e-05,
+      "loss": 0.0394,
+      "step": 14654
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 1.894893111638955e-05,
+      "loss": 0.0385,
+      "step": 14655
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 1.8942992874109263e-05,
+      "loss": 0.0266,
+      "step": 14656
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 1.893705463182898e-05,
+      "loss": 0.0328,
+      "step": 14657
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 1.8931116389548693e-05,
+      "loss": 0.0462,
+      "step": 14658
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 1.892517814726841e-05,
+      "loss": 0.0099,
+      "step": 14659
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 1.8919239904988124e-05,
+      "loss": 0.0417,
+      "step": 14660
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 1.8913301662707838e-05,
+      "loss": 0.0182,
+      "step": 14661
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 1.8907363420427555e-05,
+      "loss": 0.0246,
+      "step": 14662
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 1.8901425178147268e-05,
+      "loss": 0.0526,
+      "step": 14663
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 1.8895486935866985e-05,
+      "loss": 0.0383,
+      "step": 14664
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 1.8889548693586702e-05,
+      "loss": 0.0298,
+      "step": 14665
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 1.8883610451306412e-05,
+      "loss": 0.0324,
+      "step": 14666
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 1.887767220902613e-05,
+      "loss": 0.0356,
+      "step": 14667
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 1.8871733966745843e-05,
+      "loss": 0.0299,
+      "step": 14668
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 1.886579572446556e-05,
+      "loss": 0.0302,
+      "step": 14669
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 1.8859857482185277e-05,
+      "loss": 0.0177,
+      "step": 14670
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 1.8853919239904987e-05,
+      "loss": 0.0238,
+      "step": 14671
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 1.8847980997624704e-05,
+      "loss": 0.0231,
+      "step": 14672
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 1.8842042755344418e-05,
+      "loss": 0.0496,
+      "step": 14673
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 1.8836104513064134e-05,
+      "loss": 0.046,
+      "step": 14674
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 1.883016627078385e-05,
+      "loss": 0.0483,
+      "step": 14675
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 1.882422802850356e-05,
+      "loss": 0.0508,
+      "step": 14676
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 1.881828978622328e-05,
+      "loss": 0.032,
+      "step": 14677
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 1.8812351543942992e-05,
+      "loss": 0.0335,
+      "step": 14678
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 1.880641330166271e-05,
+      "loss": 0.053,
+      "step": 14679
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 1.8800475059382426e-05,
+      "loss": 0.054,
+      "step": 14680
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 1.8794536817102136e-05,
+      "loss": 0.0412,
+      "step": 14681
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 1.8788598574821853e-05,
+      "loss": 0.0632,
+      "step": 14682
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 1.8782660332541567e-05,
+      "loss": 0.0456,
+      "step": 14683
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 1.8776722090261284e-05,
+      "loss": 0.0476,
+      "step": 14684
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 1.8770783847981e-05,
+      "loss": 0.0323,
+      "step": 14685
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 1.876484560570071e-05,
+      "loss": 0.0469,
+      "step": 14686
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 1.8758907363420428e-05,
+      "loss": 0.0593,
+      "step": 14687
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 1.875296912114014e-05,
+      "loss": 0.0529,
+      "step": 14688
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 1.874703087885986e-05,
+      "loss": 0.0486,
+      "step": 14689
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 1.8741092636579576e-05,
+      "loss": 0.039,
+      "step": 14690
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 1.873515439429929e-05,
+      "loss": 0.0343,
+      "step": 14691
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 1.8729216152019003e-05,
+      "loss": 0.035,
+      "step": 14692
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 1.8723277909738716e-05,
+      "loss": 0.0268,
+      "step": 14693
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 1.8717339667458433e-05,
+      "loss": 0.0156,
+      "step": 14694
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 1.871140142517815e-05,
+      "loss": 0.0276,
+      "step": 14695
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 1.8705463182897864e-05,
+      "loss": 0.0234,
+      "step": 14696
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 1.8699524940617577e-05,
+      "loss": 0.0353,
+      "step": 14697
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 1.869358669833729e-05,
+      "loss": 0.0633,
+      "step": 14698
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 1.8687648456057008e-05,
+      "loss": 0.0385,
+      "step": 14699
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 1.8681710213776725e-05,
+      "loss": 0.046,
+      "step": 14700
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 1.867577197149644e-05,
+      "loss": 0.0253,
+      "step": 14701
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 1.8669833729216152e-05,
+      "loss": 0.0566,
+      "step": 14702
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 1.866389548693587e-05,
+      "loss": 0.0282,
+      "step": 14703
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 1.8657957244655583e-05,
+      "loss": 0.0314,
+      "step": 14704
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 1.86520190023753e-05,
+      "loss": 0.0267,
+      "step": 14705
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 1.8646080760095013e-05,
+      "loss": 0.0359,
+      "step": 14706
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 1.8640142517814727e-05,
+      "loss": 0.0319,
+      "step": 14707
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 1.8634204275534444e-05,
+      "loss": 0.0439,
+      "step": 14708
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 1.8628266033254157e-05,
+      "loss": 0.0371,
+      "step": 14709
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 1.8622327790973874e-05,
+      "loss": 0.0261,
+      "step": 14710
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 1.8616389548693588e-05,
+      "loss": 0.0199,
+      "step": 14711
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 1.86104513064133e-05,
+      "loss": 0.0683,
+      "step": 14712
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 1.860451306413302e-05,
+      "loss": 0.0613,
+      "step": 14713
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 1.8598574821852732e-05,
+      "loss": 0.0136,
+      "step": 14714
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 1.859263657957245e-05,
+      "loss": 0.0312,
+      "step": 14715
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 1.8586698337292162e-05,
+      "loss": 0.0252,
+      "step": 14716
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 1.858076009501188e-05,
+      "loss": 0.0319,
+      "step": 14717
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 1.8574821852731593e-05,
+      "loss": 0.0369,
+      "step": 14718
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 1.8568883610451307e-05,
+      "loss": 0.0277,
+      "step": 14719
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 1.8562945368171024e-05,
+      "loss": 0.031,
+      "step": 14720
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 1.8557007125890737e-05,
+      "loss": 0.0542,
+      "step": 14721
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 1.8551068883610454e-05,
+      "loss": 0.0332,
+      "step": 14722
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 1.8545130641330168e-05,
+      "loss": 0.0547,
+      "step": 14723
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 1.853919239904988e-05,
+      "loss": 0.0295,
+      "step": 14724
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 1.8533254156769598e-05,
+      "loss": 0.0359,
+      "step": 14725
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 1.8527315914489312e-05,
+      "loss": 0.0669,
+      "step": 14726
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 1.852137767220903e-05,
+      "loss": 0.0288,
+      "step": 14727
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 1.8515439429928742e-05,
+      "loss": 0.0431,
+      "step": 14728
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 1.8509501187648456e-05,
+      "loss": 0.0316,
+      "step": 14729
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 1.8503562945368173e-05,
+      "loss": 0.0204,
+      "step": 14730
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 1.8497624703087887e-05,
+      "loss": 0.032,
+      "step": 14731
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 1.8491686460807603e-05,
+      "loss": 0.0345,
+      "step": 14732
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 1.8485748218527317e-05,
+      "loss": 0.0429,
+      "step": 14733
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 1.847980997624703e-05,
+      "loss": 0.0528,
+      "step": 14734
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 1.8473871733966748e-05,
+      "loss": 0.0585,
+      "step": 14735
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 1.846793349168646e-05,
+      "loss": 0.0365,
+      "step": 14736
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 1.8461995249406178e-05,
+      "loss": 0.0407,
+      "step": 14737
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 1.8456057007125892e-05,
+      "loss": 0.0417,
+      "step": 14738
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 1.8450118764845605e-05,
+      "loss": 0.0225,
+      "step": 14739
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 1.8444180522565322e-05,
+      "loss": 0.0294,
+      "step": 14740
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 1.8438242280285036e-05,
+      "loss": 0.0181,
+      "step": 14741
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 1.8432304038004753e-05,
+      "loss": 0.0238,
+      "step": 14742
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 1.8426365795724466e-05,
+      "loss": 0.0491,
+      "step": 14743
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 1.842042755344418e-05,
+      "loss": 0.0439,
+      "step": 14744
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 1.8414489311163897e-05,
+      "loss": 0.0231,
+      "step": 14745
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 1.840855106888361e-05,
+      "loss": 0.0911,
+      "step": 14746
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 1.8402612826603328e-05,
+      "loss": 0.0207,
+      "step": 14747
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 1.839667458432304e-05,
+      "loss": 0.0574,
+      "step": 14748
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 1.8390736342042755e-05,
+      "loss": 0.029,
+      "step": 14749
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 1.838479809976247e-05,
+      "loss": 0.0405,
+      "step": 14750
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 1.8378859857482185e-05,
+      "loss": 0.0365,
+      "step": 14751
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 1.8372921615201902e-05,
+      "loss": 0.0339,
+      "step": 14752
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 1.8366983372921616e-05,
+      "loss": 0.0412,
+      "step": 14753
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 1.836104513064133e-05,
+      "loss": 0.0439,
+      "step": 14754
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 1.8355106888361046e-05,
+      "loss": 0.0608,
+      "step": 14755
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 1.834916864608076e-05,
+      "loss": 0.0655,
+      "step": 14756
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 1.8343230403800477e-05,
+      "loss": 0.036,
+      "step": 14757
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 1.833729216152019e-05,
+      "loss": 0.0403,
+      "step": 14758
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 1.8331353919239904e-05,
+      "loss": 0.0274,
+      "step": 14759
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 1.832541567695962e-05,
+      "loss": 0.043,
+      "step": 14760
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 1.8319477434679335e-05,
+      "loss": 0.023,
+      "step": 14761
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 1.831353919239905e-05,
+      "loss": 0.0376,
+      "step": 14762
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 1.8307600950118765e-05,
+      "loss": 0.0273,
+      "step": 14763
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 1.830166270783848e-05,
+      "loss": 0.0276,
+      "step": 14764
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 1.8295724465558196e-05,
+      "loss": 0.0379,
+      "step": 14765
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 1.828978622327791e-05,
+      "loss": 0.0318,
+      "step": 14766
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 1.8283847980997626e-05,
+      "loss": 0.0325,
+      "step": 14767
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 1.827790973871734e-05,
+      "loss": 0.042,
+      "step": 14768
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 1.8271971496437053e-05,
+      "loss": 0.0156,
+      "step": 14769
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 1.826603325415677e-05,
+      "loss": 0.0157,
+      "step": 14770
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 1.8260095011876484e-05,
+      "loss": 0.1124,
+      "step": 14771
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 1.82541567695962e-05,
+      "loss": 0.0215,
+      "step": 14772
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 1.8248218527315915e-05,
+      "loss": 0.0677,
+      "step": 14773
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 1.824228028503563e-05,
+      "loss": 0.048,
+      "step": 14774
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 1.8236342042755345e-05,
+      "loss": 0.0477,
+      "step": 14775
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 1.823040380047506e-05,
+      "loss": 0.0406,
+      "step": 14776
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 1.8224465558194776e-05,
+      "loss": 0.0366,
+      "step": 14777
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 1.8218527315914493e-05,
+      "loss": 0.044,
+      "step": 14778
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 1.8212589073634206e-05,
+      "loss": 0.0483,
+      "step": 14779
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 1.820665083135392e-05,
+      "loss": 0.0581,
+      "step": 14780
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 1.8200712589073633e-05,
+      "loss": 0.0395,
+      "step": 14781
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 1.819477434679335e-05,
+      "loss": 0.056,
+      "step": 14782
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 1.8188836104513067e-05,
+      "loss": 0.024,
+      "step": 14783
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 1.818289786223278e-05,
+      "loss": 0.0493,
+      "step": 14784
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 1.8176959619952494e-05,
+      "loss": 0.0239,
+      "step": 14785
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 1.8171021377672208e-05,
+      "loss": 0.037,
+      "step": 14786
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 1.8165083135391925e-05,
+      "loss": 0.0274,
+      "step": 14787
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 1.8159144893111642e-05,
+      "loss": 0.0304,
+      "step": 14788
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 1.8153206650831356e-05,
+      "loss": 0.0411,
+      "step": 14789
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 1.814726840855107e-05,
+      "loss": 0.0293,
+      "step": 14790
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 1.8141330166270783e-05,
+      "loss": 0.0291,
+      "step": 14791
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 1.81353919239905e-05,
+      "loss": 0.0231,
+      "step": 14792
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 1.8129453681710217e-05,
+      "loss": 0.0435,
+      "step": 14793
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 1.812351543942993e-05,
+      "loss": 0.036,
+      "step": 14794
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 1.8117577197149644e-05,
+      "loss": 0.0357,
+      "step": 14795
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 1.8111638954869357e-05,
+      "loss": 0.0354,
+      "step": 14796
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 1.8105700712589074e-05,
+      "loss": 0.0331,
+      "step": 14797
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 1.809976247030879e-05,
+      "loss": 0.0347,
+      "step": 14798
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 1.8093824228028505e-05,
+      "loss": 0.0191,
+      "step": 14799
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 1.808788598574822e-05,
+      "loss": 0.0315,
+      "step": 14800
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 1.8081947743467932e-05,
+      "loss": 0.0674,
+      "step": 14801
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 1.807600950118765e-05,
+      "loss": 0.0183,
+      "step": 14802
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 1.8070071258907366e-05,
+      "loss": 0.0433,
+      "step": 14803
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 1.806413301662708e-05,
+      "loss": 0.0582,
+      "step": 14804
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 1.8058194774346797e-05,
+      "loss": 0.0303,
+      "step": 14805
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 1.8052256532066507e-05,
+      "loss": 0.0236,
+      "step": 14806
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 1.8046318289786224e-05,
+      "loss": 0.0222,
+      "step": 14807
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 1.804038004750594e-05,
+      "loss": 0.0276,
+      "step": 14808
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 1.8034441805225654e-05,
+      "loss": 0.0237,
+      "step": 14809
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 1.802850356294537e-05,
+      "loss": 0.0332,
+      "step": 14810
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 1.802256532066508e-05,
+      "loss": 0.0243,
+      "step": 14811
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 1.80166270783848e-05,
+      "loss": 0.0193,
+      "step": 14812
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 1.8010688836104515e-05,
+      "loss": 0.1775,
+      "step": 14813
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 1.800475059382423e-05,
+      "loss": 0.0346,
+      "step": 14814
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 1.7998812351543946e-05,
+      "loss": 0.0746,
+      "step": 14815
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 1.7992874109263656e-05,
+      "loss": 0.0118,
+      "step": 14816
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 1.7986935866983373e-05,
+      "loss": 0.0448,
+      "step": 14817
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 1.798099762470309e-05,
+      "loss": 0.0332,
+      "step": 14818
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 1.7975059382422804e-05,
+      "loss": 0.0277,
+      "step": 14819
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 1.796912114014252e-05,
+      "loss": 0.0416,
+      "step": 14820
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 1.796318289786223e-05,
+      "loss": 0.0424,
+      "step": 14821
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 1.7957244655581948e-05,
+      "loss": 0.0018,
+      "step": 14822
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 1.7951306413301665e-05,
+      "loss": 0.0507,
+      "step": 14823
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 1.7945368171021378e-05,
+      "loss": 0.056,
+      "step": 14824
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 1.7939429928741095e-05,
+      "loss": 0.0706,
+      "step": 14825
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 1.793349168646081e-05,
+      "loss": 0.0208,
+      "step": 14826
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 1.7927553444180522e-05,
+      "loss": 0.0323,
+      "step": 14827
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 1.792161520190024e-05,
+      "loss": 0.0259,
+      "step": 14828
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 1.7915676959619953e-05,
+      "loss": 0.0353,
+      "step": 14829
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 1.790973871733967e-05,
+      "loss": 0.0213,
+      "step": 14830
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 1.7903800475059384e-05,
+      "loss": 0.0313,
+      "step": 14831
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 1.7897862232779097e-05,
+      "loss": 0.0443,
+      "step": 14832
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 1.7891923990498814e-05,
+      "loss": 0.0381,
+      "step": 14833
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 1.7885985748218528e-05,
+      "loss": 0.0181,
+      "step": 14834
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 1.7880047505938245e-05,
+      "loss": 0.0638,
+      "step": 14835
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 1.7874109263657958e-05,
+      "loss": 0.0401,
+      "step": 14836
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 1.7868171021377672e-05,
+      "loss": 0.0489,
+      "step": 14837
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 1.786223277909739e-05,
+      "loss": 0.0281,
+      "step": 14838
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 1.7856294536817102e-05,
+      "loss": 0.0402,
+      "step": 14839
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 1.785035629453682e-05,
+      "loss": 0.0362,
+      "step": 14840
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 1.7844418052256533e-05,
+      "loss": 0.0369,
+      "step": 14841
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 1.7838479809976246e-05,
+      "loss": 0.038,
+      "step": 14842
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 1.7832541567695963e-05,
+      "loss": 0.0373,
+      "step": 14843
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 1.7826603325415677e-05,
+      "loss": 0.0401,
+      "step": 14844
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 1.7820665083135394e-05,
+      "loss": 0.0256,
+      "step": 14845
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 1.7814726840855108e-05,
+      "loss": 0.0257,
+      "step": 14846
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 1.780878859857482e-05,
+      "loss": 0.0352,
+      "step": 14847
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 1.7802850356294538e-05,
+      "loss": 0.0369,
+      "step": 14848
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 1.779691211401425e-05,
+      "loss": 0.0459,
+      "step": 14849
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 1.779097387173397e-05,
+      "loss": 0.0396,
+      "step": 14850
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 1.7785035629453682e-05,
+      "loss": 0.0149,
+      "step": 14851
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 1.7779097387173396e-05,
+      "loss": 0.0367,
+      "step": 14852
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 1.7773159144893113e-05,
+      "loss": 0.0401,
+      "step": 14853
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 1.7767220902612826e-05,
+      "loss": 0.0503,
+      "step": 14854
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 1.7761282660332543e-05,
+      "loss": 0.0277,
+      "step": 14855
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 1.7755344418052257e-05,
+      "loss": 0.0326,
+      "step": 14856
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 1.7749406175771974e-05,
+      "loss": 0.0411,
+      "step": 14857
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 1.7743467933491687e-05,
+      "loss": 0.0349,
+      "step": 14858
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 1.77375296912114e-05,
+      "loss": 0.0278,
+      "step": 14859
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 1.7731591448931118e-05,
+      "loss": 0.011,
+      "step": 14860
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 1.772565320665083e-05,
+      "loss": 0.0324,
+      "step": 14861
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 1.771971496437055e-05,
+      "loss": 0.0476,
+      "step": 14862
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 1.7713776722090262e-05,
+      "loss": 0.0236,
+      "step": 14863
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 1.7707838479809976e-05,
+      "loss": 0.026,
+      "step": 14864
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 1.7701900237529693e-05,
+      "loss": 0.0303,
+      "step": 14865
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 1.7695961995249406e-05,
+      "loss": 0.0345,
+      "step": 14866
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 1.7690023752969123e-05,
+      "loss": 0.0496,
+      "step": 14867
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 1.7684085510688837e-05,
+      "loss": 0.0123,
+      "step": 14868
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 1.767814726840855e-05,
+      "loss": 0.0183,
+      "step": 14869
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 1.7672209026128267e-05,
+      "loss": 0.027,
+      "step": 14870
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 1.766627078384798e-05,
+      "loss": 0.0082,
+      "step": 14871
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 1.7660332541567698e-05,
+      "loss": 0.0198,
+      "step": 14872
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 1.765439429928741e-05,
+      "loss": 0.0674,
+      "step": 14873
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 1.7648456057007125e-05,
+      "loss": 0.0254,
+      "step": 14874
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 1.7642517814726842e-05,
+      "loss": 0.0824,
+      "step": 14875
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 1.7636579572446556e-05,
+      "loss": 0.0402,
+      "step": 14876
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 1.7630641330166273e-05,
+      "loss": 0.0355,
+      "step": 14877
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 1.7624703087885986e-05,
+      "loss": 0.0551,
+      "step": 14878
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 1.76187648456057e-05,
+      "loss": 0.049,
+      "step": 14879
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 1.7612826603325417e-05,
+      "loss": 0.0487,
+      "step": 14880
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 1.760688836104513e-05,
+      "loss": 0.0398,
+      "step": 14881
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 1.7600950118764847e-05,
+      "loss": 0.0315,
+      "step": 14882
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 1.759501187648456e-05,
+      "loss": 0.0609,
+      "step": 14883
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 1.7589073634204274e-05,
+      "loss": 0.052,
+      "step": 14884
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 1.758313539192399e-05,
+      "loss": 0.0289,
+      "step": 14885
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 1.7577197149643705e-05,
+      "loss": 0.0488,
+      "step": 14886
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 1.7571258907363422e-05,
+      "loss": 0.0336,
+      "step": 14887
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 1.756532066508314e-05,
+      "loss": 0.0208,
+      "step": 14888
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 1.755938242280285e-05,
+      "loss": 0.0425,
+      "step": 14889
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 1.7553444180522566e-05,
+      "loss": 0.0186,
+      "step": 14890
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 1.754750593824228e-05,
+      "loss": 0.0358,
+      "step": 14891
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 1.7541567695961997e-05,
+      "loss": 0.0617,
+      "step": 14892
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 1.7535629453681714e-05,
+      "loss": 0.038,
+      "step": 14893
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 1.7529691211401424e-05,
+      "loss": 0.0271,
+      "step": 14894
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 1.752375296912114e-05,
+      "loss": 0.0492,
+      "step": 14895
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 1.7517814726840854e-05,
+      "loss": 0.0457,
+      "step": 14896
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 1.751187648456057e-05,
+      "loss": 0.0389,
+      "step": 14897
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 1.7505938242280288e-05,
+      "loss": 0.042,
+      "step": 14898
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 1.75e-05,
+      "loss": 0.0365,
+      "step": 14899
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 1.7494061757719715e-05,
+      "loss": 0.0533,
+      "step": 14900
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 1.7488123515439432e-05,
+      "loss": 0.0272,
+      "step": 14901
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 1.7482185273159146e-05,
+      "loss": 0.0288,
+      "step": 14902
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 1.7476247030878863e-05,
+      "loss": 0.0824,
+      "step": 14903
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 1.7470308788598573e-05,
+      "loss": 0.0311,
+      "step": 14904
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 1.746437054631829e-05,
+      "loss": 0.0399,
+      "step": 14905
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 1.7458432304038007e-05,
+      "loss": 0.0326,
+      "step": 14906
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 1.745249406175772e-05,
+      "loss": 0.0225,
+      "step": 14907
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 1.7446555819477438e-05,
+      "loss": 0.013,
+      "step": 14908
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 1.7440617577197148e-05,
+      "loss": 0.0178,
+      "step": 14909
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 1.7434679334916865e-05,
+      "loss": 0.0403,
+      "step": 14910
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 1.7428741092636582e-05,
+      "loss": 0.0624,
+      "step": 14911
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 1.7422802850356295e-05,
+      "loss": 0.027,
+      "step": 14912
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 1.7416864608076012e-05,
+      "loss": 0.0211,
+      "step": 14913
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 1.7410926365795726e-05,
+      "loss": 0.0249,
+      "step": 14914
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 1.740498812351544e-05,
+      "loss": 0.0505,
+      "step": 14915
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 1.7399049881235156e-05,
+      "loss": 0.1384,
+      "step": 14916
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 1.739311163895487e-05,
+      "loss": 0.0209,
+      "step": 14917
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 1.7387173396674587e-05,
+      "loss": 0.0442,
+      "step": 14918
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 1.73812351543943e-05,
+      "loss": 0.0254,
+      "step": 14919
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 1.7375296912114014e-05,
+      "loss": 0.0455,
+      "step": 14920
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 1.736935866983373e-05,
+      "loss": 0.0289,
+      "step": 14921
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 1.7363420427553445e-05,
+      "loss": 0.0778,
+      "step": 14922
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 1.735748218527316e-05,
+      "loss": 0.081,
+      "step": 14923
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 1.7351543942992875e-05,
+      "loss": 0.0439,
+      "step": 14924
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 1.734560570071259e-05,
+      "loss": 0.0486,
+      "step": 14925
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 1.7339667458432306e-05,
+      "loss": 0.0584,
+      "step": 14926
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 1.733372921615202e-05,
+      "loss": 0.0473,
+      "step": 14927
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 1.7327790973871736e-05,
+      "loss": 0.0694,
+      "step": 14928
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 1.732185273159145e-05,
+      "loss": 0.0888,
+      "step": 14929
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 1.7315914489311164e-05,
+      "loss": 0.0513,
+      "step": 14930
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 1.730997624703088e-05,
+      "loss": 0.0383,
+      "step": 14931
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 1.7304038004750594e-05,
+      "loss": 0.0612,
+      "step": 14932
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 1.729809976247031e-05,
+      "loss": 0.0708,
+      "step": 14933
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 1.7292161520190025e-05,
+      "loss": 0.0425,
+      "step": 14934
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 1.7286223277909738e-05,
+      "loss": 0.0359,
+      "step": 14935
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 1.7280285035629455e-05,
+      "loss": 0.0394,
+      "step": 14936
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 1.727434679334917e-05,
+      "loss": 0.0425,
+      "step": 14937
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 1.7268408551068886e-05,
+      "loss": 0.0386,
+      "step": 14938
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 1.72624703087886e-05,
+      "loss": 0.0258,
+      "step": 14939
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 1.7256532066508313e-05,
+      "loss": 0.0351,
+      "step": 14940
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 1.725059382422803e-05,
+      "loss": 0.0609,
+      "step": 14941
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 1.7244655581947743e-05,
+      "loss": 0.0671,
+      "step": 14942
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 1.723871733966746e-05,
+      "loss": 0.0565,
+      "step": 14943
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 1.7232779097387174e-05,
+      "loss": 0.0293,
+      "step": 14944
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 1.722684085510689e-05,
+      "loss": 0.035,
+      "step": 14945
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 1.7220902612826605e-05,
+      "loss": 0.021,
+      "step": 14946
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 1.7214964370546318e-05,
+      "loss": 0.0516,
+      "step": 14947
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 1.7209026128266035e-05,
+      "loss": 0.044,
+      "step": 14948
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 1.720308788598575e-05,
+      "loss": 0.0771,
+      "step": 14949
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 1.7197149643705466e-05,
+      "loss": 0.0325,
+      "step": 14950
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 1.719121140142518e-05,
+      "loss": 0.0466,
+      "step": 14951
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 1.7185273159144893e-05,
+      "loss": 0.0453,
+      "step": 14952
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 1.717933491686461e-05,
+      "loss": 0.0599,
+      "step": 14953
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 1.7173396674584323e-05,
+      "loss": 0.0469,
+      "step": 14954
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 1.716745843230404e-05,
+      "loss": 0.0262,
+      "step": 14955
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 1.7161520190023754e-05,
+      "loss": 0.0583,
+      "step": 14956
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 1.7155581947743467e-05,
+      "loss": 0.031,
+      "step": 14957
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 1.7149643705463184e-05,
+      "loss": 0.0337,
+      "step": 14958
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 1.7143705463182898e-05,
+      "loss": 0.036,
+      "step": 14959
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 1.7137767220902615e-05,
+      "loss": 0.028,
+      "step": 14960
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 1.713182897862233e-05,
+      "loss": 0.0186,
+      "step": 14961
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 1.7125890736342042e-05,
+      "loss": 0.0634,
+      "step": 14962
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 1.711995249406176e-05,
+      "loss": 0.0258,
+      "step": 14963
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 1.7114014251781473e-05,
+      "loss": 0.0471,
+      "step": 14964
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 1.710807600950119e-05,
+      "loss": 0.0346,
+      "step": 14965
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 1.7102137767220903e-05,
+      "loss": 0.0259,
+      "step": 14966
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 1.7096199524940617e-05,
+      "loss": 0.0234,
+      "step": 14967
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 1.7090261282660334e-05,
+      "loss": 0.0549,
+      "step": 14968
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 1.7084323040380047e-05,
+      "loss": 0.009,
+      "step": 14969
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 1.7078384798099764e-05,
+      "loss": 0.0619,
+      "step": 14970
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 1.7072446555819478e-05,
+      "loss": 0.0067,
+      "step": 14971
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 1.706650831353919e-05,
+      "loss": 0.0618,
+      "step": 14972
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 1.706057007125891e-05,
+      "loss": 0.0458,
+      "step": 14973
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 1.7054631828978622e-05,
+      "loss": 0.0658,
+      "step": 14974
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 1.704869358669834e-05,
+      "loss": 0.0304,
+      "step": 14975
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 1.7042755344418056e-05,
+      "loss": 0.0385,
+      "step": 14976
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 1.7036817102137766e-05,
+      "loss": 0.0279,
+      "step": 14977
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 1.7030878859857483e-05,
+      "loss": 0.072,
+      "step": 14978
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 1.7024940617577197e-05,
+      "loss": 0.0228,
+      "step": 14979
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 1.7019002375296914e-05,
+      "loss": 0.0337,
+      "step": 14980
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 1.701306413301663e-05,
+      "loss": 0.0298,
+      "step": 14981
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 1.700712589073634e-05,
+      "loss": 0.0372,
+      "step": 14982
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 1.7001187648456058e-05,
+      "loss": 0.0258,
+      "step": 14983
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 1.699524940617577e-05,
+      "loss": 0.0379,
+      "step": 14984
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 1.698931116389549e-05,
+      "loss": 0.0539,
+      "step": 14985
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 1.6983372921615205e-05,
+      "loss": 0.061,
+      "step": 14986
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 1.6977434679334916e-05,
+      "loss": 0.0472,
+      "step": 14987
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 1.6971496437054633e-05,
+      "loss": 0.0273,
+      "step": 14988
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 1.6965558194774346e-05,
+      "loss": 0.0333,
+      "step": 14989
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 1.6959619952494063e-05,
+      "loss": 0.0411,
+      "step": 14990
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 1.695368171021378e-05,
+      "loss": 0.03,
+      "step": 14991
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 1.694774346793349e-05,
+      "loss": 0.0421,
+      "step": 14992
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 1.6941805225653207e-05,
+      "loss": 0.0501,
+      "step": 14993
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 1.693586698337292e-05,
+      "loss": 0.0261,
+      "step": 14994
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 1.6929928741092638e-05,
+      "loss": 0.0763,
+      "step": 14995
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 1.6923990498812355e-05,
+      "loss": 0.0661,
+      "step": 14996
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 1.6918052256532065e-05,
+      "loss": 0.0291,
+      "step": 14997
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 1.6912114014251782e-05,
+      "loss": 0.0422,
+      "step": 14998
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 1.6906175771971495e-05,
+      "loss": 0.0338,
+      "step": 14999
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 1.6900237529691212e-05,
+      "loss": 0.043,
+      "step": 15000
+    },
+    {
+      "epoch": 16.82,
+      "eval_loss": 1.8534111976623535,
+      "eval_runtime": 900.1062,
+      "eval_samples_per_second": 2.935,
+      "eval_steps_per_second": 0.368,
+      "eval_wer": 1.0096588655295518,
+      "step": 15000
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 1.689429928741093e-05,
+      "loss": 0.0284,
+      "step": 15001
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 1.6888361045130643e-05,
+      "loss": 0.0313,
+      "step": 15002
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 1.6882422802850357e-05,
+      "loss": 0.0298,
+      "step": 15003
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 1.687648456057007e-05,
+      "loss": 0.0335,
+      "step": 15004
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 1.6870546318289787e-05,
+      "loss": 0.0399,
+      "step": 15005
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 1.6864608076009504e-05,
+      "loss": 0.0269,
+      "step": 15006
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 1.6858669833729218e-05,
+      "loss": 0.0318,
+      "step": 15007
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 1.685273159144893e-05,
+      "loss": 0.0261,
+      "step": 15008
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 1.6846793349168645e-05,
+      "loss": 0.0396,
+      "step": 15009
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 1.6840855106888362e-05,
+      "loss": 0.0335,
+      "step": 15010
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 1.683491686460808e-05,
+      "loss": 0.0256,
+      "step": 15011
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 1.6828978622327792e-05,
+      "loss": 0.0571,
+      "step": 15012
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 1.6823040380047506e-05,
+      "loss": 0.0184,
+      "step": 15013
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 1.681710213776722e-05,
+      "loss": 0.0176,
+      "step": 15014
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 1.6811163895486936e-05,
+      "loss": 0.019,
+      "step": 15015
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 1.6805225653206653e-05,
+      "loss": 0.0392,
+      "step": 15016
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 1.6799287410926367e-05,
+      "loss": 0.0246,
+      "step": 15017
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 1.679334916864608e-05,
+      "loss": 0.013,
+      "step": 15018
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 1.6787410926365794e-05,
+      "loss": 0.0464,
+      "step": 15019
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 1.678147268408551e-05,
+      "loss": 0.0171,
+      "step": 15020
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 1.6775534441805228e-05,
+      "loss": 0.0808,
+      "step": 15021
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 1.676959619952494e-05,
+      "loss": 0.0243,
+      "step": 15022
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 1.6763657957244655e-05,
+      "loss": 0.0681,
+      "step": 15023
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 1.6757719714964372e-05,
+      "loss": 0.0541,
+      "step": 15024
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 1.6751781472684086e-05,
+      "loss": 0.0384,
+      "step": 15025
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 1.6745843230403803e-05,
+      "loss": 0.0294,
+      "step": 15026
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 1.6739904988123516e-05,
+      "loss": 0.0359,
+      "step": 15027
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 1.6733966745843233e-05,
+      "loss": 0.0544,
+      "step": 15028
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 1.6728028503562947e-05,
+      "loss": 0.046,
+      "step": 15029
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 1.672209026128266e-05,
+      "loss": 0.0504,
+      "step": 15030
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 1.6716152019002377e-05,
+      "loss": 0.0298,
+      "step": 15031
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 1.671021377672209e-05,
+      "loss": 0.0305,
+      "step": 15032
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 1.6704275534441808e-05,
+      "loss": 0.0753,
+      "step": 15033
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 1.669833729216152e-05,
+      "loss": 0.0581,
+      "step": 15034
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 1.6692399049881235e-05,
+      "loss": 0.0599,
+      "step": 15035
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 1.6686460807600952e-05,
+      "loss": 0.0428,
+      "step": 15036
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 1.6680522565320666e-05,
+      "loss": 0.0423,
+      "step": 15037
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 1.6674584323040383e-05,
+      "loss": 0.0389,
+      "step": 15038
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 1.6668646080760096e-05,
+      "loss": 0.0623,
+      "step": 15039
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 1.666270783847981e-05,
+      "loss": 0.025,
+      "step": 15040
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 1.6656769596199527e-05,
+      "loss": 0.0487,
+      "step": 15041
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 1.665083135391924e-05,
+      "loss": 0.0423,
+      "step": 15042
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 1.6644893111638957e-05,
+      "loss": 0.0349,
+      "step": 15043
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 1.663895486935867e-05,
+      "loss": 0.054,
+      "step": 15044
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 1.6633016627078385e-05,
+      "loss": 0.0412,
+      "step": 15045
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 1.66270783847981e-05,
+      "loss": 0.0391,
+      "step": 15046
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 1.6621140142517815e-05,
+      "loss": 0.0382,
+      "step": 15047
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 1.6615201900237532e-05,
+      "loss": 0.0374,
+      "step": 15048
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 1.6609263657957246e-05,
+      "loss": 0.0293,
+      "step": 15049
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 1.660332541567696e-05,
+      "loss": 0.0171,
+      "step": 15050
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 1.6597387173396676e-05,
+      "loss": 0.0299,
+      "step": 15051
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 1.659144893111639e-05,
+      "loss": 0.0227,
+      "step": 15052
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 1.6585510688836107e-05,
+      "loss": 0.0314,
+      "step": 15053
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 1.657957244655582e-05,
+      "loss": 0.028,
+      "step": 15054
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 1.6573634204275534e-05,
+      "loss": 0.0323,
+      "step": 15055
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 1.656769596199525e-05,
+      "loss": 0.0236,
+      "step": 15056
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 1.6561757719714964e-05,
+      "loss": 0.0223,
+      "step": 15057
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 1.655581947743468e-05,
+      "loss": 0.0267,
+      "step": 15058
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 1.6549881235154395e-05,
+      "loss": 0.031,
+      "step": 15059
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 1.654394299287411e-05,
+      "loss": 0.0544,
+      "step": 15060
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 1.6538004750593826e-05,
+      "loss": 0.042,
+      "step": 15061
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 1.653206650831354e-05,
+      "loss": 0.0261,
+      "step": 15062
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 1.6526128266033256e-05,
+      "loss": 0.0305,
+      "step": 15063
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 1.652019002375297e-05,
+      "loss": 0.0269,
+      "step": 15064
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 1.6514251781472683e-05,
+      "loss": 0.0204,
+      "step": 15065
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 1.65083135391924e-05,
+      "loss": 0.0325,
+      "step": 15066
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 1.6502375296912114e-05,
+      "loss": 0.0286,
+      "step": 15067
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 1.649643705463183e-05,
+      "loss": 0.0446,
+      "step": 15068
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 1.6490498812351544e-05,
+      "loss": 0.0255,
+      "step": 15069
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 1.6484560570071258e-05,
+      "loss": 0.0141,
+      "step": 15070
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 1.6478622327790975e-05,
+      "loss": 0.0929,
+      "step": 15071
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 1.647268408551069e-05,
+      "loss": 0.0603,
+      "step": 15072
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 1.6466745843230405e-05,
+      "loss": 0.0498,
+      "step": 15073
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 1.646080760095012e-05,
+      "loss": 0.0689,
+      "step": 15074
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 1.6454869358669833e-05,
+      "loss": 0.0676,
+      "step": 15075
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 1.644893111638955e-05,
+      "loss": 0.062,
+      "step": 15076
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 1.6442992874109263e-05,
+      "loss": 0.0365,
+      "step": 15077
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 1.643705463182898e-05,
+      "loss": 0.0368,
+      "step": 15078
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 1.6431116389548694e-05,
+      "loss": 0.0444,
+      "step": 15079
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 1.6425178147268407e-05,
+      "loss": 0.0453,
+      "step": 15080
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 1.6419239904988124e-05,
+      "loss": 0.0178,
+      "step": 15081
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 1.6413301662707838e-05,
+      "loss": 0.0178,
+      "step": 15082
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 1.6407363420427555e-05,
+      "loss": 0.0444,
+      "step": 15083
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 1.640142517814727e-05,
+      "loss": 0.0422,
+      "step": 15084
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 1.6395486935866985e-05,
+      "loss": 0.0287,
+      "step": 15085
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 1.63895486935867e-05,
+      "loss": 0.0272,
+      "step": 15086
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 1.6383610451306413e-05,
+      "loss": 0.0304,
+      "step": 15087
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 1.637767220902613e-05,
+      "loss": 0.0381,
+      "step": 15088
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 1.6371733966745843e-05,
+      "loss": 0.031,
+      "step": 15089
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 1.636579572446556e-05,
+      "loss": 0.0507,
+      "step": 15090
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 1.6359857482185274e-05,
+      "loss": 0.035,
+      "step": 15091
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 1.6353919239904987e-05,
+      "loss": 0.0217,
+      "step": 15092
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 1.6347980997624704e-05,
+      "loss": 0.0302,
+      "step": 15093
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 1.6342042755344418e-05,
+      "loss": 0.0412,
+      "step": 15094
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 1.6336104513064135e-05,
+      "loss": 0.0201,
+      "step": 15095
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 1.6330166270783848e-05,
+      "loss": 0.0373,
+      "step": 15096
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 1.6324228028503562e-05,
+      "loss": 0.0556,
+      "step": 15097
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 1.631828978622328e-05,
+      "loss": 0.0559,
+      "step": 15098
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 1.6312351543942996e-05,
+      "loss": 0.0193,
+      "step": 15099
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 1.630641330166271e-05,
+      "loss": 0.0428,
+      "step": 15100
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 1.6300475059382423e-05,
+      "loss": 0.0398,
+      "step": 15101
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 1.6294536817102137e-05,
+      "loss": 0.0273,
+      "step": 15102
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 1.6288598574821854e-05,
+      "loss": 0.0534,
+      "step": 15103
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 1.628266033254157e-05,
+      "loss": 0.0196,
+      "step": 15104
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 1.6276722090261284e-05,
+      "loss": 0.0487,
+      "step": 15105
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 1.6270783847980998e-05,
+      "loss": 0.0355,
+      "step": 15106
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 1.626484560570071e-05,
+      "loss": 0.033,
+      "step": 15107
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 1.6258907363420428e-05,
+      "loss": 0.0265,
+      "step": 15108
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 1.6252969121140145e-05,
+      "loss": 0.0289,
+      "step": 15109
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 1.624703087885986e-05,
+      "loss": 0.0221,
+      "step": 15110
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 1.6241092636579576e-05,
+      "loss": 0.0421,
+      "step": 15111
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 1.6235154394299286e-05,
+      "loss": 0.0541,
+      "step": 15112
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 1.6229216152019003e-05,
+      "loss": 0.0181,
+      "step": 15113
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 1.622327790973872e-05,
+      "loss": 0.0437,
+      "step": 15114
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 1.6217339667458433e-05,
+      "loss": 0.061,
+      "step": 15115
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 1.621140142517815e-05,
+      "loss": 0.0258,
+      "step": 15116
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 1.620546318289786e-05,
+      "loss": 0.0149,
+      "step": 15117
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 1.6199524940617578e-05,
+      "loss": 0.0568,
+      "step": 15118
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 1.6193586698337295e-05,
+      "loss": 0.0371,
+      "step": 15119
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 1.6187648456057008e-05,
+      "loss": 0.0253,
+      "step": 15120
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 1.6181710213776725e-05,
+      "loss": 0.0546,
+      "step": 15121
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 1.6175771971496435e-05,
+      "loss": 0.0287,
+      "step": 15122
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 1.6169833729216152e-05,
+      "loss": 0.0681,
+      "step": 15123
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 1.616389548693587e-05,
+      "loss": 0.0449,
+      "step": 15124
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 1.6157957244655583e-05,
+      "loss": 0.0415,
+      "step": 15125
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 1.61520190023753e-05,
+      "loss": 0.047,
+      "step": 15126
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 1.614608076009501e-05,
+      "loss": 0.0409,
+      "step": 15127
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 1.6140142517814727e-05,
+      "loss": 0.0349,
+      "step": 15128
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 1.6134204275534444e-05,
+      "loss": 0.0519,
+      "step": 15129
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 1.6128266033254157e-05,
+      "loss": 0.0409,
+      "step": 15130
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 1.6122327790973874e-05,
+      "loss": 0.0524,
+      "step": 15131
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 1.6116389548693585e-05,
+      "loss": 0.0535,
+      "step": 15132
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 1.61104513064133e-05,
+      "loss": 0.0534,
+      "step": 15133
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 1.610451306413302e-05,
+      "loss": 0.0451,
+      "step": 15134
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 1.6098574821852732e-05,
+      "loss": 0.0556,
+      "step": 15135
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 1.609263657957245e-05,
+      "loss": 0.0349,
+      "step": 15136
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 1.608669833729216e-05,
+      "loss": 0.0353,
+      "step": 15137
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 1.6080760095011876e-05,
+      "loss": 0.0552,
+      "step": 15138
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 1.6074821852731593e-05,
+      "loss": 0.0323,
+      "step": 15139
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 1.6068883610451307e-05,
+      "loss": 0.0494,
+      "step": 15140
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 1.6062945368171024e-05,
+      "loss": 0.0281,
+      "step": 15141
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 1.6057007125890737e-05,
+      "loss": 0.0212,
+      "step": 15142
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 1.605106888361045e-05,
+      "loss": 0.0348,
+      "step": 15143
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 1.6045130641330168e-05,
+      "loss": 0.0562,
+      "step": 15144
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 1.603919239904988e-05,
+      "loss": 0.0188,
+      "step": 15145
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 1.60332541567696e-05,
+      "loss": 0.0215,
+      "step": 15146
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 1.6027315914489312e-05,
+      "loss": 0.0338,
+      "step": 15147
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 1.6021377672209026e-05,
+      "loss": 0.0327,
+      "step": 15148
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 1.6015439429928743e-05,
+      "loss": 0.0214,
+      "step": 15149
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 1.6009501187648456e-05,
+      "loss": 0.0326,
+      "step": 15150
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 1.6003562945368173e-05,
+      "loss": 0.0259,
+      "step": 15151
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 1.5997624703087887e-05,
+      "loss": 0.0266,
+      "step": 15152
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 1.59916864608076e-05,
+      "loss": 0.0569,
+      "step": 15153
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 1.5985748218527317e-05,
+      "loss": 0.018,
+      "step": 15154
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 1.597980997624703e-05,
+      "loss": 0.0357,
+      "step": 15155
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 1.5973871733966748e-05,
+      "loss": 0.0275,
+      "step": 15156
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 1.596793349168646e-05,
+      "loss": 0.0319,
+      "step": 15157
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 1.5961995249406175e-05,
+      "loss": 0.0369,
+      "step": 15158
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 1.5956057007125892e-05,
+      "loss": 0.0124,
+      "step": 15159
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 1.5950118764845606e-05,
+      "loss": 0.0198,
+      "step": 15160
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 1.5944180522565323e-05,
+      "loss": 0.0469,
+      "step": 15161
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 1.5938242280285036e-05,
+      "loss": 0.0341,
+      "step": 15162
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 1.593230403800475e-05,
+      "loss": 0.0138,
+      "step": 15163
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 1.593230403800475e-05,
+      "loss": 0.2076,
+      "step": 15164
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 1.5926365795724467e-05,
+      "loss": 0.0449,
+      "step": 15165
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 1.592042755344418e-05,
+      "loss": 0.0386,
+      "step": 15166
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 1.5914489311163897e-05,
+      "loss": 0.0404,
+      "step": 15167
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 1.590855106888361e-05,
+      "loss": 0.0412,
+      "step": 15168
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 1.5902612826603328e-05,
+      "loss": 0.0348,
+      "step": 15169
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 1.589667458432304e-05,
+      "loss": 0.0268,
+      "step": 15170
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 1.5890736342042755e-05,
+      "loss": 0.0351,
+      "step": 15171
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 1.5884798099762472e-05,
+      "loss": 0.0246,
+      "step": 15172
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 1.5878859857482185e-05,
+      "loss": 0.0406,
+      "step": 15173
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 1.5872921615201902e-05,
+      "loss": 0.0213,
+      "step": 15174
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 1.5866983372921616e-05,
+      "loss": 0.0518,
+      "step": 15175
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 1.586104513064133e-05,
+      "loss": 0.0583,
+      "step": 15176
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 1.5855106888361047e-05,
+      "loss": 0.0284,
+      "step": 15177
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 1.584916864608076e-05,
+      "loss": 0.0293,
+      "step": 15178
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 1.5843230403800477e-05,
+      "loss": 0.0355,
+      "step": 15179
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 1.583729216152019e-05,
+      "loss": 0.033,
+      "step": 15180
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 1.5831353919239904e-05,
+      "loss": 0.0177,
+      "step": 15181
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 1.582541567695962e-05,
+      "loss": 0.0376,
+      "step": 15182
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 1.5819477434679335e-05,
+      "loss": 0.0557,
+      "step": 15183
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 1.5813539192399052e-05,
+      "loss": 0.0371,
+      "step": 15184
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 1.5807600950118765e-05,
+      "loss": 0.047,
+      "step": 15185
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 1.580166270783848e-05,
+      "loss": 0.0232,
+      "step": 15186
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 1.5795724465558196e-05,
+      "loss": 0.0614,
+      "step": 15187
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 1.578978622327791e-05,
+      "loss": 0.0312,
+      "step": 15188
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 1.5783847980997626e-05,
+      "loss": 0.0387,
+      "step": 15189
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 1.577790973871734e-05,
+      "loss": 0.038,
+      "step": 15190
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 1.5771971496437054e-05,
+      "loss": 0.051,
+      "step": 15191
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 1.576603325415677e-05,
+      "loss": 0.0365,
+      "step": 15192
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 1.5760095011876484e-05,
+      "loss": 0.0353,
+      "step": 15193
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 1.57541567695962e-05,
+      "loss": 0.0143,
+      "step": 15194
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 1.5748218527315918e-05,
+      "loss": 0.0273,
+      "step": 15195
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 1.574228028503563e-05,
+      "loss": 0.0182,
+      "step": 15196
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 1.5736342042755345e-05,
+      "loss": 0.0459,
+      "step": 15197
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 1.573040380047506e-05,
+      "loss": 0.0425,
+      "step": 15198
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 1.5724465558194776e-05,
+      "loss": 0.0354,
+      "step": 15199
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 1.5718527315914493e-05,
+      "loss": 0.0285,
+      "step": 15200
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 1.5712589073634203e-05,
+      "loss": 0.0157,
+      "step": 15201
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 1.570665083135392e-05,
+      "loss": 0.0276,
+      "step": 15202
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 1.5700712589073634e-05,
+      "loss": 0.044,
+      "step": 15203
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 1.569477434679335e-05,
+      "loss": 0.0255,
+      "step": 15204
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 1.5688836104513067e-05,
+      "loss": 0.0312,
+      "step": 15205
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 1.5682897862232778e-05,
+      "loss": 0.0226,
+      "step": 15206
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 1.5676959619952495e-05,
+      "loss": 0.0317,
+      "step": 15207
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 1.5671021377672208e-05,
+      "loss": 0.0303,
+      "step": 15208
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 1.5665083135391925e-05,
+      "loss": 0.0184,
+      "step": 15209
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 1.5659144893111642e-05,
+      "loss": 0.0295,
+      "step": 15210
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 1.5653206650831352e-05,
+      "loss": 0.0145,
+      "step": 15211
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 1.564726840855107e-05,
+      "loss": 0.05,
+      "step": 15212
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 1.5641330166270783e-05,
+      "loss": 0.0302,
+      "step": 15213
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 1.56353919239905e-05,
+      "loss": 0.0831,
+      "step": 15214
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 1.5629453681710217e-05,
+      "loss": 0.0443,
+      "step": 15215
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 1.5623515439429927e-05,
+      "loss": 0.0602,
+      "step": 15216
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 1.5617577197149644e-05,
+      "loss": 0.0525,
+      "step": 15217
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 1.5611638954869358e-05,
+      "loss": 0.0487,
+      "step": 15218
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 1.5605700712589075e-05,
+      "loss": 0.0314,
+      "step": 15219
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 1.559976247030879e-05,
+      "loss": 0.0444,
+      "step": 15220
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 1.5593824228028502e-05,
+      "loss": 0.0287,
+      "step": 15221
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 1.558788598574822e-05,
+      "loss": 0.033,
+      "step": 15222
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 1.5581947743467936e-05,
+      "loss": 0.056,
+      "step": 15223
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 1.557600950118765e-05,
+      "loss": 0.0493,
+      "step": 15224
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 1.5570071258907366e-05,
+      "loss": 0.0435,
+      "step": 15225
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 1.556413301662708e-05,
+      "loss": 0.0378,
+      "step": 15226
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 1.5558194774346793e-05,
+      "loss": 0.0191,
+      "step": 15227
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 1.555225653206651e-05,
+      "loss": 0.0137,
+      "step": 15228
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 1.5546318289786224e-05,
+      "loss": 0.0224,
+      "step": 15229
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 1.554038004750594e-05,
+      "loss": 0.0389,
+      "step": 15230
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 1.5534441805225654e-05,
+      "loss": 0.0311,
+      "step": 15231
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 1.5528503562945368e-05,
+      "loss": 0.0283,
+      "step": 15232
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 1.5522565320665085e-05,
+      "loss": 0.03,
+      "step": 15233
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 1.55166270783848e-05,
+      "loss": 0.0244,
+      "step": 15234
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 1.5510688836104516e-05,
+      "loss": 0.0296,
+      "step": 15235
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 1.550475059382423e-05,
+      "loss": 0.0295,
+      "step": 15236
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 1.5498812351543943e-05,
+      "loss": 0.0433,
+      "step": 15237
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 1.549287410926366e-05,
+      "loss": 0.0387,
+      "step": 15238
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 1.5486935866983373e-05,
+      "loss": 0.05,
+      "step": 15239
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 1.548099762470309e-05,
+      "loss": 0.0397,
+      "step": 15240
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 1.5475059382422804e-05,
+      "loss": 0.054,
+      "step": 15241
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 1.5469121140142517e-05,
+      "loss": 0.0405,
+      "step": 15242
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 1.5463182897862234e-05,
+      "loss": 0.0157,
+      "step": 15243
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 1.5457244655581948e-05,
+      "loss": 0.026,
+      "step": 15244
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 1.5451306413301665e-05,
+      "loss": 0.0322,
+      "step": 15245
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 1.544536817102138e-05,
+      "loss": 0.0373,
+      "step": 15246
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 1.5439429928741092e-05,
+      "loss": 0.0227,
+      "step": 15247
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 1.543349168646081e-05,
+      "loss": 0.0405,
+      "step": 15248
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 1.5427553444180523e-05,
+      "loss": 0.0469,
+      "step": 15249
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 1.542161520190024e-05,
+      "loss": 0.0153,
+      "step": 15250
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 1.5415676959619953e-05,
+      "loss": 0.0243,
+      "step": 15251
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 1.540973871733967e-05,
+      "loss": 0.056,
+      "step": 15252
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 1.5403800475059384e-05,
+      "loss": 0.0354,
+      "step": 15253
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 1.5397862232779097e-05,
+      "loss": 0.0185,
+      "step": 15254
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 1.5391923990498814e-05,
+      "loss": 0.0272,
+      "step": 15255
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 1.5385985748218528e-05,
+      "loss": 0.0646,
+      "step": 15256
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 1.5380047505938245e-05,
+      "loss": 0.0289,
+      "step": 15257
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 1.537410926365796e-05,
+      "loss": 0.0176,
+      "step": 15258
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 1.5368171021377672e-05,
+      "loss": 0.0169,
+      "step": 15259
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 1.536223277909739e-05,
+      "loss": 0.0258,
+      "step": 15260
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 1.5356294536817103e-05,
+      "loss": 0.0233,
+      "step": 15261
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 1.535035629453682e-05,
+      "loss": 0.0546,
+      "step": 15262
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 1.5344418052256533e-05,
+      "loss": 0.0064,
+      "step": 15263
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 1.5338479809976247e-05,
+      "loss": 0.0372,
+      "step": 15264
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 1.5332541567695964e-05,
+      "loss": 0.0509,
+      "step": 15265
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 1.5326603325415677e-05,
+      "loss": 0.0519,
+      "step": 15266
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 1.5320665083135394e-05,
+      "loss": 0.0301,
+      "step": 15267
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 1.5314726840855108e-05,
+      "loss": 0.0365,
+      "step": 15268
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 1.530878859857482e-05,
+      "loss": 0.0541,
+      "step": 15269
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 1.530285035629454e-05,
+      "loss": 0.0435,
+      "step": 15270
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 1.5296912114014252e-05,
+      "loss": 0.02,
+      "step": 15271
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 1.529097387173397e-05,
+      "loss": 0.0422,
+      "step": 15272
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 1.5285035629453682e-05,
+      "loss": 0.021,
+      "step": 15273
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 1.5279097387173396e-05,
+      "loss": 0.041,
+      "step": 15274
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 1.5273159144893113e-05,
+      "loss": 0.0399,
+      "step": 15275
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 1.5267220902612827e-05,
+      "loss": 0.0372,
+      "step": 15276
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 1.5261282660332544e-05,
+      "loss": 0.0635,
+      "step": 15277
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 1.5255344418052255e-05,
+      "loss": 0.0512,
+      "step": 15278
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 1.524940617577197e-05,
+      "loss": 0.0375,
+      "step": 15279
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 1.5243467933491688e-05,
+      "loss": 0.028,
+      "step": 15280
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 1.5237529691211403e-05,
+      "loss": 0.0436,
+      "step": 15281
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 1.5231591448931118e-05,
+      "loss": 0.0409,
+      "step": 15282
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 1.5225653206650833e-05,
+      "loss": 0.062,
+      "step": 15283
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 1.5219714964370545e-05,
+      "loss": 0.0406,
+      "step": 15284
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 1.5213776722090262e-05,
+      "loss": 0.0388,
+      "step": 15285
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 1.5207838479809978e-05,
+      "loss": 0.033,
+      "step": 15286
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 1.5201900237529693e-05,
+      "loss": 0.0297,
+      "step": 15287
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 1.5195961995249408e-05,
+      "loss": 0.0439,
+      "step": 15288
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 1.519002375296912e-05,
+      "loss": 0.0164,
+      "step": 15289
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 1.5184085510688837e-05,
+      "loss": 0.0277,
+      "step": 15290
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 1.5178147268408552e-05,
+      "loss": 0.0325,
+      "step": 15291
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 1.5172209026128268e-05,
+      "loss": 0.0157,
+      "step": 15292
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 1.5166270783847983e-05,
+      "loss": 0.03,
+      "step": 15293
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 1.5160332541567695e-05,
+      "loss": 0.0276,
+      "step": 15294
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 1.5154394299287412e-05,
+      "loss": 0.0253,
+      "step": 15295
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 1.5148456057007127e-05,
+      "loss": 0.0427,
+      "step": 15296
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 1.5142517814726842e-05,
+      "loss": 0.0326,
+      "step": 15297
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 1.5136579572446558e-05,
+      "loss": 0.0338,
+      "step": 15298
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 1.5130641330166271e-05,
+      "loss": 0.0513,
+      "step": 15299
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 1.5124703087885986e-05,
+      "loss": 0.0288,
+      "step": 15300
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 1.5118764845605702e-05,
+      "loss": 0.0422,
+      "step": 15301
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 1.5112826603325417e-05,
+      "loss": 0.0182,
+      "step": 15302
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 1.5106888361045132e-05,
+      "loss": 0.0395,
+      "step": 15303
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 1.5100950118764846e-05,
+      "loss": 0.0258,
+      "step": 15304
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 1.5095011876484561e-05,
+      "loss": 0.0182,
+      "step": 15305
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 1.5089073634204276e-05,
+      "loss": 0.015,
+      "step": 15306
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 1.5083135391923992e-05,
+      "loss": 0.0299,
+      "step": 15307
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 1.5077197149643707e-05,
+      "loss": 0.0207,
+      "step": 15308
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 1.5071258907363422e-05,
+      "loss": 0.0184,
+      "step": 15309
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 1.5065320665083136e-05,
+      "loss": 0.0116,
+      "step": 15310
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 1.5059382422802851e-05,
+      "loss": 0.0194,
+      "step": 15311
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 1.5053444180522566e-05,
+      "loss": 0.0408,
+      "step": 15312
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 1.5047505938242282e-05,
+      "loss": 0.0728,
+      "step": 15313
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 1.5041567695961997e-05,
+      "loss": 0.0524,
+      "step": 15314
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 1.503562945368171e-05,
+      "loss": 0.0369,
+      "step": 15315
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 1.5029691211401426e-05,
+      "loss": 0.025,
+      "step": 15316
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 1.5023752969121141e-05,
+      "loss": 0.04,
+      "step": 15317
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 1.5017814726840856e-05,
+      "loss": 0.0198,
+      "step": 15318
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 1.5011876484560572e-05,
+      "loss": 0.0522,
+      "step": 15319
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 1.5005938242280285e-05,
+      "loss": 0.0319,
+      "step": 15320
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 1.5e-05,
+      "loss": 0.0479,
+      "step": 15321
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 1.4994061757719716e-05,
+      "loss": 0.0287,
+      "step": 15322
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 1.4988123515439431e-05,
+      "loss": 0.0375,
+      "step": 15323
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 1.4982185273159146e-05,
+      "loss": 0.0407,
+      "step": 15324
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 1.497624703087886e-05,
+      "loss": 0.027,
+      "step": 15325
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 1.4970308788598575e-05,
+      "loss": 0.0612,
+      "step": 15326
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 1.496437054631829e-05,
+      "loss": 0.0335,
+      "step": 15327
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 1.4958432304038006e-05,
+      "loss": 0.0409,
+      "step": 15328
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 1.4952494061757721e-05,
+      "loss": 0.0266,
+      "step": 15329
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 1.4946555819477434e-05,
+      "loss": 0.0287,
+      "step": 15330
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 1.494061757719715e-05,
+      "loss": 0.0407,
+      "step": 15331
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 1.4934679334916865e-05,
+      "loss": 0.0224,
+      "step": 15332
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 1.492874109263658e-05,
+      "loss": 0.018,
+      "step": 15333
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 1.4922802850356296e-05,
+      "loss": 0.0312,
+      "step": 15334
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 1.4916864608076009e-05,
+      "loss": 0.0243,
+      "step": 15335
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 1.4910926365795724e-05,
+      "loss": 0.0149,
+      "step": 15336
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 1.490498812351544e-05,
+      "loss": 0.0211,
+      "step": 15337
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 1.4899049881235155e-05,
+      "loss": 0.0244,
+      "step": 15338
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 1.489311163895487e-05,
+      "loss": 0.0364,
+      "step": 15339
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 1.4887173396674586e-05,
+      "loss": 0.0363,
+      "step": 15340
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 1.4881235154394299e-05,
+      "loss": 0.0356,
+      "step": 15341
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 1.4875296912114014e-05,
+      "loss": 0.0432,
+      "step": 15342
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 1.486935866983373e-05,
+      "loss": 0.0206,
+      "step": 15343
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 1.4863420427553445e-05,
+      "loss": 0.0198,
+      "step": 15344
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 1.4857482185273162e-05,
+      "loss": 0.0236,
+      "step": 15345
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 1.4851543942992874e-05,
+      "loss": 0.0347,
+      "step": 15346
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 1.4845605700712589e-05,
+      "loss": 0.0221,
+      "step": 15347
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 1.4839667458432304e-05,
+      "loss": 0.0326,
+      "step": 15348
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 1.483372921615202e-05,
+      "loss": 0.0248,
+      "step": 15349
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 1.4827790973871737e-05,
+      "loss": 0.0209,
+      "step": 15350
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 1.4821852731591448e-05,
+      "loss": 0.0248,
+      "step": 15351
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 1.4815914489311164e-05,
+      "loss": 0.0218,
+      "step": 15352
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 1.4809976247030879e-05,
+      "loss": 0.0298,
+      "step": 15353
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 1.4804038004750594e-05,
+      "loss": 0.0259,
+      "step": 15354
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 1.4798099762470311e-05,
+      "loss": 0.0203,
+      "step": 15355
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 1.4792161520190023e-05,
+      "loss": 0.0296,
+      "step": 15356
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 1.4786223277909738e-05,
+      "loss": 0.0203,
+      "step": 15357
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 1.4780285035629454e-05,
+      "loss": 0.0341,
+      "step": 15358
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 1.4774346793349169e-05,
+      "loss": 0.0137,
+      "step": 15359
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 1.4768408551068886e-05,
+      "loss": 0.0582,
+      "step": 15360
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 1.4762470308788598e-05,
+      "loss": 0.0326,
+      "step": 15361
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 1.4756532066508313e-05,
+      "loss": 0.0393,
+      "step": 15362
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 1.4750593824228028e-05,
+      "loss": 0.0125,
+      "step": 15363
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 1.4744655581947744e-05,
+      "loss": 0.0224,
+      "step": 15364
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 1.473871733966746e-05,
+      "loss": 0.0469,
+      "step": 15365
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 1.4732779097387176e-05,
+      "loss": 0.072,
+      "step": 15366
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 1.4726840855106888e-05,
+      "loss": 0.0426,
+      "step": 15367
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 1.4720902612826603e-05,
+      "loss": 0.0353,
+      "step": 15368
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 1.4714964370546318e-05,
+      "loss": 0.0427,
+      "step": 15369
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 1.4709026128266035e-05,
+      "loss": 0.024,
+      "step": 15370
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 1.470308788598575e-05,
+      "loss": 0.0361,
+      "step": 15371
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 1.4697149643705462e-05,
+      "loss": 0.1039,
+      "step": 15372
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 1.4691211401425178e-05,
+      "loss": 0.0244,
+      "step": 15373
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 1.4685273159144895e-05,
+      "loss": 0.0428,
+      "step": 15374
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 1.467933491686461e-05,
+      "loss": 0.0268,
+      "step": 15375
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 1.4673396674584325e-05,
+      "loss": 0.0299,
+      "step": 15376
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 1.4667458432304037e-05,
+      "loss": 0.0367,
+      "step": 15377
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 1.4661520190023752e-05,
+      "loss": 0.0299,
+      "step": 15378
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 1.465558194774347e-05,
+      "loss": 0.0304,
+      "step": 15379
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 1.4649643705463185e-05,
+      "loss": 0.0305,
+      "step": 15380
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 1.46437054631829e-05,
+      "loss": 0.0623,
+      "step": 15381
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 1.4637767220902612e-05,
+      "loss": 0.0433,
+      "step": 15382
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 1.4631828978622327e-05,
+      "loss": 0.0332,
+      "step": 15383
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 1.4625890736342044e-05,
+      "loss": 0.0287,
+      "step": 15384
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 1.461995249406176e-05,
+      "loss": 0.0179,
+      "step": 15385
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 1.4614014251781475e-05,
+      "loss": 0.0243,
+      "step": 15386
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 1.4608076009501186e-05,
+      "loss": 0.0249,
+      "step": 15387
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 1.4602137767220902e-05,
+      "loss": 0.0324,
+      "step": 15388
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 1.4596199524940619e-05,
+      "loss": 0.0171,
+      "step": 15389
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 1.4590261282660334e-05,
+      "loss": 0.0309,
+      "step": 15390
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 1.458432304038005e-05,
+      "loss": 0.0369,
+      "step": 15391
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 1.4578384798099765e-05,
+      "loss": 0.0258,
+      "step": 15392
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 1.4572446555819476e-05,
+      "loss": 0.0415,
+      "step": 15393
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 1.4566508313539193e-05,
+      "loss": 0.0441,
+      "step": 15394
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 1.4560570071258909e-05,
+      "loss": 0.0354,
+      "step": 15395
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 1.4554631828978624e-05,
+      "loss": 0.0358,
+      "step": 15396
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 1.454869358669834e-05,
+      "loss": 0.0162,
+      "step": 15397
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 1.4542755344418053e-05,
+      "loss": 0.0209,
+      "step": 15398
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 1.4536817102137768e-05,
+      "loss": 0.035,
+      "step": 15399
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 1.4530878859857483e-05,
+      "loss": 0.0235,
+      "step": 15400
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 1.4524940617577199e-05,
+      "loss": 0.034,
+      "step": 15401
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 1.4519002375296914e-05,
+      "loss": 0.0201,
+      "step": 15402
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 1.4513064133016627e-05,
+      "loss": 0.0227,
+      "step": 15403
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 1.4507125890736343e-05,
+      "loss": 0.0314,
+      "step": 15404
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 1.4501187648456058e-05,
+      "loss": 0.0304,
+      "step": 15405
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 1.4495249406175773e-05,
+      "loss": 0.0482,
+      "step": 15406
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 1.4489311163895489e-05,
+      "loss": 0.0436,
+      "step": 15407
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 1.4483372921615202e-05,
+      "loss": 0.0241,
+      "step": 15408
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 1.4477434679334917e-05,
+      "loss": 0.0055,
+      "step": 15409
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 1.4471496437054633e-05,
+      "loss": 0.0417,
+      "step": 15410
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 1.4465558194774348e-05,
+      "loss": 0.0135,
+      "step": 15411
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 1.4459619952494063e-05,
+      "loss": 0.0563,
+      "step": 15412
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 1.4453681710213777e-05,
+      "loss": 0.0863,
+      "step": 15413
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 1.4447743467933492e-05,
+      "loss": 0.0096,
+      "step": 15414
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 1.4441805225653207e-05,
+      "loss": 0.0515,
+      "step": 15415
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 1.4435866983372923e-05,
+      "loss": 0.0851,
+      "step": 15416
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 1.4429928741092638e-05,
+      "loss": 0.0322,
+      "step": 15417
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 1.4423990498812352e-05,
+      "loss": 0.0332,
+      "step": 15418
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 1.4418052256532067e-05,
+      "loss": 0.0399,
+      "step": 15419
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 1.4412114014251782e-05,
+      "loss": 0.0214,
+      "step": 15420
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 1.4406175771971497e-05,
+      "loss": 0.0556,
+      "step": 15421
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 1.4400237529691213e-05,
+      "loss": 0.0355,
+      "step": 15422
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 1.4394299287410928e-05,
+      "loss": 0.0276,
+      "step": 15423
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 1.4388361045130641e-05,
+      "loss": 0.0333,
+      "step": 15424
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 1.4382422802850357e-05,
+      "loss": 0.049,
+      "step": 15425
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 1.4376484560570072e-05,
+      "loss": 0.0599,
+      "step": 15426
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 1.4370546318289787e-05,
+      "loss": 0.029,
+      "step": 15427
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 1.4364608076009503e-05,
+      "loss": 0.0475,
+      "step": 15428
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 1.4358669833729216e-05,
+      "loss": 0.0624,
+      "step": 15429
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 1.4352731591448931e-05,
+      "loss": 0.0204,
+      "step": 15430
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 1.4346793349168647e-05,
+      "loss": 0.0546,
+      "step": 15431
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 1.4340855106888362e-05,
+      "loss": 0.026,
+      "step": 15432
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 1.4334916864608077e-05,
+      "loss": 0.0283,
+      "step": 15433
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 1.4328978622327791e-05,
+      "loss": 0.0409,
+      "step": 15434
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 1.4323040380047506e-05,
+      "loss": 0.0487,
+      "step": 15435
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 1.4317102137767221e-05,
+      "loss": 0.0309,
+      "step": 15436
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 1.4311163895486937e-05,
+      "loss": 0.0182,
+      "step": 15437
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 1.4305225653206652e-05,
+      "loss": 0.0312,
+      "step": 15438
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 1.4299287410926366e-05,
+      "loss": 0.026,
+      "step": 15439
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 1.429334916864608e-05,
+      "loss": 0.0262,
+      "step": 15440
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 1.4287410926365796e-05,
+      "loss": 0.0422,
+      "step": 15441
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 1.4281472684085511e-05,
+      "loss": 0.0417,
+      "step": 15442
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 1.4275534441805227e-05,
+      "loss": 0.0444,
+      "step": 15443
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 1.426959619952494e-05,
+      "loss": 0.0214,
+      "step": 15444
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 1.4263657957244655e-05,
+      "loss": 0.0509,
+      "step": 15445
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 1.425771971496437e-05,
+      "loss": 0.037,
+      "step": 15446
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 1.4251781472684086e-05,
+      "loss": 0.0247,
+      "step": 15447
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 1.4245843230403801e-05,
+      "loss": 0.0243,
+      "step": 15448
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 1.4239904988123518e-05,
+      "loss": 0.0238,
+      "step": 15449
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 1.423396674584323e-05,
+      "loss": 0.0295,
+      "step": 15450
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 1.4228028503562945e-05,
+      "loss": 0.0308,
+      "step": 15451
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 1.422209026128266e-05,
+      "loss": 0.0285,
+      "step": 15452
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 1.4216152019002376e-05,
+      "loss": 0.0216,
+      "step": 15453
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 1.4210213776722093e-05,
+      "loss": 0.0382,
+      "step": 15454
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 1.4204275534441805e-05,
+      "loss": 0.0549,
+      "step": 15455
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 1.419833729216152e-05,
+      "loss": 0.0225,
+      "step": 15456
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 1.4192399049881235e-05,
+      "loss": 0.0473,
+      "step": 15457
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 1.418646080760095e-05,
+      "loss": 0.0352,
+      "step": 15458
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 1.4180522565320668e-05,
+      "loss": 0.0403,
+      "step": 15459
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 1.417458432304038e-05,
+      "loss": 0.0283,
+      "step": 15460
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 1.4168646080760095e-05,
+      "loss": 0.1158,
+      "step": 15461
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 1.416270783847981e-05,
+      "loss": 0.0339,
+      "step": 15462
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 1.4156769596199525e-05,
+      "loss": 0.0142,
+      "step": 15463
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 1.4150831353919242e-05,
+      "loss": 0.0343,
+      "step": 15464
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 1.4144893111638954e-05,
+      "loss": 0.0467,
+      "step": 15465
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 1.413895486935867e-05,
+      "loss": 0.032,
+      "step": 15466
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 1.4133016627078385e-05,
+      "loss": 0.0375,
+      "step": 15467
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 1.41270783847981e-05,
+      "loss": 0.053,
+      "step": 15468
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 1.4121140142517817e-05,
+      "loss": 0.0283,
+      "step": 15469
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 1.4115201900237529e-05,
+      "loss": 0.0463,
+      "step": 15470
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 1.4109263657957244e-05,
+      "loss": 0.0404,
+      "step": 15471
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 1.410332541567696e-05,
+      "loss": 0.0197,
+      "step": 15472
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 1.4097387173396676e-05,
+      "loss": 0.056,
+      "step": 15473
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 1.4091448931116392e-05,
+      "loss": 0.0268,
+      "step": 15474
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 1.4085510688836104e-05,
+      "loss": 0.0387,
+      "step": 15475
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 1.4079572446555819e-05,
+      "loss": 0.0337,
+      "step": 15476
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 1.4073634204275534e-05,
+      "loss": 0.031,
+      "step": 15477
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 1.4067695961995251e-05,
+      "loss": 0.0416,
+      "step": 15478
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 1.4061757719714966e-05,
+      "loss": 0.026,
+      "step": 15479
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 1.4055819477434682e-05,
+      "loss": 0.0258,
+      "step": 15480
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 1.4049881235154394e-05,
+      "loss": 0.0359,
+      "step": 15481
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 1.4043942992874109e-05,
+      "loss": 0.0379,
+      "step": 15482
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 1.4038004750593826e-05,
+      "loss": 0.0267,
+      "step": 15483
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 1.4032066508313541e-05,
+      "loss": 0.0325,
+      "step": 15484
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 1.4026128266033256e-05,
+      "loss": 0.0445,
+      "step": 15485
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 1.4020190023752968e-05,
+      "loss": 0.0422,
+      "step": 15486
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 1.4014251781472683e-05,
+      "loss": 0.0294,
+      "step": 15487
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 1.40083135391924e-05,
+      "loss": 0.0318,
+      "step": 15488
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 1.4002375296912116e-05,
+      "loss": 0.0338,
+      "step": 15489
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 1.3996437054631831e-05,
+      "loss": 0.0233,
+      "step": 15490
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 1.3990498812351543e-05,
+      "loss": 0.0288,
+      "step": 15491
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 1.3984560570071258e-05,
+      "loss": 0.0168,
+      "step": 15492
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 1.3978622327790975e-05,
+      "loss": 0.031,
+      "step": 15493
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 1.397268408551069e-05,
+      "loss": 0.0331,
+      "step": 15494
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 1.3966745843230406e-05,
+      "loss": 0.04,
+      "step": 15495
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 1.3960807600950118e-05,
+      "loss": 0.0313,
+      "step": 15496
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 1.3954869358669835e-05,
+      "loss": 0.0216,
+      "step": 15497
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 1.394893111638955e-05,
+      "loss": 0.0176,
+      "step": 15498
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 1.3942992874109265e-05,
+      "loss": 0.0588,
+      "step": 15499
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 1.393705463182898e-05,
+      "loss": 0.0263,
+      "step": 15500
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 1.3931116389548692e-05,
+      "loss": 0.0304,
+      "step": 15501
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 1.392517814726841e-05,
+      "loss": 0.0541,
+      "step": 15502
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 1.3919239904988124e-05,
+      "loss": 0.0725,
+      "step": 15503
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 1.391330166270784e-05,
+      "loss": 0.0298,
+      "step": 15504
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 1.3907363420427555e-05,
+      "loss": 0.0525,
+      "step": 15505
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 1.390142517814727e-05,
+      "loss": 0.0287,
+      "step": 15506
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 1.3895486935866984e-05,
+      "loss": 0.0307,
+      "step": 15507
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 1.38895486935867e-05,
+      "loss": 0.0278,
+      "step": 15508
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 1.3883610451306414e-05,
+      "loss": 0.0279,
+      "step": 15509
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 1.387767220902613e-05,
+      "loss": 0.0195,
+      "step": 15510
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 1.3871733966745845e-05,
+      "loss": 0.0587,
+      "step": 15511
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 1.3865795724465559e-05,
+      "loss": 0.0156,
+      "step": 15512
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 1.3859857482185274e-05,
+      "loss": 0.0144,
+      "step": 15513
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 1.3853919239904989e-05,
+      "loss": 0.0184,
+      "step": 15514
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 1.3847980997624704e-05,
+      "loss": 0.0478,
+      "step": 15515
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 1.384204275534442e-05,
+      "loss": 0.0623,
+      "step": 15516
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 1.3836104513064133e-05,
+      "loss": 0.0414,
+      "step": 15517
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 1.3830166270783849e-05,
+      "loss": 0.0319,
+      "step": 15518
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 1.3824228028503564e-05,
+      "loss": 0.0226,
+      "step": 15519
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 1.3818289786223279e-05,
+      "loss": 0.0334,
+      "step": 15520
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 1.3812351543942994e-05,
+      "loss": 0.054,
+      "step": 15521
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 1.3806413301662708e-05,
+      "loss": 0.039,
+      "step": 15522
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 1.3800475059382423e-05,
+      "loss": 0.0357,
+      "step": 15523
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 1.3794536817102138e-05,
+      "loss": 0.0412,
+      "step": 15524
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 1.3788598574821854e-05,
+      "loss": 0.0319,
+      "step": 15525
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 1.3782660332541569e-05,
+      "loss": 0.0308,
+      "step": 15526
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 1.3776722090261283e-05,
+      "loss": 0.0475,
+      "step": 15527
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 1.3770783847980998e-05,
+      "loss": 0.0309,
+      "step": 15528
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 1.3764845605700713e-05,
+      "loss": 0.0276,
+      "step": 15529
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 1.3758907363420428e-05,
+      "loss": 0.0293,
+      "step": 15530
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 1.3752969121140144e-05,
+      "loss": 0.0486,
+      "step": 15531
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 1.3747030878859859e-05,
+      "loss": 0.0401,
+      "step": 15532
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 1.3741092636579573e-05,
+      "loss": 0.039,
+      "step": 15533
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 1.3735154394299288e-05,
+      "loss": 0.0366,
+      "step": 15534
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 1.3729216152019003e-05,
+      "loss": 0.0411,
+      "step": 15535
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 1.3723277909738718e-05,
+      "loss": 0.0308,
+      "step": 15536
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 1.3717339667458434e-05,
+      "loss": 0.0379,
+      "step": 15537
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 1.3711401425178147e-05,
+      "loss": 0.0379,
+      "step": 15538
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 1.3705463182897863e-05,
+      "loss": 0.0244,
+      "step": 15539
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 1.3699524940617578e-05,
+      "loss": 0.0424,
+      "step": 15540
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 1.3693586698337293e-05,
+      "loss": 0.0255,
+      "step": 15541
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 1.3687648456057008e-05,
+      "loss": 0.0521,
+      "step": 15542
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 1.3681710213776722e-05,
+      "loss": 0.0308,
+      "step": 15543
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 1.3675771971496437e-05,
+      "loss": 0.0603,
+      "step": 15544
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 1.3669833729216152e-05,
+      "loss": 0.0169,
+      "step": 15545
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 1.3663895486935868e-05,
+      "loss": 0.0258,
+      "step": 15546
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 1.3657957244655583e-05,
+      "loss": 0.0455,
+      "step": 15547
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 1.3652019002375297e-05,
+      "loss": 0.03,
+      "step": 15548
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 1.3646080760095012e-05,
+      "loss": 0.0461,
+      "step": 15549
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 1.3640142517814727e-05,
+      "loss": 0.0299,
+      "step": 15550
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 1.3634204275534442e-05,
+      "loss": 0.0135,
+      "step": 15551
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 1.3628266033254158e-05,
+      "loss": 0.019,
+      "step": 15552
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 1.3622327790973871e-05,
+      "loss": 0.0392,
+      "step": 15553
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 1.3616389548693587e-05,
+      "loss": 0.0391,
+      "step": 15554
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 1.3610451306413302e-05,
+      "loss": 0.0594,
+      "step": 15555
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 1.3604513064133017e-05,
+      "loss": 0.0486,
+      "step": 15556
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 1.3598574821852732e-05,
+      "loss": 0.0477,
+      "step": 15557
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 1.3592636579572446e-05,
+      "loss": 0.0098,
+      "step": 15558
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 1.3586698337292161e-05,
+      "loss": 0.0241,
+      "step": 15559
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 1.3580760095011877e-05,
+      "loss": 0.0256,
+      "step": 15560
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 1.3574821852731592e-05,
+      "loss": 0.0038,
+      "step": 15561
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 1.3568883610451307e-05,
+      "loss": 0.0111,
+      "step": 15562
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 1.3562945368171024e-05,
+      "loss": 0.0243,
+      "step": 15563
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 1.3557007125890736e-05,
+      "loss": 0.0681,
+      "step": 15564
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 1.3551068883610451e-05,
+      "loss": 0.0562,
+      "step": 15565
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 1.3545130641330166e-05,
+      "loss": 0.0568,
+      "step": 15566
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 1.3539192399049882e-05,
+      "loss": 0.0562,
+      "step": 15567
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 1.3533254156769599e-05,
+      "loss": 0.0247,
+      "step": 15568
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 1.352731591448931e-05,
+      "loss": 0.0415,
+      "step": 15569
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 1.3521377672209026e-05,
+      "loss": 0.0277,
+      "step": 15570
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 1.3515439429928741e-05,
+      "loss": 0.0456,
+      "step": 15571
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 1.3509501187648458e-05,
+      "loss": 0.0384,
+      "step": 15572
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 1.3503562945368173e-05,
+      "loss": 0.0537,
+      "step": 15573
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 1.3497624703087885e-05,
+      "loss": 0.0211,
+      "step": 15574
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 1.34916864608076e-05,
+      "loss": 0.0217,
+      "step": 15575
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 1.3485748218527316e-05,
+      "loss": 0.029,
+      "step": 15576
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 1.3479809976247033e-05,
+      "loss": 0.0385,
+      "step": 15577
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 1.3473871733966748e-05,
+      "loss": 0.0366,
+      "step": 15578
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 1.346793349168646e-05,
+      "loss": 0.0264,
+      "step": 15579
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 1.3461995249406175e-05,
+      "loss": 0.0385,
+      "step": 15580
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 1.345605700712589e-05,
+      "loss": 0.0494,
+      "step": 15581
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 1.3450118764845607e-05,
+      "loss": 0.0361,
+      "step": 15582
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 1.3444180522565323e-05,
+      "loss": 0.0232,
+      "step": 15583
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 1.3438242280285035e-05,
+      "loss": 0.0474,
+      "step": 15584
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 1.343230403800475e-05,
+      "loss": 0.0485,
+      "step": 15585
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 1.3426365795724465e-05,
+      "loss": 0.0439,
+      "step": 15586
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 1.3420427553444182e-05,
+      "loss": 0.0233,
+      "step": 15587
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 1.3414489311163897e-05,
+      "loss": 0.0359,
+      "step": 15588
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 1.3408551068883613e-05,
+      "loss": 0.0244,
+      "step": 15589
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 1.3402612826603325e-05,
+      "loss": 0.0218,
+      "step": 15590
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 1.339667458432304e-05,
+      "loss": 0.033,
+      "step": 15591
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 1.3390736342042757e-05,
+      "loss": 0.0148,
+      "step": 15592
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 1.3384798099762472e-05,
+      "loss": 0.0373,
+      "step": 15593
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 1.3378859857482187e-05,
+      "loss": 0.0187,
+      "step": 15594
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 1.33729216152019e-05,
+      "loss": 0.0325,
+      "step": 15595
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 1.3366983372921616e-05,
+      "loss": 0.0367,
+      "step": 15596
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 1.3361045130641332e-05,
+      "loss": 0.0394,
+      "step": 15597
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 1.3355106888361047e-05,
+      "loss": 0.0077,
+      "step": 15598
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 1.3349168646080762e-05,
+      "loss": 0.0293,
+      "step": 15599
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 1.3343230403800474e-05,
+      "loss": 0.0264,
+      "step": 15600
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 1.3337292161520191e-05,
+      "loss": 0.0311,
+      "step": 15601
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 1.3331353919239906e-05,
+      "loss": 0.0257,
+      "step": 15602
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 1.3325415676959621e-05,
+      "loss": 0.0284,
+      "step": 15603
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 1.3319477434679337e-05,
+      "loss": 0.0497,
+      "step": 15604
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 1.3313539192399049e-05,
+      "loss": 0.0171,
+      "step": 15605
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 1.3307600950118766e-05,
+      "loss": 0.0254,
+      "step": 15606
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 1.3301662707838481e-05,
+      "loss": 0.0267,
+      "step": 15607
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 1.3295724465558196e-05,
+      "loss": 0.0333,
+      "step": 15608
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 1.3289786223277911e-05,
+      "loss": 0.043,
+      "step": 15609
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 1.3283847980997623e-05,
+      "loss": 0.0155,
+      "step": 15610
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 1.327790973871734e-05,
+      "loss": 0.0384,
+      "step": 15611
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 1.3271971496437056e-05,
+      "loss": 0.0045,
+      "step": 15612
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 1.326603325415677e-05,
+      "loss": 0.0584,
+      "step": 15613
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 1.3260095011876486e-05,
+      "loss": 0.0187,
+      "step": 15614
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 1.3254156769596198e-05,
+      "loss": 0.0463,
+      "step": 15615
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 1.3248218527315915e-05,
+      "loss": 0.0566,
+      "step": 15616
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 1.324228028503563e-05,
+      "loss": 0.0588,
+      "step": 15617
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 1.3236342042755345e-05,
+      "loss": 0.0261,
+      "step": 15618
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 1.323040380047506e-05,
+      "loss": 0.0372,
+      "step": 15619
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 1.3224465558194776e-05,
+      "loss": 0.0409,
+      "step": 15620
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 1.321852731591449e-05,
+      "loss": 0.0235,
+      "step": 15621
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 1.3212589073634205e-05,
+      "loss": 0.0503,
+      "step": 15622
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 1.320665083135392e-05,
+      "loss": 0.04,
+      "step": 15623
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 1.3200712589073635e-05,
+      "loss": 0.0305,
+      "step": 15624
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 1.319477434679335e-05,
+      "loss": 0.0353,
+      "step": 15625
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 1.3188836104513064e-05,
+      "loss": 0.0329,
+      "step": 15626
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 1.318289786223278e-05,
+      "loss": 0.0308,
+      "step": 15627
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 1.3176959619952495e-05,
+      "loss": 0.0386,
+      "step": 15628
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 1.317102137767221e-05,
+      "loss": 0.0429,
+      "step": 15629
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 1.3165083135391925e-05,
+      "loss": 0.0707,
+      "step": 15630
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 1.3159144893111639e-05,
+      "loss": 0.0457,
+      "step": 15631
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 1.3153206650831354e-05,
+      "loss": 0.0437,
+      "step": 15632
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 1.314726840855107e-05,
+      "loss": 0.0448,
+      "step": 15633
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 1.3141330166270785e-05,
+      "loss": 0.0202,
+      "step": 15634
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 1.31353919239905e-05,
+      "loss": 0.0395,
+      "step": 15635
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 1.3129453681710214e-05,
+      "loss": 0.0406,
+      "step": 15636
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 1.3123515439429929e-05,
+      "loss": 0.0287,
+      "step": 15637
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 1.3117577197149644e-05,
+      "loss": 0.0527,
+      "step": 15638
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 1.311163895486936e-05,
+      "loss": 0.0351,
+      "step": 15639
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 1.3105700712589075e-05,
+      "loss": 0.0462,
+      "step": 15640
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 1.3099762470308788e-05,
+      "loss": 0.0216,
+      "step": 15641
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 1.3093824228028504e-05,
+      "loss": 0.0276,
+      "step": 15642
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 1.3087885985748219e-05,
+      "loss": 0.0254,
+      "step": 15643
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 1.3081947743467934e-05,
+      "loss": 0.024,
+      "step": 15644
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 1.307600950118765e-05,
+      "loss": 0.0213,
+      "step": 15645
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 1.3070071258907365e-05,
+      "loss": 0.0281,
+      "step": 15646
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 1.3064133016627078e-05,
+      "loss": 0.0263,
+      "step": 15647
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 1.3058194774346794e-05,
+      "loss": 0.0524,
+      "step": 15648
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 1.3052256532066509e-05,
+      "loss": 0.0277,
+      "step": 15649
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 1.3046318289786224e-05,
+      "loss": 0.0272,
+      "step": 15650
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 1.304038004750594e-05,
+      "loss": 0.025,
+      "step": 15651
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 1.3034441805225653e-05,
+      "loss": 0.017,
+      "step": 15652
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 1.3028503562945368e-05,
+      "loss": 0.0315,
+      "step": 15653
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 1.3022565320665084e-05,
+      "loss": 0.0196,
+      "step": 15654
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 1.3016627078384799e-05,
+      "loss": 0.0087,
+      "step": 15655
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 1.3010688836104514e-05,
+      "loss": 0.0531,
+      "step": 15656
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 1.3004750593824228e-05,
+      "loss": 0.0215,
+      "step": 15657
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 1.2998812351543943e-05,
+      "loss": 0.0356,
+      "step": 15658
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 1.2992874109263658e-05,
+      "loss": 0.0267,
+      "step": 15659
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 1.2986935866983373e-05,
+      "loss": 0.0367,
+      "step": 15660
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 1.2980997624703089e-05,
+      "loss": 0.0127,
+      "step": 15661
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 1.2975059382422802e-05,
+      "loss": 0.0243,
+      "step": 15662
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 1.2969121140142518e-05,
+      "loss": 0.0499,
+      "step": 15663
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 1.2963182897862233e-05,
+      "loss": 0.099,
+      "step": 15664
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 1.2957244655581948e-05,
+      "loss": 0.0257,
+      "step": 15665
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 1.2951306413301663e-05,
+      "loss": 0.0337,
+      "step": 15666
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 1.2945368171021377e-05,
+      "loss": 0.0472,
+      "step": 15667
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 1.2939429928741092e-05,
+      "loss": 0.0309,
+      "step": 15668
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 1.2933491686460808e-05,
+      "loss": 0.0331,
+      "step": 15669
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 1.2927553444180523e-05,
+      "loss": 0.0428,
+      "step": 15670
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 1.292161520190024e-05,
+      "loss": 0.0338,
+      "step": 15671
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 1.2915676959619955e-05,
+      "loss": 0.0378,
+      "step": 15672
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 1.2909738717339667e-05,
+      "loss": 0.0374,
+      "step": 15673
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 1.2903800475059382e-05,
+      "loss": 0.0351,
+      "step": 15674
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 1.2897862232779098e-05,
+      "loss": 0.0285,
+      "step": 15675
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 1.2891923990498814e-05,
+      "loss": 0.0324,
+      "step": 15676
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 1.288598574821853e-05,
+      "loss": 0.0265,
+      "step": 15677
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 1.2880047505938242e-05,
+      "loss": 0.0393,
+      "step": 15678
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 1.2874109263657957e-05,
+      "loss": 0.037,
+      "step": 15679
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 1.2868171021377672e-05,
+      "loss": 0.0372,
+      "step": 15680
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 1.286223277909739e-05,
+      "loss": 0.0415,
+      "step": 15681
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 1.2856294536817104e-05,
+      "loss": 0.0296,
+      "step": 15682
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 1.2850356294536816e-05,
+      "loss": 0.0268,
+      "step": 15683
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 1.2844418052256532e-05,
+      "loss": 0.0212,
+      "step": 15684
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 1.2838479809976247e-05,
+      "loss": 0.024,
+      "step": 15685
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 1.2832541567695964e-05,
+      "loss": 0.0746,
+      "step": 15686
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 1.2826603325415679e-05,
+      "loss": 0.0345,
+      "step": 15687
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 1.2820665083135391e-05,
+      "loss": 0.1319,
+      "step": 15688
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 1.2814726840855106e-05,
+      "loss": 0.0343,
+      "step": 15689
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 1.2808788598574822e-05,
+      "loss": 0.0319,
+      "step": 15690
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 1.2802850356294539e-05,
+      "loss": 0.0578,
+      "step": 15691
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 1.2796912114014254e-05,
+      "loss": 0.0385,
+      "step": 15692
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 1.2790973871733966e-05,
+      "loss": 0.0237,
+      "step": 15693
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 1.2785035629453681e-05,
+      "loss": 0.037,
+      "step": 15694
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 1.2779097387173398e-05,
+      "loss": 0.0193,
+      "step": 15695
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 1.2773159144893113e-05,
+      "loss": 0.0283,
+      "step": 15696
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 1.2767220902612828e-05,
+      "loss": 0.0745,
+      "step": 15697
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 1.276128266033254e-05,
+      "loss": 0.0409,
+      "step": 15698
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 1.2755344418052256e-05,
+      "loss": 0.0298,
+      "step": 15699
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 1.2749406175771973e-05,
+      "loss": 0.028,
+      "step": 15700
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 1.2743467933491688e-05,
+      "loss": 0.0252,
+      "step": 15701
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 1.2737529691211403e-05,
+      "loss": 0.0475,
+      "step": 15702
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 1.2731591448931118e-05,
+      "loss": 0.0139,
+      "step": 15703
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 1.272565320665083e-05,
+      "loss": 0.0185,
+      "step": 15704
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 1.2719714964370547e-05,
+      "loss": 0.0188,
+      "step": 15705
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 1.2713776722090263e-05,
+      "loss": 0.0156,
+      "step": 15706
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 1.2707838479809978e-05,
+      "loss": 0.0804,
+      "step": 15707
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 1.2701900237529693e-05,
+      "loss": 0.0165,
+      "step": 15708
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 1.2695961995249405e-05,
+      "loss": 0.0319,
+      "step": 15709
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 1.2690023752969122e-05,
+      "loss": 0.0536,
+      "step": 15710
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 1.2684085510688837e-05,
+      "loss": 0.0084,
+      "step": 15711
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 1.2678147268408553e-05,
+      "loss": 0.0188,
+      "step": 15712
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 1.2672209026128268e-05,
+      "loss": 0.0185,
+      "step": 15713
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 1.266627078384798e-05,
+      "loss": 0.0518,
+      "step": 15714
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 1.2660332541567697e-05,
+      "loss": 0.0452,
+      "step": 15715
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 1.2654394299287412e-05,
+      "loss": 0.0228,
+      "step": 15716
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 1.2648456057007127e-05,
+      "loss": 0.0565,
+      "step": 15717
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 1.2642517814726842e-05,
+      "loss": 0.0385,
+      "step": 15718
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 1.2636579572446556e-05,
+      "loss": 0.0282,
+      "step": 15719
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 1.2630641330166271e-05,
+      "loss": 0.0477,
+      "step": 15720
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 1.2624703087885987e-05,
+      "loss": 0.0524,
+      "step": 15721
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 1.2618764845605702e-05,
+      "loss": 0.0375,
+      "step": 15722
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 1.2612826603325417e-05,
+      "loss": 0.0379,
+      "step": 15723
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 1.260688836104513e-05,
+      "loss": 0.0434,
+      "step": 15724
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 1.2600950118764846e-05,
+      "loss": 0.0378,
+      "step": 15725
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 1.2595011876484561e-05,
+      "loss": 0.0384,
+      "step": 15726
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 1.2589073634204277e-05,
+      "loss": 0.0296,
+      "step": 15727
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 1.2583135391923992e-05,
+      "loss": 0.04,
+      "step": 15728
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 1.2577197149643707e-05,
+      "loss": 0.0197,
+      "step": 15729
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 1.257125890736342e-05,
+      "loss": 0.033,
+      "step": 15730
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 1.2565320665083136e-05,
+      "loss": 0.0587,
+      "step": 15731
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 1.2559382422802851e-05,
+      "loss": 0.0258,
+      "step": 15732
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 1.2553444180522567e-05,
+      "loss": 0.0445,
+      "step": 15733
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 1.2547505938242282e-05,
+      "loss": 0.066,
+      "step": 15734
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 1.2541567695961995e-05,
+      "loss": 0.0497,
+      "step": 15735
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 1.253562945368171e-05,
+      "loss": 0.0279,
+      "step": 15736
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 1.2529691211401426e-05,
+      "loss": 0.0289,
+      "step": 15737
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 1.2523752969121141e-05,
+      "loss": 0.0223,
+      "step": 15738
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 1.2517814726840856e-05,
+      "loss": 0.0375,
+      "step": 15739
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 1.251187648456057e-05,
+      "loss": 0.0165,
+      "step": 15740
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 1.2505938242280285e-05,
+      "loss": 0.0659,
+      "step": 15741
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 1.25e-05,
+      "loss": 0.045,
+      "step": 15742
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 1.2494061757719716e-05,
+      "loss": 0.0314,
+      "step": 15743
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 1.248812351543943e-05,
+      "loss": 0.0319,
+      "step": 15744
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 1.2482185273159146e-05,
+      "loss": 0.0421,
+      "step": 15745
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 1.247624703087886e-05,
+      "loss": 0.0203,
+      "step": 15746
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 1.2470308788598575e-05,
+      "loss": 0.0211,
+      "step": 15747
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 1.246437054631829e-05,
+      "loss": 0.0281,
+      "step": 15748
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 1.2458432304038004e-05,
+      "loss": 0.0247,
+      "step": 15749
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 1.2452494061757721e-05,
+      "loss": 0.0264,
+      "step": 15750
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 1.2446555819477436e-05,
+      "loss": 0.0364,
+      "step": 15751
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 1.244061757719715e-05,
+      "loss": 0.0193,
+      "step": 15752
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 1.2434679334916865e-05,
+      "loss": 0.0253,
+      "step": 15753
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 1.2428741092636579e-05,
+      "loss": 0.0601,
+      "step": 15754
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 1.2422802850356296e-05,
+      "loss": 0.0289,
+      "step": 15755
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 1.2416864608076011e-05,
+      "loss": 0.0379,
+      "step": 15756
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 1.2410926365795725e-05,
+      "loss": 0.0242,
+      "step": 15757
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 1.240498812351544e-05,
+      "loss": 0.0227,
+      "step": 15758
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 1.2399049881235155e-05,
+      "loss": 0.0243,
+      "step": 15759
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 1.239311163895487e-05,
+      "loss": 0.0212,
+      "step": 15760
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 1.2387173396674586e-05,
+      "loss": 0.0287,
+      "step": 15761
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 1.23812351543943e-05,
+      "loss": 0.0159,
+      "step": 15762
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 1.2375296912114015e-05,
+      "loss": 0.0446,
+      "step": 15763
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 1.236935866983373e-05,
+      "loss": 0.0501,
+      "step": 15764
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 1.2363420427553445e-05,
+      "loss": 0.05,
+      "step": 15765
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 1.235748218527316e-05,
+      "loss": 0.0338,
+      "step": 15766
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 1.2351543942992874e-05,
+      "loss": 0.0623,
+      "step": 15767
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 1.234560570071259e-05,
+      "loss": 0.035,
+      "step": 15768
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 1.2339667458432305e-05,
+      "loss": 0.0486,
+      "step": 15769
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 1.233372921615202e-05,
+      "loss": 0.0676,
+      "step": 15770
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 1.2327790973871735e-05,
+      "loss": 0.0274,
+      "step": 15771
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 1.2321852731591449e-05,
+      "loss": 0.038,
+      "step": 15772
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 1.2315914489311164e-05,
+      "loss": 0.0404,
+      "step": 15773
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 1.230997624703088e-05,
+      "loss": 0.036,
+      "step": 15774
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 1.2304038004750595e-05,
+      "loss": 0.039,
+      "step": 15775
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 1.229809976247031e-05,
+      "loss": 0.0315,
+      "step": 15776
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 1.2292161520190023e-05,
+      "loss": 0.0336,
+      "step": 15777
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 1.2286223277909739e-05,
+      "loss": 0.0305,
+      "step": 15778
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 1.2280285035629454e-05,
+      "loss": 0.0148,
+      "step": 15779
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 1.227434679334917e-05,
+      "loss": 0.0324,
+      "step": 15780
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 1.2268408551068884e-05,
+      "loss": 0.0219,
+      "step": 15781
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 1.22624703087886e-05,
+      "loss": 0.0328,
+      "step": 15782
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 1.2256532066508313e-05,
+      "loss": 0.0143,
+      "step": 15783
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 1.2250593824228029e-05,
+      "loss": 0.0328,
+      "step": 15784
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 1.2244655581947744e-05,
+      "loss": 0.0182,
+      "step": 15785
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 1.2238717339667459e-05,
+      "loss": 0.0481,
+      "step": 15786
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 1.2232779097387174e-05,
+      "loss": 0.0471,
+      "step": 15787
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 1.2226840855106888e-05,
+      "loss": 0.0279,
+      "step": 15788
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 1.2220902612826603e-05,
+      "loss": 0.0354,
+      "step": 15789
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 1.2214964370546319e-05,
+      "loss": 0.0317,
+      "step": 15790
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 1.2209026128266034e-05,
+      "loss": 0.0418,
+      "step": 15791
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 1.2203087885985749e-05,
+      "loss": 0.0377,
+      "step": 15792
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 1.2197149643705463e-05,
+      "loss": 0.0294,
+      "step": 15793
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 1.219121140142518e-05,
+      "loss": 0.0444,
+      "step": 15794
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 1.2185273159144895e-05,
+      "loss": 0.0359,
+      "step": 15795
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 1.2179334916864608e-05,
+      "loss": 0.0496,
+      "step": 15796
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 1.2173396674584324e-05,
+      "loss": 0.0239,
+      "step": 15797
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 1.2167458432304037e-05,
+      "loss": 0.0321,
+      "step": 15798
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 1.2161520190023754e-05,
+      "loss": 0.026,
+      "step": 15799
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 1.215558194774347e-05,
+      "loss": 0.0218,
+      "step": 15800
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 1.2149643705463183e-05,
+      "loss": 0.0403,
+      "step": 15801
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 1.2143705463182898e-05,
+      "loss": 0.0121,
+      "step": 15802
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 1.2137767220902612e-05,
+      "loss": 0.03,
+      "step": 15803
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 1.2131828978622329e-05,
+      "loss": 0.0271,
+      "step": 15804
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 1.2125890736342044e-05,
+      "loss": 0.063,
+      "step": 15805
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 1.2119952494061758e-05,
+      "loss": 0.0501,
+      "step": 15806
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 1.2114014251781473e-05,
+      "loss": 0.022,
+      "step": 15807
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 1.2108076009501188e-05,
+      "loss": 0.0149,
+      "step": 15808
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 1.2102137767220904e-05,
+      "loss": 0.0241,
+      "step": 15809
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 1.2096199524940619e-05,
+      "loss": 0.0476,
+      "step": 15810
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 1.2090261282660333e-05,
+      "loss": 0.0287,
+      "step": 15811
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 1.2084323040380048e-05,
+      "loss": 0.0366,
+      "step": 15812
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 1.2078384798099763e-05,
+      "loss": 0.0387,
+      "step": 15813
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 1.2072446555819478e-05,
+      "loss": 0.08,
+      "step": 15814
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 1.2066508313539194e-05,
+      "loss": 0.0265,
+      "step": 15815
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 1.2060570071258907e-05,
+      "loss": 0.0489,
+      "step": 15816
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 1.2054631828978622e-05,
+      "loss": 0.0502,
+      "step": 15817
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 1.2048693586698338e-05,
+      "loss": 0.0346,
+      "step": 15818
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 1.2042755344418053e-05,
+      "loss": 0.0375,
+      "step": 15819
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 1.2036817102137768e-05,
+      "loss": 0.0145,
+      "step": 15820
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 1.2030878859857484e-05,
+      "loss": 0.0592,
+      "step": 15821
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 1.2024940617577197e-05,
+      "loss": 0.0531,
+      "step": 15822
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 1.2019002375296912e-05,
+      "loss": 0.0487,
+      "step": 15823
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 1.2013064133016628e-05,
+      "loss": 0.035,
+      "step": 15824
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 1.2007125890736343e-05,
+      "loss": 0.02,
+      "step": 15825
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 1.2001187648456058e-05,
+      "loss": 0.034,
+      "step": 15826
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 1.1995249406175772e-05,
+      "loss": 0.0375,
+      "step": 15827
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 1.1989311163895487e-05,
+      "loss": 0.0287,
+      "step": 15828
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 1.1983372921615202e-05,
+      "loss": 0.0163,
+      "step": 15829
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 1.1977434679334918e-05,
+      "loss": 0.0401,
+      "step": 15830
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 1.1971496437054633e-05,
+      "loss": 0.0319,
+      "step": 15831
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 1.1965558194774347e-05,
+      "loss": 0.0244,
+      "step": 15832
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 1.1959619952494062e-05,
+      "loss": 0.0361,
+      "step": 15833
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 1.1953681710213779e-05,
+      "loss": 0.035,
+      "step": 15834
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 1.1947743467933492e-05,
+      "loss": 0.0456,
+      "step": 15835
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 1.1941805225653208e-05,
+      "loss": 0.037,
+      "step": 15836
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 1.1935866983372921e-05,
+      "loss": 0.0236,
+      "step": 15837
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 1.1929928741092636e-05,
+      "loss": 0.0237,
+      "step": 15838
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 1.1923990498812353e-05,
+      "loss": 0.0273,
+      "step": 15839
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 1.1918052256532067e-05,
+      "loss": 0.0463,
+      "step": 15840
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 1.1912114014251782e-05,
+      "loss": 0.0199,
+      "step": 15841
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 1.1906175771971496e-05,
+      "loss": 0.0315,
+      "step": 15842
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 1.1900237529691211e-05,
+      "loss": 0.2625,
+      "step": 15843
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 1.1894299287410928e-05,
+      "loss": 0.0169,
+      "step": 15844
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 1.1888361045130642e-05,
+      "loss": 0.0432,
+      "step": 15845
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 1.1882422802850357e-05,
+      "loss": 0.0179,
+      "step": 15846
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 1.187648456057007e-05,
+      "loss": 0.0169,
+      "step": 15847
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 1.1870546318289786e-05,
+      "loss": 0.0352,
+      "step": 15848
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 1.1864608076009503e-05,
+      "loss": 0.0287,
+      "step": 15849
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 1.1858669833729216e-05,
+      "loss": 0.0386,
+      "step": 15850
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 1.1852731591448932e-05,
+      "loss": 0.0331,
+      "step": 15851
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 1.1846793349168647e-05,
+      "loss": 0.0051,
+      "step": 15852
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 1.184085510688836e-05,
+      "loss": 0.019,
+      "step": 15853
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 1.1834916864608077e-05,
+      "loss": 0.0226,
+      "step": 15854
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 1.1828978622327791e-05,
+      "loss": 0.0387,
+      "step": 15855
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 1.1823040380047506e-05,
+      "loss": 0.0189,
+      "step": 15856
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 1.1817102137767222e-05,
+      "loss": 0.0147,
+      "step": 15857
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 1.1811163895486937e-05,
+      "loss": 0.0221,
+      "step": 15858
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 1.1805225653206652e-05,
+      "loss": 0.0148,
+      "step": 15859
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 1.1799287410926366e-05,
+      "loss": 0.0213,
+      "step": 15860
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 1.1793349168646081e-05,
+      "loss": 0.0346,
+      "step": 15861
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 1.1787410926365796e-05,
+      "loss": 0.0325,
+      "step": 15862
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 1.1781472684085512e-05,
+      "loss": 0.0279,
+      "step": 15863
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 1.1775534441805227e-05,
+      "loss": 0.0258,
+      "step": 15864
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 1.1769596199524942e-05,
+      "loss": 0.0498,
+      "step": 15865
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 1.1763657957244656e-05,
+      "loss": 0.0504,
+      "step": 15866
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 1.1757719714964371e-05,
+      "loss": 0.0584,
+      "step": 15867
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 1.1751781472684086e-05,
+      "loss": 0.0328,
+      "step": 15868
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 1.1745843230403802e-05,
+      "loss": 0.0315,
+      "step": 15869
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 1.1739904988123517e-05,
+      "loss": 0.0395,
+      "step": 15870
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 1.173396674584323e-05,
+      "loss": 0.0251,
+      "step": 15871
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 1.1728028503562946e-05,
+      "loss": 0.0212,
+      "step": 15872
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 1.1722090261282661e-05,
+      "loss": 0.0485,
+      "step": 15873
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 1.1716152019002376e-05,
+      "loss": 0.0309,
+      "step": 15874
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 1.1710213776722091e-05,
+      "loss": 0.0225,
+      "step": 15875
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 1.1704275534441805e-05,
+      "loss": 0.0313,
+      "step": 15876
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 1.169833729216152e-05,
+      "loss": 0.0303,
+      "step": 15877
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 1.1692399049881236e-05,
+      "loss": 0.022,
+      "step": 15878
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 1.1686460807600951e-05,
+      "loss": 0.0355,
+      "step": 15879
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 1.1680522565320666e-05,
+      "loss": 0.0283,
+      "step": 15880
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 1.167458432304038e-05,
+      "loss": 0.0255,
+      "step": 15881
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 1.1668646080760095e-05,
+      "loss": 0.0463,
+      "step": 15882
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 1.166270783847981e-05,
+      "loss": 0.0397,
+      "step": 15883
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 1.1656769596199526e-05,
+      "loss": 0.0381,
+      "step": 15884
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 1.165083135391924e-05,
+      "loss": 0.0295,
+      "step": 15885
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 1.1644893111638954e-05,
+      "loss": 0.0339,
+      "step": 15886
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 1.163895486935867e-05,
+      "loss": 0.03,
+      "step": 15887
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 1.1633016627078385e-05,
+      "loss": 0.0232,
+      "step": 15888
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 1.16270783847981e-05,
+      "loss": 0.023,
+      "step": 15889
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 1.1621140142517816e-05,
+      "loss": 0.0325,
+      "step": 15890
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 1.161520190023753e-05,
+      "loss": 0.0447,
+      "step": 15891
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 1.1609263657957244e-05,
+      "loss": 0.0326,
+      "step": 15892
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 1.1603325415676961e-05,
+      "loss": 0.033,
+      "step": 15893
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 1.1597387173396675e-05,
+      "loss": 0.0601,
+      "step": 15894
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 1.159144893111639e-05,
+      "loss": 0.0199,
+      "step": 15895
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 1.1585510688836105e-05,
+      "loss": 0.0407,
+      "step": 15896
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 1.1579572446555819e-05,
+      "loss": 0.04,
+      "step": 15897
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 1.1573634204275536e-05,
+      "loss": 0.0501,
+      "step": 15898
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 1.156769596199525e-05,
+      "loss": 0.0308,
+      "step": 15899
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 1.1561757719714965e-05,
+      "loss": 0.0318,
+      "step": 15900
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 1.155581947743468e-05,
+      "loss": 0.0271,
+      "step": 15901
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 1.1549881235154394e-05,
+      "loss": 0.0214,
+      "step": 15902
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 1.154394299287411e-05,
+      "loss": 0.0499,
+      "step": 15903
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 1.1538004750593826e-05,
+      "loss": 0.0331,
+      "step": 15904
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 1.153206650831354e-05,
+      "loss": 0.0315,
+      "step": 15905
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 1.1526128266033255e-05,
+      "loss": 0.0477,
+      "step": 15906
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 1.1520190023752968e-05,
+      "loss": 0.0157,
+      "step": 15907
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 1.1514251781472685e-05,
+      "loss": 0.0328,
+      "step": 15908
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 1.15083135391924e-05,
+      "loss": 0.0065,
+      "step": 15909
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 1.1502375296912114e-05,
+      "loss": 0.0499,
+      "step": 15910
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 1.149643705463183e-05,
+      "loss": 0.0105,
+      "step": 15911
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 1.1490498812351543e-05,
+      "loss": 0.0127,
+      "step": 15912
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 1.148456057007126e-05,
+      "loss": 0.0141,
+      "step": 15913
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 1.1478622327790975e-05,
+      "loss": 0.0506,
+      "step": 15914
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 1.1472684085510689e-05,
+      "loss": 0.0372,
+      "step": 15915
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 1.1466745843230404e-05,
+      "loss": 0.0519,
+      "step": 15916
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 1.146080760095012e-05,
+      "loss": 0.0329,
+      "step": 15917
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 1.1454869358669835e-05,
+      "loss": 0.0837,
+      "step": 15918
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 1.144893111638955e-05,
+      "loss": 0.0373,
+      "step": 15919
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 1.1442992874109264e-05,
+      "loss": 0.0184,
+      "step": 15920
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 1.1437054631828979e-05,
+      "loss": 0.0394,
+      "step": 15921
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 1.1431116389548694e-05,
+      "loss": 0.0483,
+      "step": 15922
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 1.142517814726841e-05,
+      "loss": 0.0494,
+      "step": 15923
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 1.1419239904988125e-05,
+      "loss": 0.0573,
+      "step": 15924
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 1.1413301662707838e-05,
+      "loss": 0.0201,
+      "step": 15925
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 1.1407363420427554e-05,
+      "loss": 0.0438,
+      "step": 15926
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 1.1401425178147269e-05,
+      "loss": 0.0348,
+      "step": 15927
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 1.1395486935866984e-05,
+      "loss": 0.0254,
+      "step": 15928
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 1.13895486935867e-05,
+      "loss": 0.0222,
+      "step": 15929
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 1.1383610451306413e-05,
+      "loss": 0.0225,
+      "step": 15930
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 1.1377672209026128e-05,
+      "loss": 0.0205,
+      "step": 15931
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 1.1371733966745844e-05,
+      "loss": 0.0184,
+      "step": 15932
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 1.1365795724465559e-05,
+      "loss": 0.0139,
+      "step": 15933
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 1.1359857482185274e-05,
+      "loss": 0.0334,
+      "step": 15934
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 1.135391923990499e-05,
+      "loss": 0.0251,
+      "step": 15935
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 1.1347980997624703e-05,
+      "loss": 0.0212,
+      "step": 15936
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 1.1342042755344418e-05,
+      "loss": 0.0217,
+      "step": 15937
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 1.1336104513064133e-05,
+      "loss": 0.0289,
+      "step": 15938
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 1.1330166270783849e-05,
+      "loss": 0.0354,
+      "step": 15939
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 1.1324228028503564e-05,
+      "loss": 0.051,
+      "step": 15940
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 1.1318289786223278e-05,
+      "loss": 0.025,
+      "step": 15941
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 1.1312351543942993e-05,
+      "loss": 0.035,
+      "step": 15942
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 1.1306413301662708e-05,
+      "loss": 0.0241,
+      "step": 15943
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 1.1300475059382423e-05,
+      "loss": 0.0323,
+      "step": 15944
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 1.1294536817102139e-05,
+      "loss": 0.0225,
+      "step": 15945
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 1.1288598574821852e-05,
+      "loss": 0.0211,
+      "step": 15946
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 1.1282660332541568e-05,
+      "loss": 0.0395,
+      "step": 15947
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 1.1276722090261285e-05,
+      "loss": 0.0279,
+      "step": 15948
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 1.1270783847980998e-05,
+      "loss": 0.015,
+      "step": 15949
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 1.1264845605700713e-05,
+      "loss": 0.0134,
+      "step": 15950
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 1.1258907363420427e-05,
+      "loss": 0.0519,
+      "step": 15951
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 1.1252969121140142e-05,
+      "loss": 0.0172,
+      "step": 15952
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 1.124703087885986e-05,
+      "loss": 0.0349,
+      "step": 15953
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 1.1241092636579573e-05,
+      "loss": 0.0244,
+      "step": 15954
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 1.1235154394299288e-05,
+      "loss": 0.0277,
+      "step": 15955
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 1.1229216152019002e-05,
+      "loss": 0.0396,
+      "step": 15956
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 1.1223277909738719e-05,
+      "loss": 0.0283,
+      "step": 15957
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 1.1217339667458434e-05,
+      "loss": 0.0222,
+      "step": 15958
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 1.1211401425178147e-05,
+      "loss": 0.0312,
+      "step": 15959
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 1.1205463182897863e-05,
+      "loss": 0.049,
+      "step": 15960
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 1.1199524940617578e-05,
+      "loss": 0.0238,
+      "step": 15961
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 1.1193586698337293e-05,
+      "loss": 0.0678,
+      "step": 15962
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 1.1187648456057009e-05,
+      "loss": 0.0198,
+      "step": 15963
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 1.1181710213776722e-05,
+      "loss": 0.0721,
+      "step": 15964
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 1.1175771971496437e-05,
+      "loss": 0.0606,
+      "step": 15965
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 1.1169833729216153e-05,
+      "loss": 0.0489,
+      "step": 15966
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 1.1163895486935868e-05,
+      "loss": 0.056,
+      "step": 15967
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 1.1157957244655583e-05,
+      "loss": 0.0361,
+      "step": 15968
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 1.1152019002375297e-05,
+      "loss": 0.0421,
+      "step": 15969
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 1.1146080760095012e-05,
+      "loss": 0.0217,
+      "step": 15970
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 1.1140142517814727e-05,
+      "loss": 0.0321,
+      "step": 15971
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 1.1134204275534443e-05,
+      "loss": 0.0751,
+      "step": 15972
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 1.1128266033254158e-05,
+      "loss": 0.0177,
+      "step": 15973
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 1.1122327790973873e-05,
+      "loss": 0.0262,
+      "step": 15974
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 1.1116389548693587e-05,
+      "loss": 0.0184,
+      "step": 15975
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 1.1110451306413302e-05,
+      "loss": 0.0248,
+      "step": 15976
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 1.1104513064133017e-05,
+      "loss": 0.0156,
+      "step": 15977
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 1.1098574821852733e-05,
+      "loss": 0.0382,
+      "step": 15978
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 1.1092636579572448e-05,
+      "loss": 0.0232,
+      "step": 15979
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 1.1086698337292161e-05,
+      "loss": 0.0282,
+      "step": 15980
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 1.1080760095011877e-05,
+      "loss": 0.0578,
+      "step": 15981
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 1.1074821852731592e-05,
+      "loss": 0.0329,
+      "step": 15982
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 1.1068883610451307e-05,
+      "loss": 0.0163,
+      "step": 15983
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 1.1062945368171023e-05,
+      "loss": 0.0392,
+      "step": 15984
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 1.1057007125890736e-05,
+      "loss": 0.0371,
+      "step": 15985
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 1.1051068883610451e-05,
+      "loss": 0.0207,
+      "step": 15986
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 1.1045130641330167e-05,
+      "loss": 0.0525,
+      "step": 15987
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 1.1039192399049882e-05,
+      "loss": 0.0195,
+      "step": 15988
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 1.1033254156769597e-05,
+      "loss": 0.0329,
+      "step": 15989
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 1.102731591448931e-05,
+      "loss": 0.0275,
+      "step": 15990
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 1.1021377672209026e-05,
+      "loss": 0.0435,
+      "step": 15991
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 1.1015439429928743e-05,
+      "loss": 0.0217,
+      "step": 15992
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 1.1009501187648457e-05,
+      "loss": 0.0382,
+      "step": 15993
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 1.1003562945368172e-05,
+      "loss": 0.0289,
+      "step": 15994
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 1.0997624703087885e-05,
+      "loss": 0.0321,
+      "step": 15995
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 1.09916864608076e-05,
+      "loss": 0.0275,
+      "step": 15996
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 1.0985748218527318e-05,
+      "loss": 0.0297,
+      "step": 15997
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 1.0979809976247031e-05,
+      "loss": 0.0536,
+      "step": 15998
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 1.0973871733966747e-05,
+      "loss": 0.0281,
+      "step": 15999
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 1.096793349168646e-05,
+      "loss": 0.0376,
+      "step": 16000
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 1.0961995249406175e-05,
+      "loss": 0.0495,
+      "step": 16001
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 1.0956057007125892e-05,
+      "loss": 0.0174,
+      "step": 16002
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 1.0950118764845606e-05,
+      "loss": 0.0322,
+      "step": 16003
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 1.0944180522565321e-05,
+      "loss": 0.059,
+      "step": 16004
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 1.0938242280285037e-05,
+      "loss": 0.0337,
+      "step": 16005
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 1.093230403800475e-05,
+      "loss": 0.0233,
+      "step": 16006
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 1.0926365795724467e-05,
+      "loss": 0.0192,
+      "step": 16007
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 1.092042755344418e-05,
+      "loss": 0.0374,
+      "step": 16008
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 1.0914489311163896e-05,
+      "loss": 0.0464,
+      "step": 16009
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 1.0908551068883611e-05,
+      "loss": 0.0112,
+      "step": 16010
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 1.0902612826603325e-05,
+      "loss": 0.014,
+      "step": 16011
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 1.0896674584323042e-05,
+      "loss": 0.025,
+      "step": 16012
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 1.0890736342042755e-05,
+      "loss": 0.0258,
+      "step": 16013
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 1.088479809976247e-05,
+      "loss": 0.0206,
+      "step": 16014
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 1.0878859857482186e-05,
+      "loss": 0.0704,
+      "step": 16015
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 1.0872921615201901e-05,
+      "loss": 0.0506,
+      "step": 16016
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 1.0866983372921616e-05,
+      "loss": 0.0262,
+      "step": 16017
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 1.0861045130641332e-05,
+      "loss": 0.0333,
+      "step": 16018
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 1.0855106888361045e-05,
+      "loss": 0.0503,
+      "step": 16019
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 1.084916864608076e-05,
+      "loss": 0.0272,
+      "step": 16020
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 1.0843230403800476e-05,
+      "loss": 0.0604,
+      "step": 16021
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 1.0837292161520191e-05,
+      "loss": 0.0337,
+      "step": 16022
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 1.0831353919239906e-05,
+      "loss": 0.0457,
+      "step": 16023
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 1.082541567695962e-05,
+      "loss": 0.0254,
+      "step": 16024
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 1.0819477434679335e-05,
+      "loss": 0.04,
+      "step": 16025
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 1.081353919239905e-05,
+      "loss": 0.035,
+      "step": 16026
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 1.0807600950118766e-05,
+      "loss": 0.0534,
+      "step": 16027
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 1.0801662707838481e-05,
+      "loss": 0.0268,
+      "step": 16028
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 1.0795724465558195e-05,
+      "loss": 0.0405,
+      "step": 16029
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 1.078978622327791e-05,
+      "loss": 0.0237,
+      "step": 16030
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 1.0783847980997625e-05,
+      "loss": 0.0194,
+      "step": 16031
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 1.077790973871734e-05,
+      "loss": 0.0183,
+      "step": 16032
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 1.0771971496437056e-05,
+      "loss": 0.0304,
+      "step": 16033
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 1.076603325415677e-05,
+      "loss": 0.0284,
+      "step": 16034
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 1.0760095011876485e-05,
+      "loss": 0.0334,
+      "step": 16035
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 1.07541567695962e-05,
+      "loss": 0.0196,
+      "step": 16036
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 1.0748218527315915e-05,
+      "loss": 0.0204,
+      "step": 16037
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 1.074228028503563e-05,
+      "loss": 0.0427,
+      "step": 16038
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 1.0736342042755344e-05,
+      "loss": 0.0296,
+      "step": 16039
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 1.073040380047506e-05,
+      "loss": 0.0555,
+      "step": 16040
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 1.0724465558194775e-05,
+      "loss": 0.0306,
+      "step": 16041
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 1.071852731591449e-05,
+      "loss": 0.0359,
+      "step": 16042
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 1.0712589073634205e-05,
+      "loss": 0.0221,
+      "step": 16043
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 1.070665083135392e-05,
+      "loss": 0.0175,
+      "step": 16044
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 1.0700712589073634e-05,
+      "loss": 0.0275,
+      "step": 16045
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 1.069477434679335e-05,
+      "loss": 0.0301,
+      "step": 16046
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 1.0688836104513065e-05,
+      "loss": 0.0156,
+      "step": 16047
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 1.068289786223278e-05,
+      "loss": 0.0248,
+      "step": 16048
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 1.0676959619952495e-05,
+      "loss": 0.1264,
+      "step": 16049
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 1.0671021377672209e-05,
+      "loss": 0.03,
+      "step": 16050
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 1.0665083135391924e-05,
+      "loss": 0.0435,
+      "step": 16051
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 1.065914489311164e-05,
+      "loss": 0.0242,
+      "step": 16052
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 1.0653206650831354e-05,
+      "loss": 0.0133,
+      "step": 16053
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 1.064726840855107e-05,
+      "loss": 0.0238,
+      "step": 16054
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 1.0641330166270783e-05,
+      "loss": 0.0327,
+      "step": 16055
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 1.06353919239905e-05,
+      "loss": 0.0315,
+      "step": 16056
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 1.0629453681710214e-05,
+      "loss": 0.0409,
+      "step": 16057
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 1.062351543942993e-05,
+      "loss": 0.0278,
+      "step": 16058
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 1.0617577197149644e-05,
+      "loss": 0.0382,
+      "step": 16059
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 1.0611638954869358e-05,
+      "loss": 0.0345,
+      "step": 16060
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 1.0605700712589075e-05,
+      "loss": 0.0288,
+      "step": 16061
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 1.059976247030879e-05,
+      "loss": 0.0232,
+      "step": 16062
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 1.0593824228028504e-05,
+      "loss": 0.0288,
+      "step": 16063
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 1.0587885985748219e-05,
+      "loss": 0.035,
+      "step": 16064
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 1.0581947743467933e-05,
+      "loss": 0.0492,
+      "step": 16065
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 1.057600950118765e-05,
+      "loss": 0.032,
+      "step": 16066
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 1.0570071258907365e-05,
+      "loss": 0.0239,
+      "step": 16067
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 1.0564133016627079e-05,
+      "loss": 0.0295,
+      "step": 16068
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 1.0558194774346794e-05,
+      "loss": 0.0311,
+      "step": 16069
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 1.0552256532066507e-05,
+      "loss": 0.0293,
+      "step": 16070
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 1.0546318289786224e-05,
+      "loss": 0.0245,
+      "step": 16071
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 1.054038004750594e-05,
+      "loss": 0.0261,
+      "step": 16072
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 1.0534441805225653e-05,
+      "loss": 0.0381,
+      "step": 16073
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 1.0528503562945368e-05,
+      "loss": 0.0116,
+      "step": 16074
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 1.0522565320665084e-05,
+      "loss": 0.0455,
+      "step": 16075
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 1.0516627078384799e-05,
+      "loss": 0.0214,
+      "step": 16076
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 1.0510688836104514e-05,
+      "loss": 0.0191,
+      "step": 16077
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 1.0504750593824228e-05,
+      "loss": 0.0314,
+      "step": 16078
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 1.0498812351543943e-05,
+      "loss": 0.0226,
+      "step": 16079
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 1.0492874109263658e-05,
+      "loss": 0.0211,
+      "step": 16080
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 1.0486935866983374e-05,
+      "loss": 0.0281,
+      "step": 16081
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 1.0480997624703089e-05,
+      "loss": 0.0314,
+      "step": 16082
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 1.0475059382422803e-05,
+      "loss": 0.0182,
+      "step": 16083
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 1.0469121140142518e-05,
+      "loss": 0.0313,
+      "step": 16084
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 1.0463182897862233e-05,
+      "loss": 0.013,
+      "step": 16085
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 1.0457244655581948e-05,
+      "loss": 0.0132,
+      "step": 16086
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 1.0451306413301664e-05,
+      "loss": 0.0237,
+      "step": 16087
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 1.0445368171021379e-05,
+      "loss": 0.108,
+      "step": 16088
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 1.0439429928741093e-05,
+      "loss": 0.0184,
+      "step": 16089
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 1.0433491686460808e-05,
+      "loss": 0.0276,
+      "step": 16090
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 1.0427553444180523e-05,
+      "loss": 0.0168,
+      "step": 16091
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 1.0421615201900238e-05,
+      "loss": 0.0265,
+      "step": 16092
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 1.0415676959619954e-05,
+      "loss": 0.0374,
+      "step": 16093
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 1.0409738717339667e-05,
+      "loss": 0.0166,
+      "step": 16094
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 1.0403800475059382e-05,
+      "loss": 0.0834,
+      "step": 16095
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 1.0397862232779098e-05,
+      "loss": 0.0324,
+      "step": 16096
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 1.0391923990498813e-05,
+      "loss": 0.0303,
+      "step": 16097
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 1.0385985748218528e-05,
+      "loss": 0.0212,
+      "step": 16098
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 1.0380047505938242e-05,
+      "loss": 0.0378,
+      "step": 16099
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 1.0374109263657957e-05,
+      "loss": 0.065,
+      "step": 16100
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 1.0368171021377674e-05,
+      "loss": 0.0239,
+      "step": 16101
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 1.0362232779097388e-05,
+      "loss": 0.0191,
+      "step": 16102
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 1.0356294536817103e-05,
+      "loss": 0.013,
+      "step": 16103
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 1.0350356294536817e-05,
+      "loss": 0.0126,
+      "step": 16104
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 1.0344418052256532e-05,
+      "loss": 0.0562,
+      "step": 16105
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 1.0338479809976249e-05,
+      "loss": 0.0359,
+      "step": 16106
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 1.0332541567695962e-05,
+      "loss": 0.0443,
+      "step": 16107
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 1.0326603325415678e-05,
+      "loss": 0.0197,
+      "step": 16108
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 1.0320665083135391e-05,
+      "loss": 0.0272,
+      "step": 16109
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 1.0314726840855107e-05,
+      "loss": 0.0246,
+      "step": 16110
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 1.0308788598574823e-05,
+      "loss": 0.0297,
+      "step": 16111
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 1.0302850356294537e-05,
+      "loss": 0.0172,
+      "step": 16112
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 1.0296912114014252e-05,
+      "loss": 0.0495,
+      "step": 16113
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 1.0290973871733968e-05,
+      "loss": 0.0392,
+      "step": 16114
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 1.0285035629453683e-05,
+      "loss": 0.0286,
+      "step": 16115
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 1.0279097387173398e-05,
+      "loss": 0.0301,
+      "step": 16116
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 1.0273159144893112e-05,
+      "loss": 0.01,
+      "step": 16117
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 1.0267220902612827e-05,
+      "loss": 0.0231,
+      "step": 16118
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 1.0261282660332542e-05,
+      "loss": 0.014,
+      "step": 16119
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 1.0255344418052258e-05,
+      "loss": 0.0489,
+      "step": 16120
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 1.0249406175771973e-05,
+      "loss": 0.0186,
+      "step": 16121
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 1.0243467933491686e-05,
+      "loss": 0.0335,
+      "step": 16122
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 1.0237529691211402e-05,
+      "loss": 0.0427,
+      "step": 16123
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 1.0231591448931117e-05,
+      "loss": 0.0309,
+      "step": 16124
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 1.0225653206650832e-05,
+      "loss": 0.0149,
+      "step": 16125
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 1.0219714964370548e-05,
+      "loss": 0.0266,
+      "step": 16126
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 1.0213776722090261e-05,
+      "loss": 0.0287,
+      "step": 16127
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 1.0207838479809976e-05,
+      "loss": 0.0458,
+      "step": 16128
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 1.0201900237529692e-05,
+      "loss": 0.0209,
+      "step": 16129
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 1.0195961995249407e-05,
+      "loss": 0.0354,
+      "step": 16130
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 1.0190023752969122e-05,
+      "loss": 0.0248,
+      "step": 16131
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 1.0184085510688837e-05,
+      "loss": 0.0416,
+      "step": 16132
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 1.0178147268408551e-05,
+      "loss": 0.0155,
+      "step": 16133
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 1.0172209026128266e-05,
+      "loss": 0.0366,
+      "step": 16134
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 1.0166270783847982e-05,
+      "loss": 0.0235,
+      "step": 16135
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 1.0160332541567697e-05,
+      "loss": 0.0358,
+      "step": 16136
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 1.0154394299287412e-05,
+      "loss": 0.0218,
+      "step": 16137
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 1.0148456057007126e-05,
+      "loss": 0.0295,
+      "step": 16138
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 1.0142517814726841e-05,
+      "loss": 0.0305,
+      "step": 16139
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 1.0136579572446556e-05,
+      "loss": 0.0306,
+      "step": 16140
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 1.0130641330166272e-05,
+      "loss": 0.0244,
+      "step": 16141
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 1.0124703087885987e-05,
+      "loss": 0.0302,
+      "step": 16142
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 1.01187648456057e-05,
+      "loss": 0.0073,
+      "step": 16143
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 1.0112826603325416e-05,
+      "loss": 0.0109,
+      "step": 16144
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 1.0106888361045131e-05,
+      "loss": 0.0475,
+      "step": 16145
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 1.0100950118764846e-05,
+      "loss": 0.0255,
+      "step": 16146
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 1.0095011876484562e-05,
+      "loss": 0.0362,
+      "step": 16147
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 1.0089073634204275e-05,
+      "loss": 0.0254,
+      "step": 16148
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 1.008313539192399e-05,
+      "loss": 0.0293,
+      "step": 16149
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 1.0077197149643706e-05,
+      "loss": 0.0223,
+      "step": 16150
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 1.0071258907363421e-05,
+      "loss": 0.0624,
+      "step": 16151
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 1.0065320665083136e-05,
+      "loss": 0.0063,
+      "step": 16152
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 1.005938242280285e-05,
+      "loss": 0.0627,
+      "step": 16153
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 1.0053444180522565e-05,
+      "loss": 0.0322,
+      "step": 16154
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 1.0047505938242282e-05,
+      "loss": 0.0076,
+      "step": 16155
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 1.0041567695961996e-05,
+      "loss": 0.0206,
+      "step": 16156
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 1.0035629453681711e-05,
+      "loss": 0.0302,
+      "step": 16157
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 1.0029691211401426e-05,
+      "loss": 0.0474,
+      "step": 16158
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 1.002375296912114e-05,
+      "loss": 0.0277,
+      "step": 16159
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 1.0017814726840857e-05,
+      "loss": 0.0413,
+      "step": 16160
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 1.001187648456057e-05,
+      "loss": 0.0471,
+      "step": 16161
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 1.0005938242280286e-05,
+      "loss": 0.0196,
+      "step": 16162
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 1e-05,
+      "loss": 0.0296,
+      "step": 16163
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 9.994061757719714e-06,
+      "loss": 0.0357,
+      "step": 16164
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 9.988123515439431e-06,
+      "loss": 0.0224,
+      "step": 16165
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 9.982185273159145e-06,
+      "loss": 0.0533,
+      "step": 16166
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 9.97624703087886e-06,
+      "loss": 0.0176,
+      "step": 16167
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 9.970308788598575e-06,
+      "loss": 0.038,
+      "step": 16168
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 9.964370546318289e-06,
+      "loss": 0.0159,
+      "step": 16169
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 9.958432304038006e-06,
+      "loss": 0.039,
+      "step": 16170
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 9.952494061757721e-06,
+      "loss": 0.0425,
+      "step": 16171
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 9.946555819477435e-06,
+      "loss": 0.0189,
+      "step": 16172
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 9.94061757719715e-06,
+      "loss": 0.0438,
+      "step": 16173
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 9.934679334916864e-06,
+      "loss": 0.0342,
+      "step": 16174
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 9.92874109263658e-06,
+      "loss": 0.018,
+      "step": 16175
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 9.922802850356296e-06,
+      "loss": 0.0597,
+      "step": 16176
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 9.91686460807601e-06,
+      "loss": 0.0242,
+      "step": 16177
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 9.910926365795725e-06,
+      "loss": 0.0288,
+      "step": 16178
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 9.90498812351544e-06,
+      "loss": 0.0272,
+      "step": 16179
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 9.899049881235155e-06,
+      "loss": 0.0239,
+      "step": 16180
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 9.89311163895487e-06,
+      "loss": 0.0449,
+      "step": 16181
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 9.887173396674584e-06,
+      "loss": 0.0192,
+      "step": 16182
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 9.8812351543943e-06,
+      "loss": 0.0216,
+      "step": 16183
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 9.875296912114015e-06,
+      "loss": 0.0383,
+      "step": 16184
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 9.86935866983373e-06,
+      "loss": 0.0114,
+      "step": 16185
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 9.863420427553445e-06,
+      "loss": 0.0673,
+      "step": 16186
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 9.857482185273159e-06,
+      "loss": 0.0364,
+      "step": 16187
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 9.851543942992874e-06,
+      "loss": 0.0221,
+      "step": 16188
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 9.84560570071259e-06,
+      "loss": 0.0298,
+      "step": 16189
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 9.839667458432305e-06,
+      "loss": 0.0342,
+      "step": 16190
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 9.83372921615202e-06,
+      "loss": 0.0209,
+      "step": 16191
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 9.827790973871734e-06,
+      "loss": 0.0148,
+      "step": 16192
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 9.821852731591449e-06,
+      "loss": 0.0144,
+      "step": 16193
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 9.815914489311164e-06,
+      "loss": 0.0157,
+      "step": 16194
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 9.80997624703088e-06,
+      "loss": 0.0192,
+      "step": 16195
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 9.804038004750595e-06,
+      "loss": 0.0284,
+      "step": 16196
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 9.798099762470308e-06,
+      "loss": 0.046,
+      "step": 16197
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 9.792161520190024e-06,
+      "loss": 0.022,
+      "step": 16198
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 9.786223277909739e-06,
+      "loss": 0.0301,
+      "step": 16199
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 9.780285035629454e-06,
+      "loss": 0.0424,
+      "step": 16200
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 9.77434679334917e-06,
+      "loss": 0.0299,
+      "step": 16201
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 9.768408551068885e-06,
+      "loss": 0.047,
+      "step": 16202
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 9.762470308788598e-06,
+      "loss": 0.0229,
+      "step": 16203
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 9.756532066508314e-06,
+      "loss": 0.0049,
+      "step": 16204
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 9.750593824228029e-06,
+      "loss": 0.0441,
+      "step": 16205
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 9.744655581947744e-06,
+      "loss": 0.0244,
+      "step": 16206
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 9.73871733966746e-06,
+      "loss": 0.0384,
+      "step": 16207
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 9.732779097387173e-06,
+      "loss": 0.0521,
+      "step": 16208
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 9.726840855106888e-06,
+      "loss": 0.0305,
+      "step": 16209
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 9.720902612826603e-06,
+      "loss": 0.0268,
+      "step": 16210
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 9.714964370546319e-06,
+      "loss": 0.0277,
+      "step": 16211
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 9.709026128266034e-06,
+      "loss": 0.0387,
+      "step": 16212
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 9.703087885985748e-06,
+      "loss": 0.032,
+      "step": 16213
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 9.697149643705465e-06,
+      "loss": 0.0165,
+      "step": 16214
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 9.69121140142518e-06,
+      "loss": 0.0365,
+      "step": 16215
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 9.685273159144893e-06,
+      "loss": 0.0369,
+      "step": 16216
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 9.679334916864609e-06,
+      "loss": 0.0367,
+      "step": 16217
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 9.673396674584322e-06,
+      "loss": 0.0256,
+      "step": 16218
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 9.66745843230404e-06,
+      "loss": 0.0409,
+      "step": 16219
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 9.661520190023755e-06,
+      "loss": 0.0206,
+      "step": 16220
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 9.655581947743468e-06,
+      "loss": 0.0822,
+      "step": 16221
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 9.649643705463183e-06,
+      "loss": 0.0293,
+      "step": 16222
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 9.643705463182897e-06,
+      "loss": 0.0331,
+      "step": 16223
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 9.637767220902614e-06,
+      "loss": 0.0256,
+      "step": 16224
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 9.63182897862233e-06,
+      "loss": 0.0419,
+      "step": 16225
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 9.625890736342043e-06,
+      "loss": 0.0254,
+      "step": 16226
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 9.619952494061758e-06,
+      "loss": 0.0412,
+      "step": 16227
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 9.614014251781473e-06,
+      "loss": 0.0273,
+      "step": 16228
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 9.608076009501189e-06,
+      "loss": 0.014,
+      "step": 16229
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 9.602137767220904e-06,
+      "loss": 0.0338,
+      "step": 16230
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 9.596199524940617e-06,
+      "loss": 0.0243,
+      "step": 16231
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 9.590261282660333e-06,
+      "loss": 0.0416,
+      "step": 16232
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 9.584323040380048e-06,
+      "loss": 0.014,
+      "step": 16233
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 9.578384798099763e-06,
+      "loss": 0.0214,
+      "step": 16234
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 9.572446555819479e-06,
+      "loss": 0.0279,
+      "step": 16235
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 9.566508313539192e-06,
+      "loss": 0.0392,
+      "step": 16236
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 9.560570071258907e-06,
+      "loss": 0.0255,
+      "step": 16237
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 9.554631828978623e-06,
+      "loss": 0.0215,
+      "step": 16238
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 9.548693586698338e-06,
+      "loss": 0.0254,
+      "step": 16239
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 9.542755344418053e-06,
+      "loss": 0.0309,
+      "step": 16240
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 9.536817102137769e-06,
+      "loss": 0.0227,
+      "step": 16241
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 9.530878859857482e-06,
+      "loss": 0.0219,
+      "step": 16242
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 9.524940617577197e-06,
+      "loss": 0.0433,
+      "step": 16243
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 9.519002375296913e-06,
+      "loss": 0.0147,
+      "step": 16244
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 9.513064133016628e-06,
+      "loss": 0.0325,
+      "step": 16245
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 9.507125890736343e-06,
+      "loss": 0.0219,
+      "step": 16246
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 9.501187648456057e-06,
+      "loss": 0.0226,
+      "step": 16247
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 9.495249406175772e-06,
+      "loss": 0.0329,
+      "step": 16248
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 9.489311163895487e-06,
+      "loss": 0.0285,
+      "step": 16249
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 9.483372921615203e-06,
+      "loss": 0.0272,
+      "step": 16250
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 9.477434679334918e-06,
+      "loss": 0.0105,
+      "step": 16251
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 9.471496437054631e-06,
+      "loss": 0.0171,
+      "step": 16252
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 9.465558194774347e-06,
+      "loss": 0.0149,
+      "step": 16253
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 9.459619952494062e-06,
+      "loss": 0.0116,
+      "step": 16254
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 9.453681710213777e-06,
+      "loss": 0.0291,
+      "step": 16255
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 9.447743467933493e-06,
+      "loss": 0.0119,
+      "step": 16256
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 9.441805225653206e-06,
+      "loss": 0.0354,
+      "step": 16257
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 9.435866983372921e-06,
+      "loss": 0.0211,
+      "step": 16258
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 9.429928741092638e-06,
+      "loss": 0.0464,
+      "step": 16259
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 9.423990498812352e-06,
+      "loss": 0.0145,
+      "step": 16260
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 9.418052256532067e-06,
+      "loss": 0.0551,
+      "step": 16261
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 9.41211401425178e-06,
+      "loss": 0.0421,
+      "step": 16262
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 9.406175771971496e-06,
+      "loss": 0.0705,
+      "step": 16263
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 9.400237529691213e-06,
+      "loss": 0.0239,
+      "step": 16264
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 9.394299287410927e-06,
+      "loss": 0.035,
+      "step": 16265
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 9.388361045130642e-06,
+      "loss": 0.0298,
+      "step": 16266
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 9.382422802850356e-06,
+      "loss": 0.0242,
+      "step": 16267
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 9.37648456057007e-06,
+      "loss": 0.0457,
+      "step": 16268
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 9.370546318289788e-06,
+      "loss": 0.0281,
+      "step": 16269
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 9.364608076009501e-06,
+      "loss": 0.0212,
+      "step": 16270
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 9.358669833729217e-06,
+      "loss": 0.0135,
+      "step": 16271
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 9.352731591448932e-06,
+      "loss": 0.0294,
+      "step": 16272
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 9.346793349168645e-06,
+      "loss": 0.0156,
+      "step": 16273
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 9.340855106888362e-06,
+      "loss": 0.0259,
+      "step": 16274
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 9.334916864608076e-06,
+      "loss": 0.0305,
+      "step": 16275
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 9.328978622327791e-06,
+      "loss": 0.0547,
+      "step": 16276
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 9.323040380047507e-06,
+      "loss": 0.0121,
+      "step": 16277
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 9.317102137767222e-06,
+      "loss": 0.0183,
+      "step": 16278
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 9.311163895486937e-06,
+      "loss": 0.0329,
+      "step": 16279
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 9.30522565320665e-06,
+      "loss": 0.0171,
+      "step": 16280
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 9.299287410926366e-06,
+      "loss": 0.0702,
+      "step": 16281
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 9.293349168646081e-06,
+      "loss": 0.0382,
+      "step": 16282
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 9.287410926365797e-06,
+      "loss": 0.0204,
+      "step": 16283
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 9.281472684085512e-06,
+      "loss": 0.0455,
+      "step": 16284
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 9.275534441805227e-06,
+      "loss": 0.018,
+      "step": 16285
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 9.26959619952494e-06,
+      "loss": 0.0224,
+      "step": 16286
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 9.263657957244656e-06,
+      "loss": 0.0187,
+      "step": 16287
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 9.257719714964371e-06,
+      "loss": 0.025,
+      "step": 16288
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 9.251781472684086e-06,
+      "loss": 0.0426,
+      "step": 16289
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 9.245843230403802e-06,
+      "loss": 0.0483,
+      "step": 16290
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 9.239904988123515e-06,
+      "loss": 0.065,
+      "step": 16291
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 9.23396674584323e-06,
+      "loss": 0.02,
+      "step": 16292
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 9.228028503562946e-06,
+      "loss": 0.0161,
+      "step": 16293
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 9.222090261282661e-06,
+      "loss": 0.0447,
+      "step": 16294
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 9.216152019002376e-06,
+      "loss": 0.0741,
+      "step": 16295
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 9.21021377672209e-06,
+      "loss": 0.0342,
+      "step": 16296
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 9.204275534441805e-06,
+      "loss": 0.0212,
+      "step": 16297
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 9.19833729216152e-06,
+      "loss": 0.0755,
+      "step": 16298
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 9.192399049881236e-06,
+      "loss": 0.0672,
+      "step": 16299
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 9.186460807600951e-06,
+      "loss": 0.0136,
+      "step": 16300
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 9.180522565320665e-06,
+      "loss": 0.0353,
+      "step": 16301
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 9.17458432304038e-06,
+      "loss": 0.0334,
+      "step": 16302
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 9.168646080760095e-06,
+      "loss": 0.0104,
+      "step": 16303
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 9.16270783847981e-06,
+      "loss": 0.0069,
+      "step": 16304
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 9.156769596199526e-06,
+      "loss": 0.0382,
+      "step": 16305
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 9.15083135391924e-06,
+      "loss": 0.0568,
+      "step": 16306
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 9.144893111638955e-06,
+      "loss": 0.0447,
+      "step": 16307
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 9.13895486935867e-06,
+      "loss": 0.0423,
+      "step": 16308
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 9.133016627078385e-06,
+      "loss": 0.0589,
+      "step": 16309
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 9.1270783847981e-06,
+      "loss": 0.0798,
+      "step": 16310
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 9.121140142517816e-06,
+      "loss": 0.0462,
+      "step": 16311
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 9.11520190023753e-06,
+      "loss": 0.0234,
+      "step": 16312
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 9.109263657957246e-06,
+      "loss": 0.0267,
+      "step": 16313
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 9.10332541567696e-06,
+      "loss": 0.0326,
+      "step": 16314
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 9.097387173396675e-06,
+      "loss": 0.0401,
+      "step": 16315
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 9.09144893111639e-06,
+      "loss": 0.0199,
+      "step": 16316
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 9.085510688836104e-06,
+      "loss": 0.0188,
+      "step": 16317
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 9.079572446555821e-06,
+      "loss": 0.0232,
+      "step": 16318
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 9.073634204275535e-06,
+      "loss": 0.0574,
+      "step": 16319
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 9.06769596199525e-06,
+      "loss": 0.0217,
+      "step": 16320
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 9.061757719714965e-06,
+      "loss": 0.0302,
+      "step": 16321
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 9.055819477434679e-06,
+      "loss": 0.051,
+      "step": 16322
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 9.049881235154396e-06,
+      "loss": 0.0143,
+      "step": 16323
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 9.04394299287411e-06,
+      "loss": 0.0418,
+      "step": 16324
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 9.038004750593825e-06,
+      "loss": 0.0281,
+      "step": 16325
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 9.03206650831354e-06,
+      "loss": 0.0288,
+      "step": 16326
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 9.026128266033253e-06,
+      "loss": 0.0216,
+      "step": 16327
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 9.02019002375297e-06,
+      "loss": 0.0288,
+      "step": 16328
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 9.014251781472686e-06,
+      "loss": 0.0318,
+      "step": 16329
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 9.0083135391924e-06,
+      "loss": 0.0392,
+      "step": 16330
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 9.002375296912114e-06,
+      "loss": 0.0191,
+      "step": 16331
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 8.996437054631828e-06,
+      "loss": 0.0233,
+      "step": 16332
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 8.990498812351545e-06,
+      "loss": 0.0386,
+      "step": 16333
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 8.98456057007126e-06,
+      "loss": 0.0358,
+      "step": 16334
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 8.978622327790974e-06,
+      "loss": 0.0218,
+      "step": 16335
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 8.972684085510689e-06,
+      "loss": 0.0269,
+      "step": 16336
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 8.966745843230404e-06,
+      "loss": 0.0361,
+      "step": 16337
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 8.96080760095012e-06,
+      "loss": 0.0139,
+      "step": 16338
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 8.954869358669835e-06,
+      "loss": 0.0285,
+      "step": 16339
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 8.948931116389549e-06,
+      "loss": 0.0117,
+      "step": 16340
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 8.942992874109264e-06,
+      "loss": 0.0183,
+      "step": 16341
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 8.937054631828979e-06,
+      "loss": 0.0196,
+      "step": 16342
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 8.931116389548694e-06,
+      "loss": 0.0402,
+      "step": 16343
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 8.92517814726841e-06,
+      "loss": 0.0113,
+      "step": 16344
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 8.919239904988123e-06,
+      "loss": 0.0125,
+      "step": 16345
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 8.913301662707838e-06,
+      "loss": 0.0147,
+      "step": 16346
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 8.907363420427554e-06,
+      "loss": 0.0351,
+      "step": 16347
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 8.901425178147269e-06,
+      "loss": 0.0647,
+      "step": 16348
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 8.895486935866984e-06,
+      "loss": 0.0285,
+      "step": 16349
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 8.889548693586698e-06,
+      "loss": 0.03,
+      "step": 16350
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 8.883610451306413e-06,
+      "loss": 0.0487,
+      "step": 16351
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 8.877672209026128e-06,
+      "loss": 0.0281,
+      "step": 16352
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 8.871733966745844e-06,
+      "loss": 0.0197,
+      "step": 16353
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 8.865795724465559e-06,
+      "loss": 0.0052,
+      "step": 16354
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 8.859857482185274e-06,
+      "loss": 0.027,
+      "step": 16355
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 8.853919239904988e-06,
+      "loss": 0.0536,
+      "step": 16356
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 8.847980997624703e-06,
+      "loss": 0.0453,
+      "step": 16357
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 8.842042755344418e-06,
+      "loss": 0.0255,
+      "step": 16358
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 8.836104513064134e-06,
+      "loss": 0.0438,
+      "step": 16359
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 8.830166270783849e-06,
+      "loss": 0.0261,
+      "step": 16360
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 8.824228028503563e-06,
+      "loss": 0.0279,
+      "step": 16361
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 8.818289786223278e-06,
+      "loss": 0.0379,
+      "step": 16362
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 8.812351543942993e-06,
+      "loss": 0.0305,
+      "step": 16363
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 8.806413301662708e-06,
+      "loss": 0.0544,
+      "step": 16364
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 8.800475059382424e-06,
+      "loss": 0.0236,
+      "step": 16365
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 8.794536817102137e-06,
+      "loss": 0.046,
+      "step": 16366
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 8.788598574821852e-06,
+      "loss": 0.0419,
+      "step": 16367
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 8.78266033254157e-06,
+      "loss": 0.0594,
+      "step": 16368
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 8.776722090261283e-06,
+      "loss": 0.0227,
+      "step": 16369
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 8.770783847980998e-06,
+      "loss": 0.0154,
+      "step": 16370
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 8.764845605700712e-06,
+      "loss": 0.0248,
+      "step": 16371
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 8.758907363420427e-06,
+      "loss": 0.0348,
+      "step": 16372
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 8.752969121140144e-06,
+      "loss": 0.0293,
+      "step": 16373
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 8.747030878859858e-06,
+      "loss": 0.0262,
+      "step": 16374
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 8.741092636579573e-06,
+      "loss": 0.0243,
+      "step": 16375
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 8.735154394299287e-06,
+      "loss": 0.024,
+      "step": 16376
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 8.729216152019004e-06,
+      "loss": 0.0269,
+      "step": 16377
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 8.723277909738719e-06,
+      "loss": 0.0276,
+      "step": 16378
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 8.717339667458432e-06,
+      "loss": 0.0317,
+      "step": 16379
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 8.711401425178148e-06,
+      "loss": 0.028,
+      "step": 16380
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 8.705463182897863e-06,
+      "loss": 0.0242,
+      "step": 16381
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 8.699524940617578e-06,
+      "loss": 0.0248,
+      "step": 16382
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 8.693586698337293e-06,
+      "loss": 0.0423,
+      "step": 16383
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 8.687648456057007e-06,
+      "loss": 0.0281,
+      "step": 16384
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 8.681710213776722e-06,
+      "loss": 0.0214,
+      "step": 16385
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 8.675771971496438e-06,
+      "loss": 0.0348,
+      "step": 16386
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 8.669833729216153e-06,
+      "loss": 0.0212,
+      "step": 16387
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 8.663895486935868e-06,
+      "loss": 0.0414,
+      "step": 16388
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 8.657957244655582e-06,
+      "loss": 0.0382,
+      "step": 16389
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 8.652019002375297e-06,
+      "loss": 0.0343,
+      "step": 16390
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 8.646080760095012e-06,
+      "loss": 0.0301,
+      "step": 16391
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 8.640142517814728e-06,
+      "loss": 0.024,
+      "step": 16392
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 8.634204275534443e-06,
+      "loss": 0.0362,
+      "step": 16393
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 8.628266033254156e-06,
+      "loss": 0.0265,
+      "step": 16394
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 8.622327790973872e-06,
+      "loss": 0.009,
+      "step": 16395
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 8.616389548693587e-06,
+      "loss": 0.0499,
+      "step": 16396
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 8.610451306413302e-06,
+      "loss": 0.0257,
+      "step": 16397
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 8.604513064133018e-06,
+      "loss": 0.0101,
+      "step": 16398
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 8.598574821852733e-06,
+      "loss": 0.0083,
+      "step": 16399
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 8.592636579572446e-06,
+      "loss": 0.0244,
+      "step": 16400
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 8.586698337292162e-06,
+      "loss": 0.0457,
+      "step": 16401
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 8.580760095011877e-06,
+      "loss": 0.0226,
+      "step": 16402
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 8.574821852731592e-06,
+      "loss": 0.0127,
+      "step": 16403
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 8.568883610451307e-06,
+      "loss": 0.0515,
+      "step": 16404
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 8.562945368171021e-06,
+      "loss": 0.0324,
+      "step": 16405
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 8.557007125890736e-06,
+      "loss": 0.055,
+      "step": 16406
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 8.551068883610452e-06,
+      "loss": 0.0405,
+      "step": 16407
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 8.545130641330167e-06,
+      "loss": 0.0277,
+      "step": 16408
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 8.539192399049882e-06,
+      "loss": 0.0395,
+      "step": 16409
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 8.533254156769596e-06,
+      "loss": 0.0262,
+      "step": 16410
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 8.527315914489311e-06,
+      "loss": 0.0446,
+      "step": 16411
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 8.521377672209028e-06,
+      "loss": 0.018,
+      "step": 16412
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 8.515439429928742e-06,
+      "loss": 0.0365,
+      "step": 16413
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 8.509501187648457e-06,
+      "loss": 0.0602,
+      "step": 16414
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 8.50356294536817e-06,
+      "loss": 0.0244,
+      "step": 16415
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 8.497624703087886e-06,
+      "loss": 0.0203,
+      "step": 16416
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 8.491686460807603e-06,
+      "loss": 0.0225,
+      "step": 16417
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 8.485748218527316e-06,
+      "loss": 0.0489,
+      "step": 16418
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 8.479809976247032e-06,
+      "loss": 0.0574,
+      "step": 16419
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 8.473871733966745e-06,
+      "loss": 0.0451,
+      "step": 16420
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 8.46793349168646e-06,
+      "loss": 0.0213,
+      "step": 16421
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 8.461995249406177e-06,
+      "loss": 0.0401,
+      "step": 16422
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 8.456057007125891e-06,
+      "loss": 0.0192,
+      "step": 16423
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 8.450118764845606e-06,
+      "loss": 0.0194,
+      "step": 16424
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 8.444180522565321e-06,
+      "loss": 0.032,
+      "step": 16425
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 8.438242280285035e-06,
+      "loss": 0.0236,
+      "step": 16426
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 8.432304038004752e-06,
+      "loss": 0.0227,
+      "step": 16427
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 8.426365795724466e-06,
+      "loss": 0.0577,
+      "step": 16428
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 8.420427553444181e-06,
+      "loss": 0.0392,
+      "step": 16429
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 8.414489311163896e-06,
+      "loss": 0.0303,
+      "step": 16430
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 8.40855106888361e-06,
+      "loss": 0.0384,
+      "step": 16431
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 8.402612826603327e-06,
+      "loss": 0.0205,
+      "step": 16432
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 8.39667458432304e-06,
+      "loss": 0.0152,
+      "step": 16433
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 8.390736342042756e-06,
+      "loss": 0.0382,
+      "step": 16434
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 8.38479809976247e-06,
+      "loss": 0.0244,
+      "step": 16435
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 8.378859857482186e-06,
+      "loss": 0.0252,
+      "step": 16436
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 8.372921615201901e-06,
+      "loss": 0.0304,
+      "step": 16437
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 8.366983372921617e-06,
+      "loss": 0.028,
+      "step": 16438
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 8.36104513064133e-06,
+      "loss": 0.0277,
+      "step": 16439
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 8.355106888361046e-06,
+      "loss": 0.0308,
+      "step": 16440
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 8.34916864608076e-06,
+      "loss": 0.0222,
+      "step": 16441
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 8.343230403800476e-06,
+      "loss": 0.0426,
+      "step": 16442
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 8.337292161520191e-06,
+      "loss": 0.0247,
+      "step": 16443
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 8.331353919239905e-06,
+      "loss": 0.0411,
+      "step": 16444
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 8.32541567695962e-06,
+      "loss": 0.0517,
+      "step": 16445
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 8.319477434679335e-06,
+      "loss": 0.0189,
+      "step": 16446
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 8.31353919239905e-06,
+      "loss": 0.0131,
+      "step": 16447
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 8.307600950118766e-06,
+      "loss": 0.0271,
+      "step": 16448
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 8.30166270783848e-06,
+      "loss": 0.0666,
+      "step": 16449
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 8.295724465558195e-06,
+      "loss": 0.0221,
+      "step": 16450
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 8.28978622327791e-06,
+      "loss": 0.0185,
+      "step": 16451
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 8.283847980997625e-06,
+      "loss": 0.0295,
+      "step": 16452
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 8.27790973871734e-06,
+      "loss": 0.0208,
+      "step": 16453
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 8.271971496437054e-06,
+      "loss": 0.0318,
+      "step": 16454
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 8.26603325415677e-06,
+      "loss": 0.0527,
+      "step": 16455
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 8.260095011876485e-06,
+      "loss": 0.0226,
+      "step": 16456
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 8.2541567695962e-06,
+      "loss": 0.0573,
+      "step": 16457
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 8.248218527315915e-06,
+      "loss": 0.045,
+      "step": 16458
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 8.242280285035629e-06,
+      "loss": 0.0256,
+      "step": 16459
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 8.236342042755344e-06,
+      "loss": 0.0419,
+      "step": 16460
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 8.23040380047506e-06,
+      "loss": 0.0759,
+      "step": 16461
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 8.224465558194775e-06,
+      "loss": 0.0405,
+      "step": 16462
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 8.21852731591449e-06,
+      "loss": 0.0207,
+      "step": 16463
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 8.212589073634204e-06,
+      "loss": 0.0487,
+      "step": 16464
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 8.206650831353919e-06,
+      "loss": 0.0201,
+      "step": 16465
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 8.200712589073634e-06,
+      "loss": 0.0223,
+      "step": 16466
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 8.19477434679335e-06,
+      "loss": 0.0198,
+      "step": 16467
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 8.188836104513065e-06,
+      "loss": 0.0545,
+      "step": 16468
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 8.18289786223278e-06,
+      "loss": 0.0207,
+      "step": 16469
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 8.176959619952494e-06,
+      "loss": 0.021,
+      "step": 16470
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 8.171021377672209e-06,
+      "loss": 0.021,
+      "step": 16471
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 8.165083135391924e-06,
+      "loss": 0.0159,
+      "step": 16472
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 8.15914489311164e-06,
+      "loss": 0.0352,
+      "step": 16473
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 8.153206650831355e-06,
+      "loss": 0.0252,
+      "step": 16474
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 8.147268408551068e-06,
+      "loss": 0.0226,
+      "step": 16475
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 8.141330166270785e-06,
+      "loss": 0.0286,
+      "step": 16476
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 8.135391923990499e-06,
+      "loss": 0.0206,
+      "step": 16477
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 8.129453681710214e-06,
+      "loss": 0.0194,
+      "step": 16478
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 8.12351543942993e-06,
+      "loss": 0.0249,
+      "step": 16479
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 8.117577197149643e-06,
+      "loss": 0.0154,
+      "step": 16480
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 8.11163895486936e-06,
+      "loss": 0.0301,
+      "step": 16481
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 8.105700712589075e-06,
+      "loss": 0.0294,
+      "step": 16482
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 8.099762470308789e-06,
+      "loss": 0.0319,
+      "step": 16483
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 8.093824228028504e-06,
+      "loss": 0.0148,
+      "step": 16484
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 8.087885985748218e-06,
+      "loss": 0.0403,
+      "step": 16485
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 8.081947743467935e-06,
+      "loss": 0.0523,
+      "step": 16486
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 8.07600950118765e-06,
+      "loss": 0.0181,
+      "step": 16487
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 8.070071258907363e-06,
+      "loss": 0.0265,
+      "step": 16488
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 8.064133016627079e-06,
+      "loss": 0.0223,
+      "step": 16489
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 8.058194774346792e-06,
+      "loss": 0.026,
+      "step": 16490
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 8.05225653206651e-06,
+      "loss": 0.0186,
+      "step": 16491
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 8.046318289786225e-06,
+      "loss": 0.0306,
+      "step": 16492
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 8.040380047505938e-06,
+      "loss": 0.0365,
+      "step": 16493
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 8.034441805225653e-06,
+      "loss": 0.0275,
+      "step": 16494
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 8.028503562945369e-06,
+      "loss": 0.0384,
+      "step": 16495
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 8.022565320665084e-06,
+      "loss": 0.0445,
+      "step": 16496
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 8.0166270783848e-06,
+      "loss": 0.0509,
+      "step": 16497
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 8.010688836104513e-06,
+      "loss": 0.0349,
+      "step": 16498
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 8.004750593824228e-06,
+      "loss": 0.023,
+      "step": 16499
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 7.998812351543943e-06,
+      "loss": 0.0333,
+      "step": 16500
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 7.992874109263659e-06,
+      "loss": 0.0168,
+      "step": 16501
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 7.986935866983374e-06,
+      "loss": 0.0659,
+      "step": 16502
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 7.980997624703088e-06,
+      "loss": 0.0423,
+      "step": 16503
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 7.975059382422803e-06,
+      "loss": 0.0091,
+      "step": 16504
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 7.969121140142518e-06,
+      "loss": 0.0325,
+      "step": 16505
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 7.963182897862233e-06,
+      "loss": 0.0405,
+      "step": 16506
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 7.957244655581949e-06,
+      "loss": 0.0584,
+      "step": 16507
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 7.951306413301664e-06,
+      "loss": 0.0493,
+      "step": 16508
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 7.945368171021377e-06,
+      "loss": 0.0248,
+      "step": 16509
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 7.939429928741093e-06,
+      "loss": 0.0478,
+      "step": 16510
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 7.933491686460808e-06,
+      "loss": 0.0574,
+      "step": 16511
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 7.927553444180523e-06,
+      "loss": 0.0382,
+      "step": 16512
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 7.921615201900239e-06,
+      "loss": 0.0258,
+      "step": 16513
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 7.915676959619952e-06,
+      "loss": 0.0496,
+      "step": 16514
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 7.909738717339667e-06,
+      "loss": 0.0318,
+      "step": 16515
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 7.903800475059383e-06,
+      "loss": 0.0276,
+      "step": 16516
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 7.897862232779098e-06,
+      "loss": 0.0206,
+      "step": 16517
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 7.891923990498813e-06,
+      "loss": 0.0234,
+      "step": 16518
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 7.885985748218527e-06,
+      "loss": 0.0538,
+      "step": 16519
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 7.880047505938242e-06,
+      "loss": 0.0267,
+      "step": 16520
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 7.874109263657959e-06,
+      "loss": 0.0234,
+      "step": 16521
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 7.868171021377673e-06,
+      "loss": 0.0377,
+      "step": 16522
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 7.862232779097388e-06,
+      "loss": 0.0336,
+      "step": 16523
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 7.856294536817101e-06,
+      "loss": 0.0303,
+      "step": 16524
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 7.850356294536817e-06,
+      "loss": 0.03,
+      "step": 16525
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 7.844418052256534e-06,
+      "loss": 0.0288,
+      "step": 16526
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 7.838479809976247e-06,
+      "loss": 0.0203,
+      "step": 16527
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 7.832541567695963e-06,
+      "loss": 0.0345,
+      "step": 16528
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 7.826603325415676e-06,
+      "loss": 0.0228,
+      "step": 16529
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 7.820665083135391e-06,
+      "loss": 0.0197,
+      "step": 16530
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 7.814726840855108e-06,
+      "loss": 0.0222,
+      "step": 16531
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 7.808788598574822e-06,
+      "loss": 0.0353,
+      "step": 16532
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 7.802850356294537e-06,
+      "loss": 0.0328,
+      "step": 16533
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 7.796912114014251e-06,
+      "loss": 0.0191,
+      "step": 16534
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 7.790973871733968e-06,
+      "loss": 0.0201,
+      "step": 16535
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 7.785035629453683e-06,
+      "loss": 0.014,
+      "step": 16536
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 7.779097387173397e-06,
+      "loss": 0.0293,
+      "step": 16537
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 7.773159144893112e-06,
+      "loss": 0.0376,
+      "step": 16538
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 7.767220902612827e-06,
+      "loss": 0.0137,
+      "step": 16539
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 7.761282660332543e-06,
+      "loss": 0.0218,
+      "step": 16540
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 7.755344418052258e-06,
+      "loss": 0.0285,
+      "step": 16541
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 7.749406175771971e-06,
+      "loss": 0.0241,
+      "step": 16542
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 7.743467933491687e-06,
+      "loss": 0.0335,
+      "step": 16543
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 7.737529691211402e-06,
+      "loss": 0.0264,
+      "step": 16544
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 7.731591448931117e-06,
+      "loss": 0.0378,
+      "step": 16545
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 7.725653206650832e-06,
+      "loss": 0.0177,
+      "step": 16546
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 7.719714964370546e-06,
+      "loss": 0.0587,
+      "step": 16547
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 7.713776722090261e-06,
+      "loss": 0.019,
+      "step": 16548
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 7.707838479809977e-06,
+      "loss": 0.0287,
+      "step": 16549
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 7.701900237529692e-06,
+      "loss": 0.0268,
+      "step": 16550
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 7.695961995249407e-06,
+      "loss": 0.0285,
+      "step": 16551
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 7.690023752969122e-06,
+      "loss": 0.0034,
+      "step": 16552
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 7.684085510688836e-06,
+      "loss": 0.0221,
+      "step": 16553
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 7.678147268408551e-06,
+      "loss": 0.0296,
+      "step": 16554
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 7.672209026128267e-06,
+      "loss": 0.0211,
+      "step": 16555
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 7.666270783847982e-06,
+      "loss": 0.0961,
+      "step": 16556
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 7.660332541567697e-06,
+      "loss": 0.0556,
+      "step": 16557
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 7.65439429928741e-06,
+      "loss": 0.0594,
+      "step": 16558
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 7.648456057007126e-06,
+      "loss": 0.0637,
+      "step": 16559
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 7.642517814726841e-06,
+      "loss": 0.0276,
+      "step": 16560
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 7.636579572446556e-06,
+      "loss": 0.0319,
+      "step": 16561
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 7.630641330166272e-06,
+      "loss": 0.0266,
+      "step": 16562
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 7.624703087885985e-06,
+      "loss": 0.0274,
+      "step": 16563
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 7.6187648456057015e-06,
+      "loss": 0.0376,
+      "step": 16564
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 7.612826603325417e-06,
+      "loss": 0.0483,
+      "step": 16565
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 7.606888361045131e-06,
+      "loss": 0.0395,
+      "step": 16566
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 7.6009501187648464e-06,
+      "loss": 0.0321,
+      "step": 16567
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 7.59501187648456e-06,
+      "loss": 0.0356,
+      "step": 16568
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 7.589073634204276e-06,
+      "loss": 0.0272,
+      "step": 16569
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 7.5831353919239914e-06,
+      "loss": 0.0297,
+      "step": 16570
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 7.577197149643706e-06,
+      "loss": 0.0287,
+      "step": 16571
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 7.571258907363421e-06,
+      "loss": 0.0186,
+      "step": 16572
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 7.5653206650831356e-06,
+      "loss": 0.0285,
+      "step": 16573
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 7.559382422802851e-06,
+      "loss": 0.0138,
+      "step": 16574
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 7.553444180522566e-06,
+      "loss": 0.0284,
+      "step": 16575
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 7.5475059382422805e-06,
+      "loss": 0.03,
+      "step": 16576
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 7.541567695961996e-06,
+      "loss": 0.0275,
+      "step": 16577
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 7.535629453681711e-06,
+      "loss": 0.031,
+      "step": 16578
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 7.5296912114014255e-06,
+      "loss": 0.0316,
+      "step": 16579
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 7.523752969121141e-06,
+      "loss": 0.023,
+      "step": 16580
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 7.517814726840855e-06,
+      "loss": 0.0183,
+      "step": 16581
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 7.5118764845605705e-06,
+      "loss": 0.0271,
+      "step": 16582
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 7.505938242280286e-06,
+      "loss": 0.0169,
+      "step": 16583
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 7.5e-06,
+      "loss": 0.0281,
+      "step": 16584
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 7.4940617577197155e-06,
+      "loss": 0.0264,
+      "step": 16585
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 7.48812351543943e-06,
+      "loss": 0.0303,
+      "step": 16586
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 7.482185273159145e-06,
+      "loss": 0.0164,
+      "step": 16587
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 7.4762470308788604e-06,
+      "loss": 0.0451,
+      "step": 16588
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 7.470308788598575e-06,
+      "loss": 0.0152,
+      "step": 16589
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 7.46437054631829e-06,
+      "loss": 0.03,
+      "step": 16590
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 7.4584323040380046e-06,
+      "loss": 0.0212,
+      "step": 16591
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 7.45249406175772e-06,
+      "loss": 0.0348,
+      "step": 16592
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 7.446555819477435e-06,
+      "loss": 0.0159,
+      "step": 16593
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 7.4406175771971495e-06,
+      "loss": 0.0179,
+      "step": 16594
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 7.434679334916865e-06,
+      "loss": 0.0206,
+      "step": 16595
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 7.428741092636581e-06,
+      "loss": 0.0194,
+      "step": 16596
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 7.4228028503562945e-06,
+      "loss": 0.0433,
+      "step": 16597
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 7.41686460807601e-06,
+      "loss": 0.0429,
+      "step": 16598
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 7.410926365795724e-06,
+      "loss": 0.0296,
+      "step": 16599
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 7.4049881235154395e-06,
+      "loss": 0.0081,
+      "step": 16600
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 7.399049881235156e-06,
+      "loss": 0.0142,
+      "step": 16601
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 7.393111638954869e-06,
+      "loss": 0.0322,
+      "step": 16602
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 7.3871733966745845e-06,
+      "loss": 0.0309,
+      "step": 16603
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 7.381235154394299e-06,
+      "loss": 0.0157,
+      "step": 16604
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 7.375296912114014e-06,
+      "loss": 0.022,
+      "step": 16605
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 7.36935866983373e-06,
+      "loss": 0.0931,
+      "step": 16606
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 7.363420427553444e-06,
+      "loss": 0.0339,
+      "step": 16607
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 7.357482185273159e-06,
+      "loss": 0.0276,
+      "step": 16608
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 7.351543942992875e-06,
+      "loss": 0.0359,
+      "step": 16609
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 7.345605700712589e-06,
+      "loss": 0.0321,
+      "step": 16610
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 7.339667458432305e-06,
+      "loss": 0.0442,
+      "step": 16611
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 7.3337292161520186e-06,
+      "loss": 0.0183,
+      "step": 16612
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 7.327790973871735e-06,
+      "loss": 0.0281,
+      "step": 16613
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 7.32185273159145e-06,
+      "loss": 0.042,
+      "step": 16614
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 7.3159144893111635e-06,
+      "loss": 0.0529,
+      "step": 16615
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 7.30997624703088e-06,
+      "loss": 0.02,
+      "step": 16616
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 7.304038004750593e-06,
+      "loss": 0.0428,
+      "step": 16617
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 7.298099762470309e-06,
+      "loss": 0.0386,
+      "step": 16618
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 7.292161520190025e-06,
+      "loss": 0.0247,
+      "step": 16619
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 7.286223277909738e-06,
+      "loss": 0.0234,
+      "step": 16620
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 7.280285035629454e-06,
+      "loss": 0.0436,
+      "step": 16621
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 7.27434679334917e-06,
+      "loss": 0.0311,
+      "step": 16622
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 7.268408551068884e-06,
+      "loss": 0.0375,
+      "step": 16623
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 7.262470308788599e-06,
+      "loss": 0.0466,
+      "step": 16624
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 7.256532066508314e-06,
+      "loss": 0.0363,
+      "step": 16625
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 7.250593824228029e-06,
+      "loss": 0.0237,
+      "step": 16626
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 7.244655581947744e-06,
+      "loss": 0.0165,
+      "step": 16627
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 7.238717339667459e-06,
+      "loss": 0.0328,
+      "step": 16628
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 7.232779097387174e-06,
+      "loss": 0.0437,
+      "step": 16629
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 7.2268408551068884e-06,
+      "loss": 0.026,
+      "step": 16630
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 7.220902612826604e-06,
+      "loss": 0.0247,
+      "step": 16631
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 7.214964370546319e-06,
+      "loss": 0.0319,
+      "step": 16632
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 7.209026128266033e-06,
+      "loss": 0.0229,
+      "step": 16633
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 7.203087885985749e-06,
+      "loss": 0.0366,
+      "step": 16634
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 7.197149643705464e-06,
+      "loss": 0.0189,
+      "step": 16635
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 7.191211401425178e-06,
+      "loss": 0.009,
+      "step": 16636
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 7.185273159144894e-06,
+      "loss": 0.0291,
+      "step": 16637
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 7.179334916864608e-06,
+      "loss": 0.0109,
+      "step": 16638
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 7.173396674584323e-06,
+      "loss": 0.0392,
+      "step": 16639
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 7.167458432304039e-06,
+      "loss": 0.0275,
+      "step": 16640
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 7.161520190023753e-06,
+      "loss": 0.0282,
+      "step": 16641
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 7.155581947743468e-06,
+      "loss": 0.0156,
+      "step": 16642
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 7.149643705463183e-06,
+      "loss": 0.0185,
+      "step": 16643
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 7.143705463182898e-06,
+      "loss": 0.0109,
+      "step": 16644
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 7.137767220902613e-06,
+      "loss": 0.0068,
+      "step": 16645
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 7.131828978622328e-06,
+      "loss": 0.0236,
+      "step": 16646
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 7.125890736342043e-06,
+      "loss": 0.0381,
+      "step": 16647
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 7.119952494061759e-06,
+      "loss": 0.0173,
+      "step": 16648
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 7.114014251781473e-06,
+      "loss": 0.0222,
+      "step": 16649
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 7.108076009501188e-06,
+      "loss": 0.0395,
+      "step": 16650
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 7.102137767220902e-06,
+      "loss": 0.0178,
+      "step": 16651
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 7.096199524940618e-06,
+      "loss": 0.0087,
+      "step": 16652
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 7.090261282660334e-06,
+      "loss": 0.0319,
+      "step": 16653
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 7.084323040380047e-06,
+      "loss": 0.0141,
+      "step": 16654
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 7.078384798099763e-06,
+      "loss": 0.0397,
+      "step": 16655
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 7.072446555819477e-06,
+      "loss": 0.1724,
+      "step": 16656
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 7.066508313539192e-06,
+      "loss": 0.0433,
+      "step": 16657
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 7.0605700712589085e-06,
+      "loss": 0.0277,
+      "step": 16658
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 7.054631828978622e-06,
+      "loss": 0.029,
+      "step": 16659
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 7.048693586698338e-06,
+      "loss": 0.0534,
+      "step": 16660
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 7.042755344418052e-06,
+      "loss": 0.0418,
+      "step": 16661
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 7.036817102137767e-06,
+      "loss": 0.0267,
+      "step": 16662
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 7.030878859857483e-06,
+      "loss": 0.028,
+      "step": 16663
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 7.024940617577197e-06,
+      "loss": 0.038,
+      "step": 16664
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 7.019002375296913e-06,
+      "loss": 0.0439,
+      "step": 16665
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 7.013064133016628e-06,
+      "loss": 0.0389,
+      "step": 16666
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 7.007125890736342e-06,
+      "loss": 0.0312,
+      "step": 16667
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 7.001187648456058e-06,
+      "loss": 0.0603,
+      "step": 16668
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 6.9952494061757714e-06,
+      "loss": 0.0148,
+      "step": 16669
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 6.9893111638954876e-06,
+      "loss": 0.0477,
+      "step": 16670
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 6.983372921615203e-06,
+      "loss": 0.0337,
+      "step": 16671
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 6.977434679334917e-06,
+      "loss": 0.0231,
+      "step": 16672
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 6.9714964370546325e-06,
+      "loss": 0.0477,
+      "step": 16673
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 6.965558194774346e-06,
+      "loss": 0.0191,
+      "step": 16674
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 6.959619952494062e-06,
+      "loss": 0.0256,
+      "step": 16675
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 6.9536817102137775e-06,
+      "loss": 0.0307,
+      "step": 16676
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 6.947743467933492e-06,
+      "loss": 0.0597,
+      "step": 16677
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 6.941805225653207e-06,
+      "loss": 0.0324,
+      "step": 16678
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 6.9358669833729225e-06,
+      "loss": 0.045,
+      "step": 16679
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 6.929928741092637e-06,
+      "loss": 0.0353,
+      "step": 16680
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 6.923990498812352e-06,
+      "loss": 0.0164,
+      "step": 16681
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 6.918052256532067e-06,
+      "loss": 0.0231,
+      "step": 16682
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 6.912114014251782e-06,
+      "loss": 0.0479,
+      "step": 16683
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 6.906175771971497e-06,
+      "loss": 0.0168,
+      "step": 16684
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 6.900237529691212e-06,
+      "loss": 0.0333,
+      "step": 16685
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 6.894299287410927e-06,
+      "loss": 0.0249,
+      "step": 16686
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 6.888361045130641e-06,
+      "loss": 0.0331,
+      "step": 16687
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 6.8824228028503566e-06,
+      "loss": 0.0281,
+      "step": 16688
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 6.876484560570072e-06,
+      "loss": 0.0081,
+      "step": 16689
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 6.870546318289786e-06,
+      "loss": 0.0305,
+      "step": 16690
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 6.8646080760095016e-06,
+      "loss": 0.0412,
+      "step": 16691
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 6.858669833729217e-06,
+      "loss": 0.0506,
+      "step": 16692
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 6.852731591448931e-06,
+      "loss": 0.0219,
+      "step": 16693
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 6.8467933491686465e-06,
+      "loss": 0.021,
+      "step": 16694
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 6.840855106888361e-06,
+      "loss": 0.0296,
+      "step": 16695
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 6.834916864608076e-06,
+      "loss": 0.0311,
+      "step": 16696
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 6.8289786223277915e-06,
+      "loss": 0.0146,
+      "step": 16697
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 6.823040380047506e-06,
+      "loss": 0.0213,
+      "step": 16698
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 6.817102137767221e-06,
+      "loss": 0.0218,
+      "step": 16699
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 6.811163895486936e-06,
+      "loss": 0.0176,
+      "step": 16700
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 6.805225653206651e-06,
+      "loss": 0.0465,
+      "step": 16701
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 6.799287410926366e-06,
+      "loss": 0.0242,
+      "step": 16702
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 6.793349168646081e-06,
+      "loss": 0.0263,
+      "step": 16703
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 6.787410926365796e-06,
+      "loss": 0.0144,
+      "step": 16704
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 6.781472684085512e-06,
+      "loss": 0.0399,
+      "step": 16705
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 6.775534441805226e-06,
+      "loss": 0.031,
+      "step": 16706
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 6.769596199524941e-06,
+      "loss": 0.0249,
+      "step": 16707
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 6.763657957244655e-06,
+      "loss": 0.0375,
+      "step": 16708
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 6.7577197149643706e-06,
+      "loss": 0.016,
+      "step": 16709
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 6.751781472684087e-06,
+      "loss": 0.0386,
+      "step": 16710
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 6.7458432304038e-06,
+      "loss": 0.0355,
+      "step": 16711
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 6.739904988123516e-06,
+      "loss": 0.0416,
+      "step": 16712
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 6.73396674584323e-06,
+      "loss": 0.0267,
+      "step": 16713
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 6.728028503562945e-06,
+      "loss": 0.0482,
+      "step": 16714
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 6.722090261282661e-06,
+      "loss": 0.0427,
+      "step": 16715
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 6.716152019002375e-06,
+      "loss": 0.0368,
+      "step": 16716
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 6.710213776722091e-06,
+      "loss": 0.0488,
+      "step": 16717
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 6.704275534441806e-06,
+      "loss": 0.0334,
+      "step": 16718
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 6.69833729216152e-06,
+      "loss": 0.0444,
+      "step": 16719
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 6.692399049881236e-06,
+      "loss": 0.0244,
+      "step": 16720
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 6.68646080760095e-06,
+      "loss": 0.0194,
+      "step": 16721
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 6.680522565320666e-06,
+      "loss": 0.0577,
+      "step": 16722
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 6.674584323040381e-06,
+      "loss": 0.0233,
+      "step": 16723
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 6.6686460807600955e-06,
+      "loss": 0.0231,
+      "step": 16724
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 6.662707838479811e-06,
+      "loss": 0.0125,
+      "step": 16725
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 6.656769596199524e-06,
+      "loss": 0.0347,
+      "step": 16726
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 6.6508313539192404e-06,
+      "loss": 0.0187,
+      "step": 16727
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 6.644893111638956e-06,
+      "loss": 0.0361,
+      "step": 16728
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 6.63895486935867e-06,
+      "loss": 0.0461,
+      "step": 16729
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 6.633016627078385e-06,
+      "loss": 0.0411,
+      "step": 16730
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 6.627078384798099e-06,
+      "loss": 0.0453,
+      "step": 16731
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 6.621140142517815e-06,
+      "loss": 0.02,
+      "step": 16732
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 6.61520190023753e-06,
+      "loss": 0.0209,
+      "step": 16733
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 6.609263657957245e-06,
+      "loss": 0.0155,
+      "step": 16734
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 6.60332541567696e-06,
+      "loss": 0.0186,
+      "step": 16735
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 6.597387173396675e-06,
+      "loss": 0.0289,
+      "step": 16736
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 6.59144893111639e-06,
+      "loss": 0.0382,
+      "step": 16737
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 6.585510688836105e-06,
+      "loss": 0.0265,
+      "step": 16738
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 6.5795724465558195e-06,
+      "loss": 0.0353,
+      "step": 16739
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 6.573634204275535e-06,
+      "loss": 0.02,
+      "step": 16740
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 6.56769596199525e-06,
+      "loss": 0.0293,
+      "step": 16741
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 6.5617577197149645e-06,
+      "loss": 0.0344,
+      "step": 16742
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 6.55581947743468e-06,
+      "loss": 0.0356,
+      "step": 16743
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 6.549881235154394e-06,
+      "loss": 0.0346,
+      "step": 16744
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 6.5439429928741094e-06,
+      "loss": 0.0144,
+      "step": 16745
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 6.538004750593825e-06,
+      "loss": 0.0106,
+      "step": 16746
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 6.532066508313539e-06,
+      "loss": 0.0255,
+      "step": 16747
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 6.5261282660332544e-06,
+      "loss": 0.0365,
+      "step": 16748
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 6.52019002375297e-06,
+      "loss": 0.0447,
+      "step": 16749
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 6.514251781472684e-06,
+      "loss": 0.0206,
+      "step": 16750
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 6.508313539192399e-06,
+      "loss": 0.0121,
+      "step": 16751
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 6.502375296912114e-06,
+      "loss": 0.0177,
+      "step": 16752
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 6.496437054631829e-06,
+      "loss": 0.0078,
+      "step": 16753
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 6.490498812351544e-06,
+      "loss": 0.0227,
+      "step": 16754
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 6.484560570071259e-06,
+      "loss": 0.0537,
+      "step": 16755
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 6.478622327790974e-06,
+      "loss": 0.0477,
+      "step": 16756
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 6.4726840855106885e-06,
+      "loss": 0.0448,
+      "step": 16757
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 6.466745843230404e-06,
+      "loss": 0.0484,
+      "step": 16758
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 6.46080760095012e-06,
+      "loss": 0.0312,
+      "step": 16759
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 6.4548693586698335e-06,
+      "loss": 0.0475,
+      "step": 16760
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 6.448931116389549e-06,
+      "loss": 0.0237,
+      "step": 16761
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 6.442992874109265e-06,
+      "loss": 0.0286,
+      "step": 16762
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 6.4370546318289785e-06,
+      "loss": 0.0493,
+      "step": 16763
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 6.431116389548695e-06,
+      "loss": 0.0302,
+      "step": 16764
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 6.425178147268408e-06,
+      "loss": 0.0444,
+      "step": 16765
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 6.4192399049881234e-06,
+      "loss": 0.0659,
+      "step": 16766
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 6.4133016627078396e-06,
+      "loss": 0.0414,
+      "step": 16767
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 6.407363420427553e-06,
+      "loss": 0.0613,
+      "step": 16768
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 6.401425178147269e-06,
+      "loss": 0.0176,
+      "step": 16769
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 6.395486935866983e-06,
+      "loss": 0.0116,
+      "step": 16770
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 6.389548693586699e-06,
+      "loss": 0.0363,
+      "step": 16771
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 6.383610451306414e-06,
+      "loss": 0.034,
+      "step": 16772
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 6.377672209026128e-06,
+      "loss": 0.0395,
+      "step": 16773
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 6.371733966745844e-06,
+      "loss": 0.0309,
+      "step": 16774
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 6.365795724465559e-06,
+      "loss": 0.0186,
+      "step": 16775
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 6.359857482185274e-06,
+      "loss": 0.0193,
+      "step": 16776
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 6.353919239904989e-06,
+      "loss": 0.0366,
+      "step": 16777
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 6.3479809976247025e-06,
+      "loss": 0.087,
+      "step": 16778
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 6.342042755344419e-06,
+      "loss": 0.015,
+      "step": 16779
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 6.336104513064134e-06,
+      "loss": 0.036,
+      "step": 16780
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 6.330166270783848e-06,
+      "loss": 0.0203,
+      "step": 16781
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 6.324228028503564e-06,
+      "loss": 0.0205,
+      "step": 16782
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 6.318289786223278e-06,
+      "loss": 0.031,
+      "step": 16783
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 6.312351543942993e-06,
+      "loss": 0.0122,
+      "step": 16784
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 6.306413301662709e-06,
+      "loss": 0.0329,
+      "step": 16785
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 6.300475059382423e-06,
+      "loss": 0.0287,
+      "step": 16786
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 6.294536817102138e-06,
+      "loss": 0.0237,
+      "step": 16787
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 6.2885985748218536e-06,
+      "loss": 0.0248,
+      "step": 16788
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 6.282660332541568e-06,
+      "loss": 0.0306,
+      "step": 16789
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 6.276722090261283e-06,
+      "loss": 0.0251,
+      "step": 16790
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 6.270783847980998e-06,
+      "loss": 0.0367,
+      "step": 16791
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 6.264845605700713e-06,
+      "loss": 0.0255,
+      "step": 16792
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 6.258907363420428e-06,
+      "loss": 0.0304,
+      "step": 16793
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 6.252969121140143e-06,
+      "loss": 0.0234,
+      "step": 16794
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 6.247030878859858e-06,
+      "loss": 0.0198,
+      "step": 16795
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 6.241092636579573e-06,
+      "loss": 0.0155,
+      "step": 16796
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 6.235154394299288e-06,
+      "loss": 0.0597,
+      "step": 16797
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 6.229216152019002e-06,
+      "loss": 0.0555,
+      "step": 16798
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 6.223277909738718e-06,
+      "loss": 0.0305,
+      "step": 16799
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 6.217339667458433e-06,
+      "loss": 0.0328,
+      "step": 16800
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 6.211401425178148e-06,
+      "loss": 0.0284,
+      "step": 16801
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 6.205463182897862e-06,
+      "loss": 0.0061,
+      "step": 16802
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 6.199524940617578e-06,
+      "loss": 0.0502,
+      "step": 16803
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 6.193586698337293e-06,
+      "loss": 0.0288,
+      "step": 16804
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 6.187648456057007e-06,
+      "loss": 0.0043,
+      "step": 16805
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 6.1817102137767226e-06,
+      "loss": 0.0084,
+      "step": 16806
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 6.175771971496437e-06,
+      "loss": 0.0372,
+      "step": 16807
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 6.169833729216152e-06,
+      "loss": 0.0312,
+      "step": 16808
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 6.1638954869358675e-06,
+      "loss": 0.0262,
+      "step": 16809
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 6.157957244655582e-06,
+      "loss": 0.0257,
+      "step": 16810
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 6.152019002375297e-06,
+      "loss": 0.0255,
+      "step": 16811
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 6.146080760095012e-06,
+      "loss": 0.0286,
+      "step": 16812
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 6.140142517814727e-06,
+      "loss": 0.0434,
+      "step": 16813
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 6.134204275534442e-06,
+      "loss": 0.0353,
+      "step": 16814
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 6.128266033254157e-06,
+      "loss": 0.0169,
+      "step": 16815
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 6.122327790973872e-06,
+      "loss": 0.0385,
+      "step": 16816
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 6.116389548693587e-06,
+      "loss": 0.0217,
+      "step": 16817
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 6.110451306413302e-06,
+      "loss": 0.0294,
+      "step": 16818
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 6.104513064133017e-06,
+      "loss": 0.0439,
+      "step": 16819
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 6.098574821852731e-06,
+      "loss": 0.0158,
+      "step": 16820
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 6.0926365795724475e-06,
+      "loss": 0.0662,
+      "step": 16821
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 6.086698337292162e-06,
+      "loss": 0.0217,
+      "step": 16822
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 6.080760095011877e-06,
+      "loss": 0.0341,
+      "step": 16823
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 6.074821852731592e-06,
+      "loss": 0.0251,
+      "step": 16824
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 6.068883610451306e-06,
+      "loss": 0.0217,
+      "step": 16825
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 6.062945368171022e-06,
+      "loss": 0.0217,
+      "step": 16826
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 6.0570071258907366e-06,
+      "loss": 0.0249,
+      "step": 16827
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 6.051068883610452e-06,
+      "loss": 0.0587,
+      "step": 16828
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 6.045130641330166e-06,
+      "loss": 0.0301,
+      "step": 16829
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 6.0391923990498815e-06,
+      "loss": 0.0245,
+      "step": 16830
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 6.033254156769597e-06,
+      "loss": 0.0162,
+      "step": 16831
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 6.027315914489311e-06,
+      "loss": 0.0417,
+      "step": 16832
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 6.0213776722090265e-06,
+      "loss": 0.0701,
+      "step": 16833
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 6.015439429928742e-06,
+      "loss": 0.0197,
+      "step": 16834
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 6.009501187648456e-06,
+      "loss": 0.0289,
+      "step": 16835
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 6.0035629453681715e-06,
+      "loss": 0.0267,
+      "step": 16836
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 5.997624703087886e-06,
+      "loss": 0.0366,
+      "step": 16837
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 5.991686460807601e-06,
+      "loss": 0.0483,
+      "step": 16838
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 5.9857482185273165e-06,
+      "loss": 0.0186,
+      "step": 16839
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 5.979809976247031e-06,
+      "loss": 0.0277,
+      "step": 16840
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 5.973871733966746e-06,
+      "loss": 0.0378,
+      "step": 16841
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 5.967933491686461e-06,
+      "loss": 0.0249,
+      "step": 16842
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 5.961995249406177e-06,
+      "loss": 0.029,
+      "step": 16843
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 5.956057007125891e-06,
+      "loss": 0.0264,
+      "step": 16844
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 5.950118764845606e-06,
+      "loss": 0.0174,
+      "step": 16845
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 5.944180522565321e-06,
+      "loss": 0.0351,
+      "step": 16846
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 5.938242280285035e-06,
+      "loss": 0.0218,
+      "step": 16847
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 5.932304038004751e-06,
+      "loss": 0.027,
+      "step": 16848
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 5.926365795724466e-06,
+      "loss": 0.0383,
+      "step": 16849
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 5.92042755344418e-06,
+      "loss": 0.0201,
+      "step": 16850
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 5.9144893111638955e-06,
+      "loss": 0.0211,
+      "step": 16851
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 5.908551068883611e-06,
+      "loss": 0.0202,
+      "step": 16852
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 5.902612826603326e-06,
+      "loss": 0.0048,
+      "step": 16853
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 5.8966745843230405e-06,
+      "loss": 0.0256,
+      "step": 16854
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 5.890736342042756e-06,
+      "loss": 0.0287,
+      "step": 16855
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 5.884798099762471e-06,
+      "loss": 0.0536,
+      "step": 16856
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 5.8788598574821855e-06,
+      "loss": 0.0485,
+      "step": 16857
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 5.872921615201901e-06,
+      "loss": 0.0607,
+      "step": 16858
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 5.866983372921615e-06,
+      "loss": 0.0321,
+      "step": 16859
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 5.8610451306413305e-06,
+      "loss": 0.0298,
+      "step": 16860
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 5.855106888361046e-06,
+      "loss": 0.0375,
+      "step": 16861
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 5.84916864608076e-06,
+      "loss": 0.0443,
+      "step": 16862
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 5.8432304038004754e-06,
+      "loss": 0.0427,
+      "step": 16863
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 5.83729216152019e-06,
+      "loss": 0.0312,
+      "step": 16864
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 5.831353919239905e-06,
+      "loss": 0.0617,
+      "step": 16865
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 5.82541567695962e-06,
+      "loss": 0.0457,
+      "step": 16866
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 5.819477434679335e-06,
+      "loss": 0.0236,
+      "step": 16867
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 5.81353919239905e-06,
+      "loss": 0.0313,
+      "step": 16868
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 5.807600950118765e-06,
+      "loss": 0.0951,
+      "step": 16869
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 5.801662707838481e-06,
+      "loss": 0.0132,
+      "step": 16870
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 5.795724465558195e-06,
+      "loss": 0.0135,
+      "step": 16871
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 5.7897862232779095e-06,
+      "loss": 0.0153,
+      "step": 16872
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 5.783847980997625e-06,
+      "loss": 0.0327,
+      "step": 16873
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 5.77790973871734e-06,
+      "loss": 0.0239,
+      "step": 16874
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 5.771971496437055e-06,
+      "loss": 0.0232,
+      "step": 16875
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 5.76603325415677e-06,
+      "loss": 0.0457,
+      "step": 16876
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 5.760095011876484e-06,
+      "loss": 0.0337,
+      "step": 16877
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 5.7541567695962e-06,
+      "loss": 0.0412,
+      "step": 16878
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 5.748218527315915e-06,
+      "loss": 0.0201,
+      "step": 16879
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 5.74228028503563e-06,
+      "loss": 0.0238,
+      "step": 16880
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 5.7363420427553445e-06,
+      "loss": 0.0346,
+      "step": 16881
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 5.73040380047506e-06,
+      "loss": 0.021,
+      "step": 16882
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 5.724465558194775e-06,
+      "loss": 0.0347,
+      "step": 16883
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 5.7185273159144894e-06,
+      "loss": 0.0165,
+      "step": 16884
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 5.712589073634205e-06,
+      "loss": 0.0206,
+      "step": 16885
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 5.706650831353919e-06,
+      "loss": 0.0167,
+      "step": 16886
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 5.700712589073634e-06,
+      "loss": 0.0237,
+      "step": 16887
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 5.69477434679335e-06,
+      "loss": 0.0193,
+      "step": 16888
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 5.688836104513064e-06,
+      "loss": 0.016,
+      "step": 16889
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 5.682897862232779e-06,
+      "loss": 0.0498,
+      "step": 16890
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 5.676959619952495e-06,
+      "loss": 0.0566,
+      "step": 16891
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 5.671021377672209e-06,
+      "loss": 0.0201,
+      "step": 16892
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 5.665083135391924e-06,
+      "loss": 0.0225,
+      "step": 16893
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 5.659144893111639e-06,
+      "loss": 0.0378,
+      "step": 16894
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 5.653206650831354e-06,
+      "loss": 0.0184,
+      "step": 16895
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 5.647268408551069e-06,
+      "loss": 0.0159,
+      "step": 16896
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 5.641330166270784e-06,
+      "loss": 0.0194,
+      "step": 16897
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 5.635391923990499e-06,
+      "loss": 0.0195,
+      "step": 16898
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 5.6294536817102135e-06,
+      "loss": 0.0144,
+      "step": 16899
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 5.62351543942993e-06,
+      "loss": 0.022,
+      "step": 16900
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 5.617577197149644e-06,
+      "loss": 0.0154,
+      "step": 16901
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 5.611638954869359e-06,
+      "loss": 0.0052,
+      "step": 16902
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 5.605700712589074e-06,
+      "loss": 0.0242,
+      "step": 16903
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 5.599762470308789e-06,
+      "loss": 0.0325,
+      "step": 16904
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 5.593824228028504e-06,
+      "loss": 0.0241,
+      "step": 16905
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 5.587885985748219e-06,
+      "loss": 0.0505,
+      "step": 16906
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 5.581947743467934e-06,
+      "loss": 0.0308,
+      "step": 16907
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 5.576009501187648e-06,
+      "loss": 0.0453,
+      "step": 16908
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 5.570071258907364e-06,
+      "loss": 0.045,
+      "step": 16909
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 5.564133016627079e-06,
+      "loss": 0.049,
+      "step": 16910
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 5.558194774346793e-06,
+      "loss": 0.0237,
+      "step": 16911
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 5.552256532066509e-06,
+      "loss": 0.0275,
+      "step": 16912
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 5.546318289786224e-06,
+      "loss": 0.037,
+      "step": 16913
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 5.540380047505938e-06,
+      "loss": 0.0242,
+      "step": 16914
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 5.534441805225654e-06,
+      "loss": 0.0331,
+      "step": 16915
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 5.528503562945368e-06,
+      "loss": 0.028,
+      "step": 16916
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 5.522565320665083e-06,
+      "loss": 0.0361,
+      "step": 16917
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 5.516627078384799e-06,
+      "loss": 0.0317,
+      "step": 16918
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 5.510688836104513e-06,
+      "loss": 0.0274,
+      "step": 16919
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 5.504750593824228e-06,
+      "loss": 0.0372,
+      "step": 16920
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 5.498812351543943e-06,
+      "loss": 0.0352,
+      "step": 16921
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 5.492874109263659e-06,
+      "loss": 0.0309,
+      "step": 16922
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 5.486935866983373e-06,
+      "loss": 0.0131,
+      "step": 16923
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 5.480997624703088e-06,
+      "loss": 0.0239,
+      "step": 16924
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 5.475059382422803e-06,
+      "loss": 0.0257,
+      "step": 16925
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 5.469121140142518e-06,
+      "loss": 0.0374,
+      "step": 16926
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 5.4631828978622335e-06,
+      "loss": 0.0363,
+      "step": 16927
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 5.457244655581948e-06,
+      "loss": 0.016,
+      "step": 16928
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 5.451306413301662e-06,
+      "loss": 0.0387,
+      "step": 16929
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 5.445368171021378e-06,
+      "loss": 0.0201,
+      "step": 16930
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 5.439429928741093e-06,
+      "loss": 0.0321,
+      "step": 16931
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 5.433491686460808e-06,
+      "loss": 0.0232,
+      "step": 16932
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 5.427553444180523e-06,
+      "loss": 0.0187,
+      "step": 16933
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 5.421615201900238e-06,
+      "loss": 0.0165,
+      "step": 16934
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 5.415676959619953e-06,
+      "loss": 0.0222,
+      "step": 16935
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 5.409738717339668e-06,
+      "loss": 0.0085,
+      "step": 16936
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 5.403800475059383e-06,
+      "loss": 0.0428,
+      "step": 16937
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 5.397862232779097e-06,
+      "loss": 0.032,
+      "step": 16938
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 5.391923990498813e-06,
+      "loss": 0.0292,
+      "step": 16939
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 5.385985748218528e-06,
+      "loss": 0.0377,
+      "step": 16940
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 5.380047505938242e-06,
+      "loss": 0.0348,
+      "step": 16941
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 5.374109263657958e-06,
+      "loss": 0.0072,
+      "step": 16942
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 5.368171021377672e-06,
+      "loss": 0.0289,
+      "step": 16943
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 5.362232779097387e-06,
+      "loss": 0.0075,
+      "step": 16944
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 5.3562945368171026e-06,
+      "loss": 0.0184,
+      "step": 16945
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 5.350356294536817e-06,
+      "loss": 0.03,
+      "step": 16946
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 5.344418052256532e-06,
+      "loss": 0.0182,
+      "step": 16947
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 5.3384798099762475e-06,
+      "loss": 0.0064,
+      "step": 16948
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 5.332541567695962e-06,
+      "loss": 0.0438,
+      "step": 16949
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 5.326603325415677e-06,
+      "loss": 0.0247,
+      "step": 16950
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 5.320665083135392e-06,
+      "loss": 0.0347,
+      "step": 16951
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 5.314726840855107e-06,
+      "loss": 0.0234,
+      "step": 16952
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 5.308788598574822e-06,
+      "loss": 0.0264,
+      "step": 16953
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 5.3028503562945375e-06,
+      "loss": 0.0323,
+      "step": 16954
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 5.296912114014252e-06,
+      "loss": 0.0264,
+      "step": 16955
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 5.290973871733966e-06,
+      "loss": 0.0416,
+      "step": 16956
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 5.2850356294536825e-06,
+      "loss": 0.0214,
+      "step": 16957
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 5.279097387173397e-06,
+      "loss": 0.0298,
+      "step": 16958
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 5.273159144893112e-06,
+      "loss": 0.0357,
+      "step": 16959
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 5.267220902612827e-06,
+      "loss": 0.0496,
+      "step": 16960
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 5.261282660332542e-06,
+      "loss": 0.0703,
+      "step": 16961
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 5.255344418052257e-06,
+      "loss": 0.0125,
+      "step": 16962
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 5.2494061757719716e-06,
+      "loss": 0.0212,
+      "step": 16963
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 5.243467933491687e-06,
+      "loss": 0.0435,
+      "step": 16964
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 5.237529691211401e-06,
+      "loss": 0.0303,
+      "step": 16965
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 5.2315914489311166e-06,
+      "loss": 0.0284,
+      "step": 16966
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 5.225653206650832e-06,
+      "loss": 0.0132,
+      "step": 16967
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 5.219714964370546e-06,
+      "loss": 0.0248,
+      "step": 16968
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 5.2137767220902615e-06,
+      "loss": 0.0392,
+      "step": 16969
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 5.207838479809977e-06,
+      "loss": 0.0243,
+      "step": 16970
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 5.201900237529691e-06,
+      "loss": 0.0276,
+      "step": 16971
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 5.1959619952494065e-06,
+      "loss": 0.0203,
+      "step": 16972
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 5.190023752969121e-06,
+      "loss": 0.0159,
+      "step": 16973
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 5.184085510688837e-06,
+      "loss": 0.029,
+      "step": 16974
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 5.1781472684085515e-06,
+      "loss": 0.0224,
+      "step": 16975
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 5.172209026128266e-06,
+      "loss": 0.0203,
+      "step": 16976
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 5.166270783847981e-06,
+      "loss": 0.0388,
+      "step": 16977
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 5.160332541567696e-06,
+      "loss": 0.0317,
+      "step": 16978
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 5.154394299287412e-06,
+      "loss": 0.0496,
+      "step": 16979
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 5.148456057007126e-06,
+      "loss": 0.0263,
+      "step": 16980
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 5.1425178147268414e-06,
+      "loss": 0.027,
+      "step": 16981
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 5.136579572446556e-06,
+      "loss": 0.0261,
+      "step": 16982
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 5.130641330166271e-06,
+      "loss": 0.0168,
+      "step": 16983
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 5.124703087885986e-06,
+      "loss": 0.032,
+      "step": 16984
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 5.118764845605701e-06,
+      "loss": 0.0187,
+      "step": 16985
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 5.112826603325416e-06,
+      "loss": 0.0243,
+      "step": 16986
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 5.1068883610451305e-06,
+      "loss": 0.0094,
+      "step": 16987
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 5.100950118764846e-06,
+      "loss": 0.0436,
+      "step": 16988
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 5.095011876484561e-06,
+      "loss": 0.0429,
+      "step": 16989
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 5.0890736342042755e-06,
+      "loss": 0.0704,
+      "step": 16990
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 5.083135391923991e-06,
+      "loss": 0.0216,
+      "step": 16991
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 5.077197149643706e-06,
+      "loss": 0.0243,
+      "step": 16992
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 5.0712589073634205e-06,
+      "loss": 0.0389,
+      "step": 16993
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 5.065320665083136e-06,
+      "loss": 0.0277,
+      "step": 16994
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 5.05938242280285e-06,
+      "loss": 0.0345,
+      "step": 16995
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 5.0534441805225655e-06,
+      "loss": 0.0178,
+      "step": 16996
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 5.047505938242281e-06,
+      "loss": 0.0525,
+      "step": 16997
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 5.041567695961995e-06,
+      "loss": 0.1588,
+      "step": 16998
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 5.0356294536817105e-06,
+      "loss": 0.0401,
+      "step": 16999
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 5.029691211401425e-06,
+      "loss": 0.0382,
+      "step": 17000
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 5.023752969121141e-06,
+      "loss": 0.019,
+      "step": 17001
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 5.0178147268408554e-06,
+      "loss": 0.0328,
+      "step": 17002
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 5.01187648456057e-06,
+      "loss": 0.0374,
+      "step": 17003
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 5.005938242280285e-06,
+      "loss": 0.0215,
+      "step": 17004
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 5e-06,
+      "loss": 0.0258,
+      "step": 17005
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 4.994061757719716e-06,
+      "loss": 0.0179,
+      "step": 17006
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 4.98812351543943e-06,
+      "loss": 0.0197,
+      "step": 17007
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 4.9821852731591445e-06,
+      "loss": 0.0344,
+      "step": 17008
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 4.976247030878861e-06,
+      "loss": 0.0306,
+      "step": 17009
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 4.970308788598575e-06,
+      "loss": 0.0257,
+      "step": 17010
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 4.96437054631829e-06,
+      "loss": 0.027,
+      "step": 17011
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 4.958432304038005e-06,
+      "loss": 0.0371,
+      "step": 17012
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 4.95249406175772e-06,
+      "loss": 0.0687,
+      "step": 17013
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 4.946555819477435e-06,
+      "loss": 0.019,
+      "step": 17014
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 4.94061757719715e-06,
+      "loss": 0.0295,
+      "step": 17015
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 4.934679334916865e-06,
+      "loss": 0.0353,
+      "step": 17016
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 4.9287410926365795e-06,
+      "loss": 0.0245,
+      "step": 17017
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 4.922802850356295e-06,
+      "loss": 0.0339,
+      "step": 17018
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 4.91686460807601e-06,
+      "loss": 0.025,
+      "step": 17019
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 4.9109263657957244e-06,
+      "loss": 0.0317,
+      "step": 17020
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 4.90498812351544e-06,
+      "loss": 0.051,
+      "step": 17021
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 4.899049881235154e-06,
+      "loss": 0.0533,
+      "step": 17022
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 4.8931116389548694e-06,
+      "loss": 0.0584,
+      "step": 17023
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 4.887173396674585e-06,
+      "loss": 0.0157,
+      "step": 17024
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 4.881235154394299e-06,
+      "loss": 0.0222,
+      "step": 17025
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 4.875296912114014e-06,
+      "loss": 0.0325,
+      "step": 17026
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 4.86935866983373e-06,
+      "loss": 0.0203,
+      "step": 17027
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 4.863420427553444e-06,
+      "loss": 0.0169,
+      "step": 17028
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 4.857482185273159e-06,
+      "loss": 0.0398,
+      "step": 17029
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 4.851543942992874e-06,
+      "loss": 0.0188,
+      "step": 17030
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 4.84560570071259e-06,
+      "loss": 0.0141,
+      "step": 17031
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 4.839667458432304e-06,
+      "loss": 0.0188,
+      "step": 17032
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 4.83372921615202e-06,
+      "loss": 0.0436,
+      "step": 17033
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 4.827790973871734e-06,
+      "loss": 0.0319,
+      "step": 17034
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 4.8218527315914485e-06,
+      "loss": 0.021,
+      "step": 17035
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 4.815914489311165e-06,
+      "loss": 0.0218,
+      "step": 17036
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 4.809976247030879e-06,
+      "loss": 0.0223,
+      "step": 17037
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 4.804038004750594e-06,
+      "loss": 0.0567,
+      "step": 17038
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 4.798099762470309e-06,
+      "loss": 0.0216,
+      "step": 17039
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 4.792161520190024e-06,
+      "loss": 0.031,
+      "step": 17040
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 4.786223277909739e-06,
+      "loss": 0.0379,
+      "step": 17041
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 4.780285035629454e-06,
+      "loss": 0.0271,
+      "step": 17042
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 4.774346793349169e-06,
+      "loss": 0.0295,
+      "step": 17043
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 4.768408551068884e-06,
+      "loss": 0.0296,
+      "step": 17044
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 4.762470308788599e-06,
+      "loss": 0.0119,
+      "step": 17045
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 4.756532066508314e-06,
+      "loss": 0.0202,
+      "step": 17046
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 4.750593824228028e-06,
+      "loss": 0.0114,
+      "step": 17047
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 4.744655581947744e-06,
+      "loss": 0.1265,
+      "step": 17048
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 4.738717339667459e-06,
+      "loss": 0.0423,
+      "step": 17049
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 4.732779097387173e-06,
+      "loss": 0.0553,
+      "step": 17050
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 4.726840855106889e-06,
+      "loss": 0.0257,
+      "step": 17051
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 4.720902612826603e-06,
+      "loss": 0.0332,
+      "step": 17052
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 4.714964370546319e-06,
+      "loss": 0.02,
+      "step": 17053
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 4.709026128266034e-06,
+      "loss": 0.0171,
+      "step": 17054
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 4.703087885985748e-06,
+      "loss": 0.0411,
+      "step": 17055
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 4.697149643705463e-06,
+      "loss": 0.0302,
+      "step": 17056
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 4.691211401425178e-06,
+      "loss": 0.0254,
+      "step": 17057
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 4.685273159144894e-06,
+      "loss": 0.0351,
+      "step": 17058
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 4.679334916864608e-06,
+      "loss": 0.0226,
+      "step": 17059
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 4.673396674584323e-06,
+      "loss": 0.0609,
+      "step": 17060
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 4.667458432304038e-06,
+      "loss": 0.031,
+      "step": 17061
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 4.661520190023753e-06,
+      "loss": 0.0287,
+      "step": 17062
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 4.6555819477434686e-06,
+      "loss": 0.0436,
+      "step": 17063
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 4.649643705463183e-06,
+      "loss": 0.0422,
+      "step": 17064
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 4.643705463182898e-06,
+      "loss": 0.0308,
+      "step": 17065
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 4.6377672209026135e-06,
+      "loss": 0.0411,
+      "step": 17066
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 4.631828978622328e-06,
+      "loss": 0.0262,
+      "step": 17067
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 4.625890736342043e-06,
+      "loss": 0.0119,
+      "step": 17068
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 4.619952494061758e-06,
+      "loss": 0.0265,
+      "step": 17069
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 4.614014251781473e-06,
+      "loss": 0.0271,
+      "step": 17070
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 4.608076009501188e-06,
+      "loss": 0.02,
+      "step": 17071
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 4.602137767220903e-06,
+      "loss": 0.0576,
+      "step": 17072
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 4.596199524940618e-06,
+      "loss": 0.0362,
+      "step": 17073
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 4.590261282660332e-06,
+      "loss": 0.0357,
+      "step": 17074
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 4.584323040380048e-06,
+      "loss": 0.0488,
+      "step": 17075
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 4.578384798099763e-06,
+      "loss": 0.0213,
+      "step": 17076
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 4.572446555819477e-06,
+      "loss": 0.0294,
+      "step": 17077
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 4.566508313539193e-06,
+      "loss": 0.0259,
+      "step": 17078
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 4.560570071258908e-06,
+      "loss": 0.0266,
+      "step": 17079
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 4.554631828978623e-06,
+      "loss": 0.0353,
+      "step": 17080
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 4.5486935866983376e-06,
+      "loss": 0.0231,
+      "step": 17081
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 4.542755344418052e-06,
+      "loss": 0.0214,
+      "step": 17082
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 4.536817102137767e-06,
+      "loss": 0.0153,
+      "step": 17083
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 4.5308788598574825e-06,
+      "loss": 0.0209,
+      "step": 17084
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 4.524940617577198e-06,
+      "loss": 0.0483,
+      "step": 17085
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 4.519002375296912e-06,
+      "loss": 0.0175,
+      "step": 17086
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 4.513064133016627e-06,
+      "loss": 0.0223,
+      "step": 17087
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 4.507125890736343e-06,
+      "loss": 0.0349,
+      "step": 17088
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 4.501187648456057e-06,
+      "loss": 0.0211,
+      "step": 17089
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 4.4952494061757725e-06,
+      "loss": 0.0265,
+      "step": 17090
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 4.489311163895487e-06,
+      "loss": 0.0537,
+      "step": 17091
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 4.483372921615202e-06,
+      "loss": 0.029,
+      "step": 17092
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 4.4774346793349175e-06,
+      "loss": 0.0097,
+      "step": 17093
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 4.471496437054632e-06,
+      "loss": 0.0311,
+      "step": 17094
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 4.465558194774347e-06,
+      "loss": 0.0323,
+      "step": 17095
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 4.459619952494062e-06,
+      "loss": 0.0338,
+      "step": 17096
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 4.453681710213777e-06,
+      "loss": 0.0141,
+      "step": 17097
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 4.447743467933492e-06,
+      "loss": 0.0193,
+      "step": 17098
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 4.441805225653207e-06,
+      "loss": 0.0363,
+      "step": 17099
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 4.435866983372922e-06,
+      "loss": 0.0387,
+      "step": 17100
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 4.429928741092637e-06,
+      "loss": 0.0412,
+      "step": 17101
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 4.4239904988123516e-06,
+      "loss": 0.0274,
+      "step": 17102
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 4.418052256532067e-06,
+      "loss": 0.0424,
+      "step": 17103
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 4.412114014251781e-06,
+      "loss": 0.0373,
+      "step": 17104
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 4.4061757719714965e-06,
+      "loss": 0.0311,
+      "step": 17105
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 4.400237529691212e-06,
+      "loss": 0.0264,
+      "step": 17106
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 4.394299287410926e-06,
+      "loss": 0.0213,
+      "step": 17107
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 4.3883610451306415e-06,
+      "loss": 0.0358,
+      "step": 17108
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 4.382422802850356e-06,
+      "loss": 0.0246,
+      "step": 17109
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 4.376484560570072e-06,
+      "loss": 0.027,
+      "step": 17110
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 4.3705463182897865e-06,
+      "loss": 0.0316,
+      "step": 17111
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 4.364608076009502e-06,
+      "loss": 0.0348,
+      "step": 17112
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 4.358669833729216e-06,
+      "loss": 0.0421,
+      "step": 17113
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 4.3527315914489315e-06,
+      "loss": 0.0175,
+      "step": 17114
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 4.346793349168647e-06,
+      "loss": 0.0221,
+      "step": 17115
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 4.340855106888361e-06,
+      "loss": 0.0177,
+      "step": 17116
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 4.3349168646080765e-06,
+      "loss": 0.0447,
+      "step": 17117
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 4.328978622327791e-06,
+      "loss": 0.0417,
+      "step": 17118
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 4.323040380047506e-06,
+      "loss": 0.028,
+      "step": 17119
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 4.3171021377672214e-06,
+      "loss": 0.0407,
+      "step": 17120
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 4.311163895486936e-06,
+      "loss": 0.04,
+      "step": 17121
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 4.305225653206651e-06,
+      "loss": 0.0274,
+      "step": 17122
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 4.299287410926366e-06,
+      "loss": 0.0542,
+      "step": 17123
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 4.293349168646081e-06,
+      "loss": 0.0185,
+      "step": 17124
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 4.287410926365796e-06,
+      "loss": 0.0408,
+      "step": 17125
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 4.2814726840855105e-06,
+      "loss": 0.0175,
+      "step": 17126
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 4.275534441805226e-06,
+      "loss": 0.0183,
+      "step": 17127
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 4.269596199524941e-06,
+      "loss": 0.0158,
+      "step": 17128
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 4.2636579572446555e-06,
+      "loss": 0.0395,
+      "step": 17129
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 4.257719714964371e-06,
+      "loss": 0.0121,
+      "step": 17130
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 4.251781472684085e-06,
+      "loss": 0.0347,
+      "step": 17131
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 4.245843230403801e-06,
+      "loss": 0.0277,
+      "step": 17132
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 4.239904988123516e-06,
+      "loss": 0.0206,
+      "step": 17133
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 4.23396674584323e-06,
+      "loss": 0.0183,
+      "step": 17134
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 4.2280285035629455e-06,
+      "loss": 0.0366,
+      "step": 17135
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 4.222090261282661e-06,
+      "loss": 0.0254,
+      "step": 17136
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 4.216152019002376e-06,
+      "loss": 0.044,
+      "step": 17137
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 4.2102137767220904e-06,
+      "loss": 0.0673,
+      "step": 17138
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 4.204275534441805e-06,
+      "loss": 0.0264,
+      "step": 17139
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 4.19833729216152e-06,
+      "loss": 0.0352,
+      "step": 17140
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 4.192399049881235e-06,
+      "loss": 0.0269,
+      "step": 17141
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 4.186460807600951e-06,
+      "loss": 0.0133,
+      "step": 17142
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 4.180522565320665e-06,
+      "loss": 0.0385,
+      "step": 17143
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 4.17458432304038e-06,
+      "loss": 0.0274,
+      "step": 17144
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 4.168646080760096e-06,
+      "loss": 0.0196,
+      "step": 17145
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 4.16270783847981e-06,
+      "loss": 0.0436,
+      "step": 17146
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 4.156769596199525e-06,
+      "loss": 0.0277,
+      "step": 17147
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 4.15083135391924e-06,
+      "loss": 0.0185,
+      "step": 17148
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 4.144893111638955e-06,
+      "loss": 0.0339,
+      "step": 17149
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 4.13895486935867e-06,
+      "loss": 0.027,
+      "step": 17150
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 4.133016627078385e-06,
+      "loss": 0.0319,
+      "step": 17151
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 4.1270783847981e-06,
+      "loss": 0.0264,
+      "step": 17152
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 4.1211401425178145e-06,
+      "loss": 0.0446,
+      "step": 17153
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 4.11520190023753e-06,
+      "loss": 0.0362,
+      "step": 17154
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 4.109263657957245e-06,
+      "loss": 0.0264,
+      "step": 17155
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 4.1033254156769595e-06,
+      "loss": 0.0476,
+      "step": 17156
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 4.097387173396675e-06,
+      "loss": 0.0287,
+      "step": 17157
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 4.09144893111639e-06,
+      "loss": 0.0228,
+      "step": 17158
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 4.0855106888361044e-06,
+      "loss": 0.0285,
+      "step": 17159
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 4.07957244655582e-06,
+      "loss": 0.0303,
+      "step": 17160
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 4.073634204275534e-06,
+      "loss": 0.0371,
+      "step": 17161
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 4.067695961995249e-06,
+      "loss": 0.0385,
+      "step": 17162
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 4.061757719714965e-06,
+      "loss": 0.0265,
+      "step": 17163
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 4.05581947743468e-06,
+      "loss": 0.0206,
+      "step": 17164
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 4.049881235154394e-06,
+      "loss": 0.0323,
+      "step": 17165
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 4.043942992874109e-06,
+      "loss": 0.0252,
+      "step": 17166
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 4.038004750593825e-06,
+      "loss": 0.0217,
+      "step": 17167
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 4.032066508313539e-06,
+      "loss": 0.0275,
+      "step": 17168
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 4.026128266033255e-06,
+      "loss": 0.0357,
+      "step": 17169
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 4.020190023752969e-06,
+      "loss": 0.0153,
+      "step": 17170
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 4.014251781472684e-06,
+      "loss": 0.0239,
+      "step": 17171
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 4.0083135391924e-06,
+      "loss": 0.0264,
+      "step": 17172
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 4.002375296912114e-06,
+      "loss": 0.0206,
+      "step": 17173
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 3.996437054631829e-06,
+      "loss": 0.0351,
+      "step": 17174
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 3.990498812351544e-06,
+      "loss": 0.036,
+      "step": 17175
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 3.984560570071259e-06,
+      "loss": 0.0223,
+      "step": 17176
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 3.978622327790974e-06,
+      "loss": 0.0212,
+      "step": 17177
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 3.972684085510689e-06,
+      "loss": 0.0264,
+      "step": 17178
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 3.966745843230404e-06,
+      "loss": 0.0203,
+      "step": 17179
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 3.960807600950119e-06,
+      "loss": 0.0221,
+      "step": 17180
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 3.954869358669834e-06,
+      "loss": 0.0167,
+      "step": 17181
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 3.948931116389549e-06,
+      "loss": 0.0403,
+      "step": 17182
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 3.942992874109263e-06,
+      "loss": 0.0146,
+      "step": 17183
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 3.9370546318289795e-06,
+      "loss": 0.0243,
+      "step": 17184
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 3.931116389548694e-06,
+      "loss": 0.0168,
+      "step": 17185
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 3.925178147268408e-06,
+      "loss": 0.0345,
+      "step": 17186
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 3.919239904988124e-06,
+      "loss": 0.0363,
+      "step": 17187
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 3.913301662707838e-06,
+      "loss": 0.0206,
+      "step": 17188
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 3.907363420427554e-06,
+      "loss": 0.0126,
+      "step": 17189
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 3.901425178147269e-06,
+      "loss": 0.0181,
+      "step": 17190
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 3.895486935866984e-06,
+      "loss": 0.021,
+      "step": 17191
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 3.889548693586698e-06,
+      "loss": 0.0319,
+      "step": 17192
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 3.883610451306414e-06,
+      "loss": 0.029,
+      "step": 17193
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 3.877672209026129e-06,
+      "loss": 0.015,
+      "step": 17194
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 3.871733966745843e-06,
+      "loss": 0.0052,
+      "step": 17195
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 3.865795724465559e-06,
+      "loss": 0.0239,
+      "step": 17196
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 3.859857482185273e-06,
+      "loss": 0.0361,
+      "step": 17197
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 3.853919239904988e-06,
+      "loss": 0.0249,
+      "step": 17198
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 3.8479809976247036e-06,
+      "loss": 0.0399,
+      "step": 17199
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 3.842042755344418e-06,
+      "loss": 0.0391,
+      "step": 17200
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 3.836104513064133e-06,
+      "loss": 0.0286,
+      "step": 17201
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 3.8301662707838485e-06,
+      "loss": 0.0361,
+      "step": 17202
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 3.824228028503563e-06,
+      "loss": 0.027,
+      "step": 17203
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 3.818289786223278e-06,
+      "loss": 0.0659,
+      "step": 17204
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 3.8123515439429927e-06,
+      "loss": 0.0295,
+      "step": 17205
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 3.8064133016627084e-06,
+      "loss": 0.0463,
+      "step": 17206
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 3.8004750593824232e-06,
+      "loss": 0.0552,
+      "step": 17207
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 3.794536817102138e-06,
+      "loss": 0.0167,
+      "step": 17208
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 3.788598574821853e-06,
+      "loss": 0.0241,
+      "step": 17209
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 3.7826603325415678e-06,
+      "loss": 0.0404,
+      "step": 17210
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 3.776722090261283e-06,
+      "loss": 0.0425,
+      "step": 17211
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 3.770783847980998e-06,
+      "loss": 0.0283,
+      "step": 17212
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 3.7648456057007128e-06,
+      "loss": 0.04,
+      "step": 17213
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 3.7589073634204276e-06,
+      "loss": 0.0511,
+      "step": 17214
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 3.752969121140143e-06,
+      "loss": 0.0258,
+      "step": 17215
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 3.7470308788598577e-06,
+      "loss": 0.0368,
+      "step": 17216
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 3.7410926365795726e-06,
+      "loss": 0.0058,
+      "step": 17217
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 3.7351543942992874e-06,
+      "loss": 0.0367,
+      "step": 17218
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 3.7292161520190023e-06,
+      "loss": 0.0292,
+      "step": 17219
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 3.7232779097387176e-06,
+      "loss": 0.0214,
+      "step": 17220
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 3.7173396674584324e-06,
+      "loss": 0.0206,
+      "step": 17221
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 3.7114014251781473e-06,
+      "loss": 0.0302,
+      "step": 17222
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 3.705463182897862e-06,
+      "loss": 0.0309,
+      "step": 17223
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 3.699524940617578e-06,
+      "loss": 0.0231,
+      "step": 17224
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 3.6935866983372922e-06,
+      "loss": 0.0213,
+      "step": 17225
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 3.687648456057007e-06,
+      "loss": 0.0187,
+      "step": 17226
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 3.681710213776722e-06,
+      "loss": 0.0206,
+      "step": 17227
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 3.6757719714964376e-06,
+      "loss": 0.032,
+      "step": 17228
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 3.6698337292161525e-06,
+      "loss": 0.0224,
+      "step": 17229
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 3.6638954869358673e-06,
+      "loss": 0.0237,
+      "step": 17230
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 3.6579572446555818e-06,
+      "loss": 0.0166,
+      "step": 17231
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 3.6520190023752966e-06,
+      "loss": 0.0197,
+      "step": 17232
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 3.6460807600950123e-06,
+      "loss": 0.0227,
+      "step": 17233
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 3.640142517814727e-06,
+      "loss": 0.0127,
+      "step": 17234
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 3.634204275534442e-06,
+      "loss": 0.0215,
+      "step": 17235
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 3.628266033254157e-06,
+      "loss": 0.0217,
+      "step": 17236
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 3.622327790973872e-06,
+      "loss": 0.0205,
+      "step": 17237
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 3.616389548693587e-06,
+      "loss": 0.0291,
+      "step": 17238
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 3.610451306413302e-06,
+      "loss": 0.0516,
+      "step": 17239
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 3.6045130641330167e-06,
+      "loss": 0.0209,
+      "step": 17240
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 3.598574821852732e-06,
+      "loss": 0.0232,
+      "step": 17241
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 3.592636579572447e-06,
+      "loss": 0.0294,
+      "step": 17242
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 3.5866983372921617e-06,
+      "loss": 0.0128,
+      "step": 17243
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 3.5807600950118765e-06,
+      "loss": 0.0258,
+      "step": 17244
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 3.5748218527315914e-06,
+      "loss": 0.0076,
+      "step": 17245
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 3.5688836104513067e-06,
+      "loss": 0.038,
+      "step": 17246
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 3.5629453681710215e-06,
+      "loss": 0.0185,
+      "step": 17247
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 3.5570071258907364e-06,
+      "loss": 0.0171,
+      "step": 17248
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 3.551068883610451e-06,
+      "loss": 0.034,
+      "step": 17249
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 3.545130641330167e-06,
+      "loss": 0.0335,
+      "step": 17250
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 3.5391923990498813e-06,
+      "loss": 0.042,
+      "step": 17251
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 3.533254156769596e-06,
+      "loss": 0.0293,
+      "step": 17252
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 3.527315914489311e-06,
+      "loss": 0.027,
+      "step": 17253
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 3.521377672209026e-06,
+      "loss": 0.027,
+      "step": 17254
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 3.5154394299287416e-06,
+      "loss": 0.0386,
+      "step": 17255
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 3.5095011876484564e-06,
+      "loss": 0.0414,
+      "step": 17256
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 3.503562945368171e-06,
+      "loss": 0.0329,
+      "step": 17257
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 3.4976247030878857e-06,
+      "loss": 0.055,
+      "step": 17258
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 3.4916864608076014e-06,
+      "loss": 0.0362,
+      "step": 17259
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 3.4857482185273163e-06,
+      "loss": 0.0327,
+      "step": 17260
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 3.479809976247031e-06,
+      "loss": 0.0407,
+      "step": 17261
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 3.473871733966746e-06,
+      "loss": 0.0172,
+      "step": 17262
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 3.4679334916864612e-06,
+      "loss": 0.0262,
+      "step": 17263
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 3.461995249406176e-06,
+      "loss": 0.0353,
+      "step": 17264
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 3.456057007125891e-06,
+      "loss": 0.0156,
+      "step": 17265
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 3.450118764845606e-06,
+      "loss": 0.0393,
+      "step": 17266
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 3.4441805225653207e-06,
+      "loss": 0.0157,
+      "step": 17267
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 3.438242280285036e-06,
+      "loss": 0.0401,
+      "step": 17268
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 3.4323040380047508e-06,
+      "loss": 0.0285,
+      "step": 17269
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 3.4263657957244656e-06,
+      "loss": 0.026,
+      "step": 17270
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 3.4204275534441805e-06,
+      "loss": 0.0173,
+      "step": 17271
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 3.4144893111638958e-06,
+      "loss": 0.0204,
+      "step": 17272
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 3.4085510688836106e-06,
+      "loss": 0.0432,
+      "step": 17273
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 3.4026128266033255e-06,
+      "loss": 0.027,
+      "step": 17274
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 3.3966745843230403e-06,
+      "loss": 0.0231,
+      "step": 17275
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 3.390736342042756e-06,
+      "loss": 0.0258,
+      "step": 17276
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 3.3847980997624704e-06,
+      "loss": 0.018,
+      "step": 17277
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 3.3788598574821853e-06,
+      "loss": 0.0189,
+      "step": 17278
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 3.3729216152019e-06,
+      "loss": 0.0212,
+      "step": 17279
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 3.366983372921615e-06,
+      "loss": 0.0234,
+      "step": 17280
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 3.3610451306413307e-06,
+      "loss": 0.0175,
+      "step": 17281
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 3.3551068883610455e-06,
+      "loss": 0.031,
+      "step": 17282
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 3.34916864608076e-06,
+      "loss": 0.0329,
+      "step": 17283
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 3.343230403800475e-06,
+      "loss": 0.0207,
+      "step": 17284
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 3.3372921615201905e-06,
+      "loss": 0.0177,
+      "step": 17285
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 3.3313539192399054e-06,
+      "loss": 0.0135,
+      "step": 17286
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 3.3254156769596202e-06,
+      "loss": 0.03,
+      "step": 17287
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 3.319477434679335e-06,
+      "loss": 0.0116,
+      "step": 17288
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 3.3135391923990495e-06,
+      "loss": 0.0166,
+      "step": 17289
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 3.307600950118765e-06,
+      "loss": 0.0137,
+      "step": 17290
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 3.30166270783848e-06,
+      "loss": 0.0365,
+      "step": 17291
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 3.295724465558195e-06,
+      "loss": 0.034,
+      "step": 17292
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 3.2897862232779097e-06,
+      "loss": 0.0254,
+      "step": 17293
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 3.283847980997625e-06,
+      "loss": 0.0388,
+      "step": 17294
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 3.27790973871734e-06,
+      "loss": 0.004,
+      "step": 17295
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 3.2719714964370547e-06,
+      "loss": 0.0248,
+      "step": 17296
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 3.2660332541567696e-06,
+      "loss": 0.0152,
+      "step": 17297
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 3.260095011876485e-06,
+      "loss": 0.0502,
+      "step": 17298
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 3.2541567695961997e-06,
+      "loss": 0.0376,
+      "step": 17299
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 3.2482185273159146e-06,
+      "loss": 0.0247,
+      "step": 17300
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 3.2422802850356294e-06,
+      "loss": 0.0279,
+      "step": 17301
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 3.2363420427553443e-06,
+      "loss": 0.0306,
+      "step": 17302
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 3.23040380047506e-06,
+      "loss": 0.0236,
+      "step": 17303
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 3.2244655581947744e-06,
+      "loss": 0.0331,
+      "step": 17304
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 3.2185273159144892e-06,
+      "loss": 0.0211,
+      "step": 17305
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 3.212589073634204e-06,
+      "loss": 0.0619,
+      "step": 17306
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 3.2066508313539198e-06,
+      "loss": 0.0276,
+      "step": 17307
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 3.2007125890736346e-06,
+      "loss": 0.0448,
+      "step": 17308
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 3.1947743467933495e-06,
+      "loss": 0.0184,
+      "step": 17309
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 3.188836104513064e-06,
+      "loss": 0.0187,
+      "step": 17310
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 3.1828978622327796e-06,
+      "loss": 0.0166,
+      "step": 17311
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 3.1769596199524945e-06,
+      "loss": 0.0367,
+      "step": 17312
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 3.1710213776722093e-06,
+      "loss": 0.0312,
+      "step": 17313
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 3.165083135391924e-06,
+      "loss": 0.0198,
+      "step": 17314
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 3.159144893111639e-06,
+      "loss": 0.0376,
+      "step": 17315
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 3.1532066508313543e-06,
+      "loss": 0.0153,
+      "step": 17316
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 3.147268408551069e-06,
+      "loss": 0.0147,
+      "step": 17317
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 3.141330166270784e-06,
+      "loss": 0.0179,
+      "step": 17318
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 3.135391923990499e-06,
+      "loss": 0.0454,
+      "step": 17319
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 3.129453681710214e-06,
+      "loss": 0.0227,
+      "step": 17320
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 3.123515439429929e-06,
+      "loss": 0.0485,
+      "step": 17321
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 3.117577197149644e-06,
+      "loss": 0.0175,
+      "step": 17322
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 3.111638954869359e-06,
+      "loss": 0.0203,
+      "step": 17323
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 3.105700712589074e-06,
+      "loss": 0.0272,
+      "step": 17324
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 3.099762470308789e-06,
+      "loss": 0.0238,
+      "step": 17325
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 3.0938242280285036e-06,
+      "loss": 0.0234,
+      "step": 17326
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 3.0878859857482185e-06,
+      "loss": 0.024,
+      "step": 17327
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 3.0819477434679338e-06,
+      "loss": 0.0235,
+      "step": 17328
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 3.0760095011876486e-06,
+      "loss": 0.0506,
+      "step": 17329
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 3.0700712589073635e-06,
+      "loss": 0.0115,
+      "step": 17330
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 3.0641330166270783e-06,
+      "loss": 0.0268,
+      "step": 17331
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 3.0581947743467936e-06,
+      "loss": 0.0272,
+      "step": 17332
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 3.0522565320665085e-06,
+      "loss": 0.0186,
+      "step": 17333
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 3.0463182897862237e-06,
+      "loss": 0.0234,
+      "step": 17334
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 3.0403800475059386e-06,
+      "loss": 0.0365,
+      "step": 17335
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 3.034441805225653e-06,
+      "loss": 0.0203,
+      "step": 17336
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 3.0285035629453683e-06,
+      "loss": 0.0068,
+      "step": 17337
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 3.022565320665083e-06,
+      "loss": 0.0327,
+      "step": 17338
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 3.0166270783847984e-06,
+      "loss": 0.017,
+      "step": 17339
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 3.0106888361045133e-06,
+      "loss": 0.0296,
+      "step": 17340
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 3.004750593824228e-06,
+      "loss": 0.0373,
+      "step": 17341
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.998812351543943e-06,
+      "loss": 0.0138,
+      "step": 17342
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.9928741092636582e-06,
+      "loss": 0.0145,
+      "step": 17343
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.986935866983373e-06,
+      "loss": 0.0096,
+      "step": 17344
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.9809976247030884e-06,
+      "loss": 0.0062,
+      "step": 17345
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.975059382422803e-06,
+      "loss": 0.0331,
+      "step": 17346
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.9691211401425176e-06,
+      "loss": 0.0039,
+      "step": 17347
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.963182897862233e-06,
+      "loss": 0.06,
+      "step": 17348
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.9572446555819478e-06,
+      "loss": 0.0368,
+      "step": 17349
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.951306413301663e-06,
+      "loss": 0.0153,
+      "step": 17350
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.945368171021378e-06,
+      "loss": 0.0297,
+      "step": 17351
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.9394299287410927e-06,
+      "loss": 0.0408,
+      "step": 17352
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.9334916864608076e-06,
+      "loss": 0.0231,
+      "step": 17353
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.927553444180523e-06,
+      "loss": 0.0286,
+      "step": 17354
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.9216152019002377e-06,
+      "loss": 0.0331,
+      "step": 17355
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.9156769596199526e-06,
+      "loss": 0.0317,
+      "step": 17356
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.9097387173396674e-06,
+      "loss": 0.0224,
+      "step": 17357
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.9038004750593827e-06,
+      "loss": 0.0282,
+      "step": 17358
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.8978622327790975e-06,
+      "loss": 0.0299,
+      "step": 17359
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.8919239904988124e-06,
+      "loss": 0.0274,
+      "step": 17360
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.8859857482185277e-06,
+      "loss": 0.0359,
+      "step": 17361
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.880047505938242e-06,
+      "loss": 0.0224,
+      "step": 17362
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.8741092636579574e-06,
+      "loss": 0.0465,
+      "step": 17363
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.8681710213776722e-06,
+      "loss": 0.0196,
+      "step": 17364
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.8622327790973875e-06,
+      "loss": 0.0336,
+      "step": 17365
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.8562945368171024e-06,
+      "loss": 0.0434,
+      "step": 17366
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.850356294536817e-06,
+      "loss": 0.0157,
+      "step": 17367
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.844418052256532e-06,
+      "loss": 0.0223,
+      "step": 17368
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.8384798099762473e-06,
+      "loss": 0.021,
+      "step": 17369
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.832541567695962e-06,
+      "loss": 0.0192,
+      "step": 17370
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.826603325415677e-06,
+      "loss": 0.0163,
+      "step": 17371
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.820665083135392e-06,
+      "loss": 0.0283,
+      "step": 17372
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.8147268408551067e-06,
+      "loss": 0.0241,
+      "step": 17373
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.808788598574822e-06,
+      "loss": 0.0297,
+      "step": 17374
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.802850356294537e-06,
+      "loss": 0.0292,
+      "step": 17375
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.796912114014252e-06,
+      "loss": 0.0118,
+      "step": 17376
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.790973871733967e-06,
+      "loss": 0.0138,
+      "step": 17377
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.785035629453682e-06,
+      "loss": 0.0307,
+      "step": 17378
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.7790973871733967e-06,
+      "loss": 0.046,
+      "step": 17379
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.773159144893112e-06,
+      "loss": 0.0404,
+      "step": 17380
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.767220902612827e-06,
+      "loss": 0.0166,
+      "step": 17381
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.7612826603325417e-06,
+      "loss": 0.0271,
+      "step": 17382
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.7553444180522565e-06,
+      "loss": 0.0462,
+      "step": 17383
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.7494061757719714e-06,
+      "loss": 0.034,
+      "step": 17384
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.7434679334916866e-06,
+      "loss": 0.0129,
+      "step": 17385
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.7375296912114015e-06,
+      "loss": 0.0274,
+      "step": 17386
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.7315914489311168e-06,
+      "loss": 0.0303,
+      "step": 17387
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.725653206650831e-06,
+      "loss": 0.0215,
+      "step": 17388
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.7197149643705465e-06,
+      "loss": 0.0254,
+      "step": 17389
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.7137767220902613e-06,
+      "loss": 0.0103,
+      "step": 17390
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.7078384798099766e-06,
+      "loss": 0.0403,
+      "step": 17391
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.7019002375296915e-06,
+      "loss": 0.0069,
+      "step": 17392
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.6959619952494063e-06,
+      "loss": 0.0245,
+      "step": 17393
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.690023752969121e-06,
+      "loss": 0.0126,
+      "step": 17394
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.684085510688836e-06,
+      "loss": 0.0069,
+      "step": 17395
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.6781472684085513e-06,
+      "loss": 0.0032,
+      "step": 17396
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.672209026128266e-06,
+      "loss": 0.0283,
+      "step": 17397
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.666270783847981e-06,
+      "loss": 0.0567,
+      "step": 17398
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.660332541567696e-06,
+      "loss": 0.0385,
+      "step": 17399
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.654394299287411e-06,
+      "loss": 0.0487,
+      "step": 17400
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.648456057007126e-06,
+      "loss": 0.0271,
+      "step": 17401
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.6425178147268412e-06,
+      "loss": 0.0264,
+      "step": 17402
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.636579572446556e-06,
+      "loss": 0.0373,
+      "step": 17403
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.630641330166271e-06,
+      "loss": 0.0445,
+      "step": 17404
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.6247030878859858e-06,
+      "loss": 0.0259,
+      "step": 17405
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.6187648456057006e-06,
+      "loss": 0.0171,
+      "step": 17406
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.612826603325416e-06,
+      "loss": 0.0314,
+      "step": 17407
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.6068883610451308e-06,
+      "loss": 0.0354,
+      "step": 17408
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.6009501187648456e-06,
+      "loss": 0.0332,
+      "step": 17409
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.5950118764845605e-06,
+      "loss": 0.0235,
+      "step": 17410
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.5890736342042757e-06,
+      "loss": 0.0255,
+      "step": 17411
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.5831353919239906e-06,
+      "loss": 0.02,
+      "step": 17412
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.577197149643706e-06,
+      "loss": 0.048,
+      "step": 17413
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.5712589073634207e-06,
+      "loss": 0.0417,
+      "step": 17414
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.5653206650831356e-06,
+      "loss": 0.0287,
+      "step": 17415
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.5593824228028504e-06,
+      "loss": 0.0401,
+      "step": 17416
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.5534441805225653e-06,
+      "loss": 0.025,
+      "step": 17417
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.5475059382422805e-06,
+      "loss": 0.0284,
+      "step": 17418
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.5415676959619954e-06,
+      "loss": 0.0337,
+      "step": 17419
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.5356294536817103e-06,
+      "loss": 0.0345,
+      "step": 17420
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.529691211401425e-06,
+      "loss": 0.0297,
+      "step": 17421
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.5237529691211404e-06,
+      "loss": 0.0263,
+      "step": 17422
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.5178147268408552e-06,
+      "loss": 0.016,
+      "step": 17423
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.5118764845605705e-06,
+      "loss": 0.0481,
+      "step": 17424
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.505938242280285e-06,
+      "loss": 0.0267,
+      "step": 17425
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.5e-06,
+      "loss": 0.0307,
+      "step": 17426
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.494061757719715e-06,
+      "loss": 0.0365,
+      "step": 17427
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.4881235154394303e-06,
+      "loss": 0.0348,
+      "step": 17428
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.482185273159145e-06,
+      "loss": 0.019,
+      "step": 17429
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.47624703087886e-06,
+      "loss": 0.0201,
+      "step": 17430
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.470308788598575e-06,
+      "loss": 0.017,
+      "step": 17431
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.4643705463182897e-06,
+      "loss": 0.0333,
+      "step": 17432
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.458432304038005e-06,
+      "loss": 0.062,
+      "step": 17433
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.45249406175772e-06,
+      "loss": 0.0176,
+      "step": 17434
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.4465558194774347e-06,
+      "loss": 0.0202,
+      "step": 17435
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.4406175771971496e-06,
+      "loss": 0.0142,
+      "step": 17436
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.434679334916865e-06,
+      "loss": 0.0138,
+      "step": 17437
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.4287410926365797e-06,
+      "loss": 0.0146,
+      "step": 17438
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.422802850356295e-06,
+      "loss": 0.0345,
+      "step": 17439
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.41686460807601e-06,
+      "loss": 0.0182,
+      "step": 17440
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.4109263657957242e-06,
+      "loss": 0.0321,
+      "step": 17441
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.4049881235154395e-06,
+      "loss": 0.0064,
+      "step": 17442
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.3990498812351544e-06,
+      "loss": 0.0457,
+      "step": 17443
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.3931116389548696e-06,
+      "loss": 0.0078,
+      "step": 17444
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.3871733966745845e-06,
+      "loss": 0.0411,
+      "step": 17445
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.3812351543942993e-06,
+      "loss": 0.1437,
+      "step": 17446
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.375296912114014e-06,
+      "loss": 0.085,
+      "step": 17447
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.3693586698337295e-06,
+      "loss": 0.0051,
+      "step": 17448
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.3634204275534443e-06,
+      "loss": 0.0485,
+      "step": 17449
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.3574821852731596e-06,
+      "loss": 0.044,
+      "step": 17450
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.351543942992874e-06,
+      "loss": 0.0582,
+      "step": 17451
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.345605700712589e-06,
+      "loss": 0.0389,
+      "step": 17452
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.339667458432304e-06,
+      "loss": 0.016,
+      "step": 17453
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.333729216152019e-06,
+      "loss": 0.0381,
+      "step": 17454
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.3277909738717343e-06,
+      "loss": 0.0242,
+      "step": 17455
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.321852731591449e-06,
+      "loss": 0.045,
+      "step": 17456
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.315914489311164e-06,
+      "loss": 0.0206,
+      "step": 17457
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.309976247030879e-06,
+      "loss": 0.0263,
+      "step": 17458
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.304038004750594e-06,
+      "loss": 0.0518,
+      "step": 17459
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.298099762470309e-06,
+      "loss": 0.0208,
+      "step": 17460
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.292161520190024e-06,
+      "loss": 0.0373,
+      "step": 17461
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.2862232779097387e-06,
+      "loss": 0.0254,
+      "step": 17462
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.280285035629454e-06,
+      "loss": 0.0257,
+      "step": 17463
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.2743467933491688e-06,
+      "loss": 0.0471,
+      "step": 17464
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.2684085510688836e-06,
+      "loss": 0.0297,
+      "step": 17465
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.262470308788599e-06,
+      "loss": 0.0351,
+      "step": 17466
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.2565320665083133e-06,
+      "loss": 0.0237,
+      "step": 17467
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.2505938242280286e-06,
+      "loss": 0.0323,
+      "step": 17468
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.2446555819477435e-06,
+      "loss": 0.0239,
+      "step": 17469
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.2387173396674587e-06,
+      "loss": 0.0299,
+      "step": 17470
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.2327790973871736e-06,
+      "loss": 0.0314,
+      "step": 17471
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.2268408551068884e-06,
+      "loss": 0.0224,
+      "step": 17472
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.2209026128266033e-06,
+      "loss": 0.043,
+      "step": 17473
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.2149643705463186e-06,
+      "loss": 0.0332,
+      "step": 17474
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.2090261282660334e-06,
+      "loss": 0.0464,
+      "step": 17475
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.2030878859857483e-06,
+      "loss": 0.025,
+      "step": 17476
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.197149643705463e-06,
+      "loss": 0.0158,
+      "step": 17477
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.191211401425178e-06,
+      "loss": 0.0296,
+      "step": 17478
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.1852731591448932e-06,
+      "loss": 0.0136,
+      "step": 17479
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.179334916864608e-06,
+      "loss": 0.0196,
+      "step": 17480
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.1733966745843234e-06,
+      "loss": 0.0182,
+      "step": 17481
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.1674584323040382e-06,
+      "loss": 0.0209,
+      "step": 17482
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.161520190023753e-06,
+      "loss": 0.0425,
+      "step": 17483
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.155581947743468e-06,
+      "loss": 0.0194,
+      "step": 17484
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.149643705463183e-06,
+      "loss": 0.0323,
+      "step": 17485
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.143705463182898e-06,
+      "loss": 0.0238,
+      "step": 17486
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.137767220902613e-06,
+      "loss": 0.0678,
+      "step": 17487
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.1318289786223278e-06,
+      "loss": 0.034,
+      "step": 17488
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.1258907363420426e-06,
+      "loss": 0.0185,
+      "step": 17489
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.119952494061758e-06,
+      "loss": 0.0345,
+      "step": 17490
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.1140142517814727e-06,
+      "loss": 0.0152,
+      "step": 17491
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.108076009501188e-06,
+      "loss": 0.0248,
+      "step": 17492
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.1021377672209024e-06,
+      "loss": 0.0265,
+      "step": 17493
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.0961995249406177e-06,
+      "loss": 0.038,
+      "step": 17494
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.0902612826603326e-06,
+      "loss": 0.0188,
+      "step": 17495
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.084323040380048e-06,
+      "loss": 0.0105,
+      "step": 17496
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.0783847980997627e-06,
+      "loss": 0.0536,
+      "step": 17497
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.0724465558194775e-06,
+      "loss": 0.0098,
+      "step": 17498
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.0665083135391924e-06,
+      "loss": 0.0352,
+      "step": 17499
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.0605700712589072e-06,
+      "loss": 0.0434,
+      "step": 17500
+    },
+    {
+      "epoch": 19.62,
+      "eval_loss": 1.85287344455719,
+      "eval_runtime": 896.1552,
+      "eval_samples_per_second": 2.948,
+      "eval_steps_per_second": 0.369,
+      "eval_wer": 0.9976993256644189,
+      "step": 17500
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.0546318289786225e-06,
+      "loss": 0.0183,
+      "step": 17501
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.0486935866983374e-06,
+      "loss": 0.0399,
+      "step": 17502
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.0427553444180522e-06,
+      "loss": 0.0297,
+      "step": 17503
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.036817102137767e-06,
+      "loss": 0.042,
+      "step": 17504
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.0308788598574823e-06,
+      "loss": 0.0329,
+      "step": 17505
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.024940617577197e-06,
+      "loss": 0.0374,
+      "step": 17506
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.0190023752969125e-06,
+      "loss": 0.0507,
+      "step": 17507
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.0130641330166273e-06,
+      "loss": 0.0243,
+      "step": 17508
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.007125890736342e-06,
+      "loss": 0.0126,
+      "step": 17509
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.001187648456057e-06,
+      "loss": 0.0461,
+      "step": 17510
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 1.995249406175772e-06,
+      "loss": 0.0352,
+      "step": 17511
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 1.989311163895487e-06,
+      "loss": 0.0169,
+      "step": 17512
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 1.983372921615202e-06,
+      "loss": 0.0432,
+      "step": 17513
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 1.977434679334917e-06,
+      "loss": 0.0392,
+      "step": 17514
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 1.9714964370546317e-06,
+      "loss": 0.052,
+      "step": 17515
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 1.965558194774347e-06,
+      "loss": 0.0284,
+      "step": 17516
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 1.959619952494062e-06,
+      "loss": 0.0433,
+      "step": 17517
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 1.953681710213777e-06,
+      "loss": 0.0245,
+      "step": 17518
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 1.947743467933492e-06,
+      "loss": 0.0269,
+      "step": 17519
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 1.941805225653207e-06,
+      "loss": 0.0311,
+      "step": 17520
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 1.9358669833729217e-06,
+      "loss": 0.0329,
+      "step": 17521
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 1.9299287410926365e-06,
+      "loss": 0.0251,
+      "step": 17522
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 1.9239904988123518e-06,
+      "loss": 0.0277,
+      "step": 17523
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 1.9180522565320666e-06,
+      "loss": 0.0388,
+      "step": 17524
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 1.9121140142517815e-06,
+      "loss": 0.0184,
+      "step": 17525
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 1.9061757719714963e-06,
+      "loss": 0.034,
+      "step": 17526
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 1.9002375296912116e-06,
+      "loss": 0.0115,
+      "step": 17527
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 1.8942992874109265e-06,
+      "loss": 0.0281,
+      "step": 17528
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 1.8883610451306415e-06,
+      "loss": 0.0404,
+      "step": 17529
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 1.8824228028503564e-06,
+      "loss": 0.0437,
+      "step": 17530
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 1.8764845605700714e-06,
+      "loss": 0.0178,
+      "step": 17531
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 1.8705463182897863e-06,
+      "loss": 0.0299,
+      "step": 17532
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 1.8646080760095011e-06,
+      "loss": 0.0233,
+      "step": 17533
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 1.8586698337292162e-06,
+      "loss": 0.0302,
+      "step": 17534
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 1.852731591448931e-06,
+      "loss": 0.0202,
+      "step": 17535
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 1.8467933491686461e-06,
+      "loss": 0.0492,
+      "step": 17536
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 1.840855106888361e-06,
+      "loss": 0.0201,
+      "step": 17537
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 1.8349168646080762e-06,
+      "loss": 0.0148,
+      "step": 17538
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 1.8289786223277909e-06,
+      "loss": 0.0271,
+      "step": 17539
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 1.8230403800475062e-06,
+      "loss": 0.0154,
+      "step": 17540
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 1.817102137767221e-06,
+      "loss": 0.0174,
+      "step": 17541
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 1.811163895486936e-06,
+      "loss": 0.0523,
+      "step": 17542
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 1.805225653206651e-06,
+      "loss": 0.0547,
+      "step": 17543
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 1.799287410926366e-06,
+      "loss": 0.0104,
+      "step": 17544
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 1.7933491686460808e-06,
+      "loss": 0.056,
+      "step": 17545
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 1.7874109263657957e-06,
+      "loss": 0.0067,
+      "step": 17546
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 1.7814726840855108e-06,
+      "loss": 0.0036,
+      "step": 17547
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 1.7755344418052256e-06,
+      "loss": 0.0193,
+      "step": 17548
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 1.7695961995249407e-06,
+      "loss": 0.024,
+      "step": 17549
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 1.7636579572446555e-06,
+      "loss": 0.0503,
+      "step": 17550
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 1.7577197149643708e-06,
+      "loss": 0.0392,
+      "step": 17551
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 1.7517814726840854e-06,
+      "loss": 0.0467,
+      "step": 17552
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 1.7458432304038007e-06,
+      "loss": 0.0338,
+      "step": 17553
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 1.7399049881235156e-06,
+      "loss": 0.049,
+      "step": 17554
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 1.7339667458432306e-06,
+      "loss": 0.0478,
+      "step": 17555
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 1.7280285035629455e-06,
+      "loss": 0.0262,
+      "step": 17556
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 1.7220902612826603e-06,
+      "loss": 0.0292,
+      "step": 17557
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 1.7161520190023754e-06,
+      "loss": 0.0233,
+      "step": 17558
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 1.7102137767220902e-06,
+      "loss": 0.0504,
+      "step": 17559
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 1.7042755344418053e-06,
+      "loss": 0.0233,
+      "step": 17560
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 1.6983372921615202e-06,
+      "loss": 0.022,
+      "step": 17561
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 1.6923990498812352e-06,
+      "loss": 0.0338,
+      "step": 17562
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 1.68646080760095e-06,
+      "loss": 0.0469,
+      "step": 17563
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 1.6805225653206653e-06,
+      "loss": 0.0457,
+      "step": 17564
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 1.67458432304038e-06,
+      "loss": 0.0288,
+      "step": 17565
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 1.6686460807600953e-06,
+      "loss": 0.0208,
+      "step": 17566
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 1.6627078384798101e-06,
+      "loss": 0.0169,
+      "step": 17567
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 1.6567695961995247e-06,
+      "loss": 0.0367,
+      "step": 17568
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 1.65083135391924e-06,
+      "loss": 0.0254,
+      "step": 17569
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 1.6448931116389549e-06,
+      "loss": 0.0263,
+      "step": 17570
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 1.63895486935867e-06,
+      "loss": 0.0497,
+      "step": 17571
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 1.6330166270783848e-06,
+      "loss": 0.0265,
+      "step": 17572
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 1.6270783847980999e-06,
+      "loss": 0.0405,
+      "step": 17573
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 1.6211401425178147e-06,
+      "loss": 0.0219,
+      "step": 17574
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 1.61520190023753e-06,
+      "loss": 0.0193,
+      "step": 17575
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 1.6092636579572446e-06,
+      "loss": 0.0389,
+      "step": 17576
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 1.6033254156769599e-06,
+      "loss": 0.0147,
+      "step": 17577
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 1.5973871733966747e-06,
+      "loss": 0.0338,
+      "step": 17578
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 1.5914489311163898e-06,
+      "loss": 0.0224,
+      "step": 17579
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 1.5855106888361047e-06,
+      "loss": 0.0227,
+      "step": 17580
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 1.5795724465558195e-06,
+      "loss": 0.0257,
+      "step": 17581
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 1.5736342042755346e-06,
+      "loss": 0.0257,
+      "step": 17582
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 1.5676959619952494e-06,
+      "loss": 0.0488,
+      "step": 17583
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 1.5617577197149645e-06,
+      "loss": 0.0351,
+      "step": 17584
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 1.5558194774346795e-06,
+      "loss": 0.0434,
+      "step": 17585
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 1.5498812351543944e-06,
+      "loss": 0.0203,
+      "step": 17586
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 1.5439429928741092e-06,
+      "loss": 0.0254,
+      "step": 17587
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 1.5380047505938243e-06,
+      "loss": 0.0178,
+      "step": 17588
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 1.5320665083135392e-06,
+      "loss": 0.012,
+      "step": 17589
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 1.5261282660332542e-06,
+      "loss": 0.02,
+      "step": 17590
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 1.5201900237529693e-06,
+      "loss": 0.0258,
+      "step": 17591
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 1.5142517814726841e-06,
+      "loss": 0.0265,
+      "step": 17592
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 1.5083135391923992e-06,
+      "loss": 0.0128,
+      "step": 17593
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 1.502375296912114e-06,
+      "loss": 0.0129,
+      "step": 17594
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 1.4964370546318291e-06,
+      "loss": 0.034,
+      "step": 17595
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 1.4904988123515442e-06,
+      "loss": 0.0882,
+      "step": 17596
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 1.4845605700712588e-06,
+      "loss": 0.0524,
+      "step": 17597
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 1.4786223277909739e-06,
+      "loss": 0.0584,
+      "step": 17598
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 1.472684085510689e-06,
+      "loss": 0.0383,
+      "step": 17599
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 1.4667458432304038e-06,
+      "loss": 0.0307,
+      "step": 17600
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 1.4608076009501189e-06,
+      "loss": 0.0339,
+      "step": 17601
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 1.4548693586698337e-06,
+      "loss": 0.0507,
+      "step": 17602
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 1.4489311163895488e-06,
+      "loss": 0.0305,
+      "step": 17603
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 1.4429928741092638e-06,
+      "loss": 0.0227,
+      "step": 17604
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 1.4370546318289787e-06,
+      "loss": 0.0295,
+      "step": 17605
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 1.4311163895486938e-06,
+      "loss": 0.0225,
+      "step": 17606
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 1.4251781472684086e-06,
+      "loss": 0.0319,
+      "step": 17607
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 1.4192399049881237e-06,
+      "loss": 0.0258,
+      "step": 17608
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 1.4133016627078385e-06,
+      "loss": 0.0161,
+      "step": 17609
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 1.4073634204275534e-06,
+      "loss": 0.0251,
+      "step": 17610
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 1.4014251781472684e-06,
+      "loss": 0.021,
+      "step": 17611
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 1.3954869358669835e-06,
+      "loss": 0.0282,
+      "step": 17612
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 1.3895486935866983e-06,
+      "loss": 0.0394,
+      "step": 17613
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 1.3836104513064134e-06,
+      "loss": 0.026,
+      "step": 17614
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 1.3776722090261283e-06,
+      "loss": 0.0357,
+      "step": 17615
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 1.3717339667458433e-06,
+      "loss": 0.0266,
+      "step": 17616
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 1.3657957244655584e-06,
+      "loss": 0.0199,
+      "step": 17617
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 1.3598574821852732e-06,
+      "loss": 0.0329,
+      "step": 17618
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 1.3539192399049883e-06,
+      "loss": 0.0276,
+      "step": 17619
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 1.3479809976247032e-06,
+      "loss": 0.0503,
+      "step": 17620
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 1.342042755344418e-06,
+      "loss": 0.028,
+      "step": 17621
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 1.336104513064133e-06,
+      "loss": 0.0129,
+      "step": 17622
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 1.330166270783848e-06,
+      "loss": 0.0214,
+      "step": 17623
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 1.324228028503563e-06,
+      "loss": 0.0217,
+      "step": 17624
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 1.318289786223278e-06,
+      "loss": 0.0398,
+      "step": 17625
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 1.3123515439429929e-06,
+      "loss": 0.043,
+      "step": 17626
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 1.306413301662708e-06,
+      "loss": 0.0217,
+      "step": 17627
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 1.3004750593824228e-06,
+      "loss": 0.0233,
+      "step": 17628
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 1.2945368171021379e-06,
+      "loss": 0.0275,
+      "step": 17629
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 1.288598574821853e-06,
+      "loss": 0.0242,
+      "step": 17630
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 1.2826603325415678e-06,
+      "loss": 0.0174,
+      "step": 17631
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 1.2767220902612826e-06,
+      "loss": 0.0583,
+      "step": 17632
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 1.2707838479809977e-06,
+      "loss": 0.0207,
+      "step": 17633
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 1.2648456057007126e-06,
+      "loss": 0.03,
+      "step": 17634
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 1.2589073634204276e-06,
+      "loss": 0.0472,
+      "step": 17635
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 1.2529691211401425e-06,
+      "loss": 0.0196,
+      "step": 17636
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 1.2470308788598575e-06,
+      "loss": 0.0407,
+      "step": 17637
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 1.2410926365795726e-06,
+      "loss": 0.0336,
+      "step": 17638
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 1.2351543942992874e-06,
+      "loss": 0.0122,
+      "step": 17639
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 1.2292161520190025e-06,
+      "loss": 0.0283,
+      "step": 17640
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 1.2232779097387174e-06,
+      "loss": 0.0287,
+      "step": 17641
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 1.2173396674584324e-06,
+      "loss": 0.0144,
+      "step": 17642
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 1.2114014251781475e-06,
+      "loss": 0.0146,
+      "step": 17643
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 1.2054631828978621e-06,
+      "loss": 0.0342,
+      "step": 17644
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 1.1995249406175772e-06,
+      "loss": 0.0241,
+      "step": 17645
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 1.1935866983372922e-06,
+      "loss": 0.0541,
+      "step": 17646
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 1.187648456057007e-06,
+      "loss": 0.0222,
+      "step": 17647
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 1.1817102137767222e-06,
+      "loss": 0.0081,
+      "step": 17648
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 1.175771971496437e-06,
+      "loss": 0.0333,
+      "step": 17649
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 1.169833729216152e-06,
+      "loss": 0.0466,
+      "step": 17650
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 1.1638954869358671e-06,
+      "loss": 0.0272,
+      "step": 17651
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 1.157957244655582e-06,
+      "loss": 0.0205,
+      "step": 17652
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 1.152019002375297e-06,
+      "loss": 0.0345,
+      "step": 17653
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 1.146080760095012e-06,
+      "loss": 0.0152,
+      "step": 17654
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 1.140142517814727e-06,
+      "loss": 0.0179,
+      "step": 17655
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 1.1342042755344418e-06,
+      "loss": 0.0352,
+      "step": 17656
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 1.1282660332541567e-06,
+      "loss": 0.0293,
+      "step": 17657
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 1.1223277909738717e-06,
+      "loss": 0.0225,
+      "step": 17658
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 1.1163895486935868e-06,
+      "loss": 0.0136,
+      "step": 17659
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 1.1104513064133016e-06,
+      "loss": 0.0289,
+      "step": 17660
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 1.1045130641330167e-06,
+      "loss": 0.0235,
+      "step": 17661
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 1.0985748218527316e-06,
+      "loss": 0.0195,
+      "step": 17662
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 1.0926365795724466e-06,
+      "loss": 0.0119,
+      "step": 17663
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 1.0866983372921617e-06,
+      "loss": 0.0357,
+      "step": 17664
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 1.0807600950118765e-06,
+      "loss": 0.037,
+      "step": 17665
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 1.0748218527315916e-06,
+      "loss": 0.0332,
+      "step": 17666
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 1.0688836104513065e-06,
+      "loss": 0.0492,
+      "step": 17667
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 1.0629453681710213e-06,
+      "loss": 0.025,
+      "step": 17668
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 1.0570071258907364e-06,
+      "loss": 0.0455,
+      "step": 17669
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 1.0510688836104512e-06,
+      "loss": 0.0216,
+      "step": 17670
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 1.0451306413301663e-06,
+      "loss": 0.0218,
+      "step": 17671
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 1.0391923990498813e-06,
+      "loss": 0.033,
+      "step": 17672
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 1.0332541567695962e-06,
+      "loss": 0.0433,
+      "step": 17673
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 1.0273159144893113e-06,
+      "loss": 0.0306,
+      "step": 17674
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 1.0213776722090261e-06,
+      "loss": 0.0453,
+      "step": 17675
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 1.0154394299287412e-06,
+      "loss": 0.021,
+      "step": 17676
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 1.0095011876484562e-06,
+      "loss": 0.0109,
+      "step": 17677
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 1.003562945368171e-06,
+      "loss": 0.0204,
+      "step": 17678
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 9.97624703087886e-07,
+      "loss": 0.0275,
+      "step": 17679
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 9.91686460807601e-07,
+      "loss": 0.025,
+      "step": 17680
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 9.857482185273159e-07,
+      "loss": 0.0159,
+      "step": 17681
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 9.79809976247031e-07,
+      "loss": 0.0166,
+      "step": 17682
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 9.73871733966746e-07,
+      "loss": 0.0175,
+      "step": 17683
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 9.679334916864608e-07,
+      "loss": 0.0225,
+      "step": 17684
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 9.619952494061759e-07,
+      "loss": 0.0224,
+      "step": 17685
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 9.560570071258907e-07,
+      "loss": 0.0371,
+      "step": 17686
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 9.501187648456058e-07,
+      "loss": 0.0184,
+      "step": 17687
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 9.441805225653208e-07,
+      "loss": 0.0205,
+      "step": 17688
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 9.382422802850357e-07,
+      "loss": 0.0353,
+      "step": 17689
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 9.323040380047506e-07,
+      "loss": 0.0313,
+      "step": 17690
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 9.263657957244655e-07,
+      "loss": 0.0104,
+      "step": 17691
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 9.204275534441805e-07,
+      "loss": 0.0528,
+      "step": 17692
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 9.144893111638954e-07,
+      "loss": 0.0311,
+      "step": 17693
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 9.085510688836105e-07,
+      "loss": 0.03,
+      "step": 17694
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 9.026128266033255e-07,
+      "loss": 0.0048,
+      "step": 17695
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 8.966745843230404e-07,
+      "loss": 0.0254,
+      "step": 17696
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 8.907363420427554e-07,
+      "loss": 0.0435,
+      "step": 17697
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 8.847980997624703e-07,
+      "loss": 0.0095,
+      "step": 17698
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 8.788598574821854e-07,
+      "loss": 0.0589,
+      "step": 17699
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 8.729216152019004e-07,
+      "loss": 0.0251,
+      "step": 17700
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 8.669833729216153e-07,
+      "loss": 0.0389,
+      "step": 17701
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 8.610451306413302e-07,
+      "loss": 0.0681,
+      "step": 17702
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 8.551068883610451e-07,
+      "loss": 0.0175,
+      "step": 17703
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 8.491686460807601e-07,
+      "loss": 0.0323,
+      "step": 17704
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 8.43230403800475e-07,
+      "loss": 0.018,
+      "step": 17705
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 8.3729216152019e-07,
+      "loss": 0.0158,
+      "step": 17706
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 8.313539192399051e-07,
+      "loss": 0.0252,
+      "step": 17707
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 8.2541567695962e-07,
+      "loss": 0.0319,
+      "step": 17708
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 8.19477434679335e-07,
+      "loss": 0.0523,
+      "step": 17709
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 8.135391923990499e-07,
+      "loss": 0.0299,
+      "step": 17710
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 8.07600950118765e-07,
+      "loss": 0.0278,
+      "step": 17711
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 8.016627078384799e-07,
+      "loss": 0.0219,
+      "step": 17712
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 7.957244655581949e-07,
+      "loss": 0.0242,
+      "step": 17713
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 7.897862232779098e-07,
+      "loss": 0.0296,
+      "step": 17714
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 7.838479809976247e-07,
+      "loss": 0.0453,
+      "step": 17715
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 7.779097387173398e-07,
+      "loss": 0.0148,
+      "step": 17716
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 7.719714964370546e-07,
+      "loss": 0.0282,
+      "step": 17717
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 7.660332541567696e-07,
+      "loss": 0.0217,
+      "step": 17718
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 7.600950118764846e-07,
+      "loss": 0.0322,
+      "step": 17719
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 7.541567695961996e-07,
+      "loss": 0.0354,
+      "step": 17720
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 7.482185273159146e-07,
+      "loss": 0.0124,
+      "step": 17721
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 7.422802850356294e-07,
+      "loss": 0.0411,
+      "step": 17722
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 7.363420427553445e-07,
+      "loss": 0.0488,
+      "step": 17723
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 7.304038004750594e-07,
+      "loss": 0.024,
+      "step": 17724
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 7.244655581947744e-07,
+      "loss": 0.0474,
+      "step": 17725
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 7.185273159144893e-07,
+      "loss": 0.019,
+      "step": 17726
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 7.125890736342043e-07,
+      "loss": 0.0378,
+      "step": 17727
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 7.066508313539193e-07,
+      "loss": 0.0236,
+      "step": 17728
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 7.007125890736342e-07,
+      "loss": 0.0343,
+      "step": 17729
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 6.947743467933492e-07,
+      "loss": 0.026,
+      "step": 17730
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 6.888361045130641e-07,
+      "loss": 0.0333,
+      "step": 17731
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 6.828978622327792e-07,
+      "loss": 0.0241,
+      "step": 17732
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 6.769596199524942e-07,
+      "loss": 0.0231,
+      "step": 17733
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 6.71021377672209e-07,
+      "loss": 0.022,
+      "step": 17734
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 6.65083135391924e-07,
+      "loss": 0.024,
+      "step": 17735
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 6.59144893111639e-07,
+      "loss": 0.0218,
+      "step": 17736
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 6.53206650831354e-07,
+      "loss": 0.0362,
+      "step": 17737
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 6.472684085510689e-07,
+      "loss": 0.0199,
+      "step": 17738
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 6.413301662707839e-07,
+      "loss": 0.0214,
+      "step": 17739
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 6.353919239904988e-07,
+      "loss": 0.0165,
+      "step": 17740
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 6.294536817102138e-07,
+      "loss": 0.0153,
+      "step": 17741
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 6.235154394299288e-07,
+      "loss": 0.0402,
+      "step": 17742
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 6.175771971496437e-07,
+      "loss": 0.0243,
+      "step": 17743
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 6.116389548693587e-07,
+      "loss": 0.0402,
+      "step": 17744
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 6.057007125890737e-07,
+      "loss": 0.0181,
+      "step": 17745
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 5.997624703087886e-07,
+      "loss": 0.0309,
+      "step": 17746
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 5.938242280285035e-07,
+      "loss": 0.0339,
+      "step": 17747
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 5.878859857482185e-07,
+      "loss": 0.0355,
+      "step": 17748
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 5.819477434679336e-07,
+      "loss": 0.0523,
+      "step": 17749
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 5.760095011876485e-07,
+      "loss": 0.0293,
+      "step": 17750
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 5.700712589073635e-07,
+      "loss": 0.0216,
+      "step": 17751
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 5.641330166270783e-07,
+      "loss": 0.0253,
+      "step": 17752
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 5.581947743467934e-07,
+      "loss": 0.0283,
+      "step": 17753
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 5.522565320665084e-07,
+      "loss": 0.0644,
+      "step": 17754
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 5.463182897862233e-07,
+      "loss": 0.0193,
+      "step": 17755
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 5.403800475059383e-07,
+      "loss": 0.0366,
+      "step": 17756
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 5.344418052256532e-07,
+      "loss": 0.046,
+      "step": 17757
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 5.285035629453682e-07,
+      "loss": 0.029,
+      "step": 17758
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 5.225653206650831e-07,
+      "loss": 0.0345,
+      "step": 17759
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 5.166270783847981e-07,
+      "loss": 0.0175,
+      "step": 17760
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 5.106888361045131e-07,
+      "loss": 0.0313,
+      "step": 17761
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 5.047505938242281e-07,
+      "loss": 0.0208,
+      "step": 17762
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 4.98812351543943e-07,
+      "loss": 0.0372,
+      "step": 17763
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 4.928741092636579e-07,
+      "loss": 0.0244,
+      "step": 17764
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 4.86935866983373e-07,
+      "loss": 0.0528,
+      "step": 17765
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 4.809976247030879e-07,
+      "loss": 0.0389,
+      "step": 17766
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 4.750593824228029e-07,
+      "loss": 0.0244,
+      "step": 17767
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 4.6912114014251786e-07,
+      "loss": 0.055,
+      "step": 17768
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 4.6318289786223276e-07,
+      "loss": 0.023,
+      "step": 17769
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 4.572446555819477e-07,
+      "loss": 0.0386,
+      "step": 17770
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 4.5130641330166273e-07,
+      "loss": 0.0252,
+      "step": 17771
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 4.453681710213777e-07,
+      "loss": 0.0167,
+      "step": 17772
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 4.394299287410927e-07,
+      "loss": 0.0349,
+      "step": 17773
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 4.3349168646080766e-07,
+      "loss": 0.0436,
+      "step": 17774
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 4.2755344418052256e-07,
+      "loss": 0.0254,
+      "step": 17775
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 4.216152019002375e-07,
+      "loss": 0.0186,
+      "step": 17776
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 4.1567695961995253e-07,
+      "loss": 0.0321,
+      "step": 17777
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 4.097387173396675e-07,
+      "loss": 0.0201,
+      "step": 17778
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 4.038004750593825e-07,
+      "loss": 0.0181,
+      "step": 17779
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 3.9786223277909745e-07,
+      "loss": 0.025,
+      "step": 17780
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 3.9192399049881236e-07,
+      "loss": 0.0257,
+      "step": 17781
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 3.859857482185273e-07,
+      "loss": 0.0307,
+      "step": 17782
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 3.800475059382423e-07,
+      "loss": 0.0248,
+      "step": 17783
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 3.741092636579573e-07,
+      "loss": 0.0225,
+      "step": 17784
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 3.6817102137767224e-07,
+      "loss": 0.0295,
+      "step": 17785
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 3.622327790973872e-07,
+      "loss": 0.0225,
+      "step": 17786
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 3.5629453681710215e-07,
+      "loss": 0.0201,
+      "step": 17787
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 3.503562945368171e-07,
+      "loss": 0.0347,
+      "step": 17788
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 3.4441805225653207e-07,
+      "loss": 0.0203,
+      "step": 17789
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 3.384798099762471e-07,
+      "loss": 0.0644,
+      "step": 17790
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 3.32541567695962e-07,
+      "loss": 0.0156,
+      "step": 17791
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 3.26603325415677e-07,
+      "loss": 0.0203,
+      "step": 17792
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 3.2066508313539195e-07,
+      "loss": 0.0156,
+      "step": 17793
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 3.147268408551069e-07,
+      "loss": 0.0237,
+      "step": 17794
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 3.0878859857482186e-07,
+      "loss": 0.0195,
+      "step": 17795
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 3.0285035629453687e-07,
+      "loss": 0.0258,
+      "step": 17796
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.969121140142518e-07,
+      "loss": 0.0128,
+      "step": 17797
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.909738717339668e-07,
+      "loss": 0.0054,
+      "step": 17798
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.8503562945368174e-07,
+      "loss": 0.0313,
+      "step": 17799
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.790973871733967e-07,
+      "loss": 0.0349,
+      "step": 17800
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.7315914489311166e-07,
+      "loss": 0.046,
+      "step": 17801
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.672209026128266e-07,
+      "loss": 0.0263,
+      "step": 17802
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.6128266033254157e-07,
+      "loss": 0.0377,
+      "step": 17803
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.5534441805225653e-07,
+      "loss": 0.0257,
+      "step": 17804
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.494061757719715e-07,
+      "loss": 0.0522,
+      "step": 17805
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.434679334916865e-07,
+      "loss": 0.0555,
+      "step": 17806
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.3752969121140145e-07,
+      "loss": 0.0289,
+      "step": 17807
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.3159144893111638e-07,
+      "loss": 0.0389,
+      "step": 17808
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.2565320665083137e-07,
+      "loss": 0.0326,
+      "step": 17809
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.1971496437054635e-07,
+      "loss": 0.0247,
+      "step": 17810
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.1377672209026128e-07,
+      "loss": 0.0242,
+      "step": 17811
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.0783847980997626e-07,
+      "loss": 0.0189,
+      "step": 17812
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.0190023752969125e-07,
+      "loss": 0.0207,
+      "step": 17813
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 1.9596199524940618e-07,
+      "loss": 0.0465,
+      "step": 17814
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 1.9002375296912116e-07,
+      "loss": 0.0323,
+      "step": 17815
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 1.8408551068883612e-07,
+      "loss": 0.0312,
+      "step": 17816
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 1.7814726840855108e-07,
+      "loss": 0.0309,
+      "step": 17817
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 1.7220902612826603e-07,
+      "loss": 0.0396,
+      "step": 17818
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 1.66270783847981e-07,
+      "loss": 0.0102,
+      "step": 17819
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 1.6033254156769597e-07,
+      "loss": 0.0264,
+      "step": 17820
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 1.5439429928741093e-07,
+      "loss": 0.022,
+      "step": 17821
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 1.484560570071259e-07,
+      "loss": 0.0329,
+      "step": 17822
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 1.4251781472684087e-07,
+      "loss": 0.0154,
+      "step": 17823
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 1.3657957244655583e-07,
+      "loss": 0.0237,
+      "step": 17824
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 1.3064133016627079e-07,
+      "loss": 0.0274,
+      "step": 17825
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 1.2470308788598574e-07,
+      "loss": 0.029,
+      "step": 17826
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 1.1876484560570073e-07,
+      "loss": 0.0115,
+      "step": 17827
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 1.1282660332541568e-07,
+      "loss": 0.0268,
+      "step": 17828
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 1.0688836104513064e-07,
+      "loss": 0.0181,
+      "step": 17829
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 1.0095011876484562e-07,
+      "loss": 0.0164,
+      "step": 17830
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 9.501187648456058e-08,
+      "loss": 0.0204,
+      "step": 17831
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 8.907363420427554e-08,
+      "loss": 0.0461,
+      "step": 17832
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 8.31353919239905e-08,
+      "loss": 0.0205,
+      "step": 17833
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 7.719714964370547e-08,
+      "loss": 0.0196,
+      "step": 17834
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 7.125890736342044e-08,
+      "loss": 0.0293,
+      "step": 17835
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 6.532066508313539e-08,
+      "loss": 0.0215,
+      "step": 17836
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 5.938242280285036e-08,
+      "loss": 0.0294,
+      "step": 17837
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 5.344418052256532e-08,
+      "loss": 0.0086,
+      "step": 17838
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 4.750593824228029e-08,
+      "loss": 0.0246,
+      "step": 17839
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 4.156769596199525e-08,
+      "loss": 0.0608,
+      "step": 17840
+    },
+    {
+      "epoch": 20.0,
+      "step": 17840,
+      "total_flos": 0.0,
+      "train_loss": 0.9826719127995244,
+      "train_runtime": 114528.9287,
+      "train_samples_per_second": 4.984,
+      "train_steps_per_second": 0.156
+    }
+  ],
+  "max_steps": 17840,
+  "num_train_epochs": 20,
+  "total_flos": 0.0,
+  "trial_name": null,
+  "trial_params": null
+}