diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,50257 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 8372,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-08,
+      "loss": 3.8985,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-08,
+      "loss": 3.9676,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 3.8037,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-08,
+      "loss": 3.8795,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 3.7558,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 3.7668,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4e-07,
+      "loss": 3.7264,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6e-07,
+      "loss": 3.8607,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-07,
+      "loss": 3.9243,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 3.9088,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2e-07,
+      "loss": 3.9361,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 3.9018,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6e-07,
+      "loss": 3.809,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8e-07,
+      "loss": 3.8588,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 3.8557,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2e-07,
+      "loss": 3.9139,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 3.8195,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-07,
+      "loss": 3.733,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8e-07,
+      "loss": 3.8347,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 3.8319,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 3.8035,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4e-07,
+      "loss": 3.771,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 3.8331,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 3.7964,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 3.7758,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2e-07,
+      "loss": 3.8268,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4e-07,
+      "loss": 3.6804,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.6e-07,
+      "loss": 3.7971,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 3.7038,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 3.8058,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 3.7717,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.4e-07,
+      "loss": 3.7852,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.6e-07,
+      "loss": 3.913,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 3.6964,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 3.8357,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2e-07,
+      "loss": 3.7321,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.4e-07,
+      "loss": 3.7974,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.6e-07,
+      "loss": 3.8722,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.8e-07,
+      "loss": 3.8006,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 3.7595,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 3.7296,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 3.7808,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.6e-07,
+      "loss": 3.704,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.8e-07,
+      "loss": 3.7205,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 3.7701,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 3.8456,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 3.7506,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 3.7092,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 3.7437,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 3.842,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.02e-06,
+      "loss": 3.8106,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.04e-06,
+      "loss": 3.7872,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.06e-06,
+      "loss": 3.7275,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.08e-06,
+      "loss": 3.7983,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1e-06,
+      "loss": 3.6847,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.12e-06,
+      "loss": 3.8405,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.14e-06,
+      "loss": 3.7125,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 3.7554,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 3.6508,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 3.723,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 3.7775,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 3.6425,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.26e-06,
+      "loss": 3.7886,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.28e-06,
+      "loss": 3.7835,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3e-06,
+      "loss": 3.7377,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.32e-06,
+      "loss": 3.7013,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.34e-06,
+      "loss": 3.6558,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 3.805,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 3.658,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 3.7237,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.42e-06,
+      "loss": 3.6212,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.44e-06,
+      "loss": 3.7804,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.46e-06,
+      "loss": 3.6221,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.48e-06,
+      "loss": 3.7157,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-06,
+      "loss": 3.5707,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.52e-06,
+      "loss": 3.714,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.54e-06,
+      "loss": 3.73,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.56e-06,
+      "loss": 3.698,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 3.7339,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 3.7057,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 3.731,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 3.5928,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 3.7794,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 3.7305,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 3.6609,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.72e-06,
+      "loss": 3.6755,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.74e-06,
+      "loss": 3.5747,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.76e-06,
+      "loss": 3.8097,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 3.746,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 3.708,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 3.647,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 3.692,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8600000000000002e-06,
+      "loss": 3.6386,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 3.7464,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 3.7239,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 3.6872,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.94e-06,
+      "loss": 3.5843,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 3.6359,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.98e-06,
+      "loss": 3.5997,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 3.6303,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.02e-06,
+      "loss": 3.6339,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.04e-06,
+      "loss": 3.6606,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.06e-06,
+      "loss": 3.7024,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.08e-06,
+      "loss": 3.7978,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 3.4576,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.12e-06,
+      "loss": 3.6358,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1400000000000003e-06,
+      "loss": 3.6367,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.16e-06,
+      "loss": 3.7169,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 3.582,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2e-06,
+      "loss": 3.6598,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 3.6929,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.24e-06,
+      "loss": 3.7283,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2600000000000004e-06,
+      "loss": 3.6273,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.28e-06,
+      "loss": 3.6746,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 3.7009,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 3.5666,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 3.6521,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 3.6093,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.38e-06,
+      "loss": 3.6252,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 3.6445,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.42e-06,
+      "loss": 3.6903,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 3.6604,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.46e-06,
+      "loss": 3.5256,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 3.6729,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-06,
+      "loss": 3.7525,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.52e-06,
+      "loss": 3.673,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5400000000000002e-06,
+      "loss": 3.5944,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.56e-06,
+      "loss": 3.5687,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 3.5883,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6e-06,
+      "loss": 3.5206,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 3.5704,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.64e-06,
+      "loss": 3.5294,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6600000000000004e-06,
+      "loss": 3.4963,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.68e-06,
+      "loss": 3.6174,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 3.5491,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 3.648,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7400000000000004e-06,
+      "loss": 3.6226,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 3.6751,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 3.5936,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 3.5267,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.82e-06,
+      "loss": 3.5474,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.84e-06,
+      "loss": 3.6663,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.86e-06,
+      "loss": 3.6454,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.88e-06,
+      "loss": 3.6454,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9e-06,
+      "loss": 3.6188,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.92e-06,
+      "loss": 3.6545,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 3.5483,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.96e-06,
+      "loss": 3.6255,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 3.6774,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3e-06,
+      "loss": 3.6898,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 3.5297,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.04e-06,
+      "loss": 3.5732,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0600000000000003e-06,
+      "loss": 3.5196,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.08e-06,
+      "loss": 3.7004,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 3.4863,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.12e-06,
+      "loss": 3.5595,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1400000000000004e-06,
+      "loss": 3.6466,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 3.6071,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 3.6044,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 3.524,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2200000000000005e-06,
+      "loss": 3.5998,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 3.7073,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2600000000000006e-06,
+      "loss": 3.5558,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 3.5562,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 3.5048,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 3.5298,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 3.5429,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 3.5596,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 3.6291,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 3.6635,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 3.7003,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.44e-06,
+      "loss": 3.4481,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.46e-06,
+      "loss": 3.5758,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.48e-06,
+      "loss": 3.6038,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5e-06,
+      "loss": 3.5105,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.52e-06,
+      "loss": 3.631,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.54e-06,
+      "loss": 3.6639,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 3.6319,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.58e-06,
+      "loss": 3.6372,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 3.5545,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.62e-06,
+      "loss": 3.5509,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 3.4724,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.66e-06,
+      "loss": 3.4752,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 3.5841,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7e-06,
+      "loss": 3.5091,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 3.5835,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.74e-06,
+      "loss": 3.5392,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 3.562,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 3.6836,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 3.679,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 3.6514,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 3.6153,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.86e-06,
+      "loss": 3.5517,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.88e-06,
+      "loss": 3.4983,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 3.6972,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 3.6154,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.94e-06,
+      "loss": 3.5469,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.96e-06,
+      "loss": 3.6211,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 3.6066,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 3.534,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 3.5634,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.04e-06,
+      "loss": 3.6281,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.060000000000001e-06,
+      "loss": 3.6215,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.08e-06,
+      "loss": 3.522,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1e-06,
+      "loss": 3.5076,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.12e-06,
+      "loss": 3.5728,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.14e-06,
+      "loss": 3.559,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.16e-06,
+      "loss": 3.5392,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.18e-06,
+      "loss": 3.5208,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 3.5061,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.22e-06,
+      "loss": 3.5039,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.24e-06,
+      "loss": 3.5804,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.26e-06,
+      "loss": 3.5836,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 3.4476,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3e-06,
+      "loss": 3.627,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.32e-06,
+      "loss": 3.5898,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.34e-06,
+      "loss": 3.6349,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 3.6543,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.38e-06,
+      "loss": 3.5136,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4e-06,
+      "loss": 3.5482,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.42e-06,
+      "loss": 3.621,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 3.6233,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 3.5274,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.48e-06,
+      "loss": 3.4743,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5e-06,
+      "loss": 3.4795,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 3.5258,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 3.6325,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.56e-06,
+      "loss": 3.4002,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.58e-06,
+      "loss": 3.5674,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 3.4766,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 3.4985,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 3.5495,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.66e-06,
+      "loss": 3.5232,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 3.5032,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7e-06,
+      "loss": 3.5169,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 3.5932,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.74e-06,
+      "loss": 3.7029,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.76e-06,
+      "loss": 3.6386,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.78e-06,
+      "loss": 3.6061,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 3.556,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 3.5557,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.84e-06,
+      "loss": 3.5023,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.86e-06,
+      "loss": 3.5929,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 3.6341,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 3.5248,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.92e-06,
+      "loss": 3.5861,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.94e-06,
+      "loss": 3.5323,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 3.5684,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 3.4539,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5e-06,
+      "loss": 3.5653,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.02e-06,
+      "loss": 3.5114,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.04e-06,
+      "loss": 3.6342,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.060000000000001e-06,
+      "loss": 3.5279,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 3.4932,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1e-06,
+      "loss": 3.5684,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.12e-06,
+      "loss": 3.5673,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.140000000000001e-06,
+      "loss": 3.4347,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1600000000000006e-06,
+      "loss": 3.5512,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.18e-06,
+      "loss": 3.5261,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.2e-06,
+      "loss": 3.5376,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.220000000000001e-06,
+      "loss": 3.6135,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 3.6282,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.2600000000000005e-06,
+      "loss": 3.5554,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.28e-06,
+      "loss": 3.5284,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 3.5936,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.320000000000001e-06,
+      "loss": 3.616,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.3400000000000005e-06,
+      "loss": 3.5766,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.36e-06,
+      "loss": 3.5787,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 3.5456,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 3.4992,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.420000000000001e-06,
+      "loss": 3.6263,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 3.452,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.460000000000001e-06,
+      "loss": 3.5558,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 3.5357,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 3.5351,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 3.3575,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.540000000000001e-06,
+      "loss": 3.4974,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.560000000000001e-06,
+      "loss": 3.4849,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 3.4188,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 3.664,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.620000000000001e-06,
+      "loss": 3.6015,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.64e-06,
+      "loss": 3.5164,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.66e-06,
+      "loss": 3.5891,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.68e-06,
+      "loss": 3.5771,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7e-06,
+      "loss": 3.5047,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.72e-06,
+      "loss": 3.565,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.74e-06,
+      "loss": 3.5207,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.76e-06,
+      "loss": 3.5712,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.78e-06,
+      "loss": 3.5512,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.8e-06,
+      "loss": 3.5396,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.82e-06,
+      "loss": 3.4562,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.84e-06,
+      "loss": 3.5835,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.86e-06,
+      "loss": 3.4174,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 3.5173,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9e-06,
+      "loss": 3.5081,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.92e-06,
+      "loss": 3.4491,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.94e-06,
+      "loss": 3.5805,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9600000000000005e-06,
+      "loss": 3.5627,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.98e-06,
+      "loss": 3.5774,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6e-06,
+      "loss": 3.5363,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.02e-06,
+      "loss": 3.5466,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.040000000000001e-06,
+      "loss": 3.6104,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 3.5467,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.08e-06,
+      "loss": 3.52,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.1e-06,
+      "loss": 3.5324,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.120000000000001e-06,
+      "loss": 3.6317,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.1400000000000005e-06,
+      "loss": 3.5573,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.16e-06,
+      "loss": 3.441,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.18e-06,
+      "loss": 3.5131,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 3.4777,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.220000000000001e-06,
+      "loss": 3.5242,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.24e-06,
+      "loss": 3.4621,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.26e-06,
+      "loss": 3.3419,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 3.5037,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 3.4321,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 3.5762,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.34e-06,
+      "loss": 3.57,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.360000000000001e-06,
+      "loss": 3.5336,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 3.6409,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 3.5235,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.42e-06,
+      "loss": 3.6526,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.440000000000001e-06,
+      "loss": 3.5293,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.460000000000001e-06,
+      "loss": 3.4952,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 3.515,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 3.4361,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.520000000000001e-06,
+      "loss": 3.415,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.540000000000001e-06,
+      "loss": 3.5689,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 3.5456,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 3.4916,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 3.5184,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.620000000000001e-06,
+      "loss": 3.5307,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 3.5578,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 3.5547,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 3.4802,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 3.4977,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 3.4796,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 3.4703,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.760000000000001e-06,
+      "loss": 3.4915,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 3.4597,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 3.6189,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.820000000000001e-06,
+      "loss": 3.5419,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.8400000000000014e-06,
+      "loss": 3.5287,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.860000000000001e-06,
+      "loss": 3.5254,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.88e-06,
+      "loss": 3.6214,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.9e-06,
+      "loss": 3.5982,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.92e-06,
+      "loss": 3.5468,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 3.5347,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.96e-06,
+      "loss": 3.4927,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.98e-06,
+      "loss": 3.4982,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7e-06,
+      "loss": 3.57,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 3.5014,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.04e-06,
+      "loss": 3.6661,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.06e-06,
+      "loss": 3.5723,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.08e-06,
+      "loss": 3.6442,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 3.4915,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 3.5111,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.14e-06,
+      "loss": 3.5004,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.16e-06,
+      "loss": 3.469,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 3.486,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 3.545,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.22e-06,
+      "loss": 3.4236,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.24e-06,
+      "loss": 3.3786,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.260000000000001e-06,
+      "loss": 3.3814,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 3.4316,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.3e-06,
+      "loss": 3.6023,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.32e-06,
+      "loss": 3.4872,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.340000000000001e-06,
+      "loss": 3.4602,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 3.468,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 3.4891,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.4e-06,
+      "loss": 3.615,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.420000000000001e-06,
+      "loss": 3.4123,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.440000000000001e-06,
+      "loss": 3.4424,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.4600000000000006e-06,
+      "loss": 3.5211,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.48e-06,
+      "loss": 3.5022,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 3.4235,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 3.5374,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.540000000000001e-06,
+      "loss": 3.5203,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 3.4676,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.58e-06,
+      "loss": 3.5289,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 3.5788,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.620000000000001e-06,
+      "loss": 3.5367,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.640000000000001e-06,
+      "loss": 3.655,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.660000000000001e-06,
+      "loss": 3.5317,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 3.4517,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.7e-06,
+      "loss": 3.5172,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.72e-06,
+      "loss": 3.509,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.74e-06,
+      "loss": 3.4323,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.76e-06,
+      "loss": 3.486,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.78e-06,
+      "loss": 3.4063,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 3.4376,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.820000000000001e-06,
+      "loss": 3.49,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 3.4926,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 3.4904,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.88e-06,
+      "loss": 3.5267,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.9e-06,
+      "loss": 3.4835,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.92e-06,
+      "loss": 3.5475,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.94e-06,
+      "loss": 3.5048,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.960000000000002e-06,
+      "loss": 3.5826,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 3.4729,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 3.4931,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.020000000000001e-06,
+      "loss": 3.4976,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 3.4994,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.06e-06,
+      "loss": 3.5024,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.08e-06,
+      "loss": 3.5663,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.1e-06,
+      "loss": 3.5492,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.120000000000002e-06,
+      "loss": 3.4954,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.14e-06,
+      "loss": 3.5102,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.16e-06,
+      "loss": 3.539,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.18e-06,
+      "loss": 3.5008,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.2e-06,
+      "loss": 3.5204,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 3.5115,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.24e-06,
+      "loss": 3.3583,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.26e-06,
+      "loss": 3.519,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.28e-06,
+      "loss": 3.5653,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.3e-06,
+      "loss": 3.6157,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.32e-06,
+      "loss": 3.5141,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.34e-06,
+      "loss": 3.3275,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.36e-06,
+      "loss": 3.3692,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 3.5056,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 3.6151,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.42e-06,
+      "loss": 3.5976,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.44e-06,
+      "loss": 3.6185,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.46e-06,
+      "loss": 3.4193,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.48e-06,
+      "loss": 3.5691,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.5e-06,
+      "loss": 3.5386,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.52e-06,
+      "loss": 3.5204,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.540000000000001e-06,
+      "loss": 3.5085,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.560000000000001e-06,
+      "loss": 3.5655,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 3.4862,
+      "step": 429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.6e-06,
+      "loss": 3.6018,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.62e-06,
+      "loss": 3.4218,
+      "step": 431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.64e-06,
+      "loss": 3.4102,
+      "step": 432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.66e-06,
+      "loss": 3.4803,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.68e-06,
+      "loss": 3.4744,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 3.4974,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 3.6238,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.740000000000001e-06,
+      "loss": 3.4661,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.76e-06,
+      "loss": 3.4542,
+      "step": 438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.78e-06,
+      "loss": 3.5468,
+      "step": 439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.8e-06,
+      "loss": 3.5363,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.82e-06,
+      "loss": 3.5258,
+      "step": 441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.84e-06,
+      "loss": 3.537,
+      "step": 442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.860000000000002e-06,
+      "loss": 3.4792,
+      "step": 443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 3.4594,
+      "step": 444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 3.4785,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.920000000000001e-06,
+      "loss": 3.5554,
+      "step": 446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.94e-06,
+      "loss": 3.4309,
+      "step": 447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.96e-06,
+      "loss": 3.3718,
+      "step": 448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.98e-06,
+      "loss": 3.5266,
+      "step": 449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9e-06,
+      "loss": 3.4291,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.020000000000002e-06,
+      "loss": 3.3068,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.040000000000002e-06,
+      "loss": 3.4432,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.060000000000001e-06,
+      "loss": 3.6156,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 3.4309,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 3.6563,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.12e-06,
+      "loss": 3.5165,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.14e-06,
+      "loss": 3.5458,
+      "step": 457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.16e-06,
+      "loss": 3.4744,
+      "step": 458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 3.4925,
+      "step": 459
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 3.4804,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.220000000000002e-06,
+      "loss": 3.5697,
+      "step": 461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.240000000000001e-06,
+      "loss": 3.4594,
+      "step": 462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.260000000000001e-06,
+      "loss": 3.4538,
+      "step": 463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 3.5725,
+      "step": 464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.3e-06,
+      "loss": 3.4056,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.32e-06,
+      "loss": 3.4452,
+      "step": 466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.340000000000002e-06,
+      "loss": 3.5996,
+      "step": 467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.360000000000002e-06,
+      "loss": 3.4671,
+      "step": 468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.38e-06,
+      "loss": 3.5151,
+      "step": 469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.4e-06,
+      "loss": 3.5539,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.42e-06,
+      "loss": 3.4483,
+      "step": 471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 3.4765,
+      "step": 472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.460000000000001e-06,
+      "loss": 3.4619,
+      "step": 473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.48e-06,
+      "loss": 3.5627,
+      "step": 474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.5e-06,
+      "loss": 3.4597,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.52e-06,
+      "loss": 3.4291,
+      "step": 476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.54e-06,
+      "loss": 3.4539,
+      "step": 477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.56e-06,
+      "loss": 3.5258,
+      "step": 478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.58e-06,
+      "loss": 3.4553,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 3.6458,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.620000000000001e-06,
+      "loss": 3.545,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.640000000000001e-06,
+      "loss": 3.4684,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.66e-06,
+      "loss": 3.5551,
+      "step": 483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.68e-06,
+      "loss": 3.4593,
+      "step": 484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.7e-06,
+      "loss": 3.5263,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.72e-06,
+      "loss": 3.491,
+      "step": 486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.74e-06,
+      "loss": 3.5891,
+      "step": 487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 3.4571,
+      "step": 488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 3.5363,
+      "step": 489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 3.4951,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.820000000000001e-06,
+      "loss": 3.5253,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.84e-06,
+      "loss": 3.4522,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.86e-06,
+      "loss": 3.5278,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.88e-06,
+      "loss": 3.4898,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9e-06,
+      "loss": 3.4631,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 3.3929,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.940000000000001e-06,
+      "loss": 3.3835,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 3.5321,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 3.4383,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1e-05,
+      "loss": 3.3691,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999999601829045e-06,
+      "loss": 3.403,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999998407316242e-06,
+      "loss": 3.4934,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999996416461782e-06,
+      "loss": 3.4457,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999993629265979e-06,
+      "loss": 3.4934,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999990045729283e-06,
+      "loss": 3.5346,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999985665852257e-06,
+      "loss": 3.4897,
+      "step": 506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999980489635606e-06,
+      "loss": 3.5066,
+      "step": 507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99997451708015e-06,
+      "loss": 3.4283,
+      "step": 508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99996774818684e-06,
+      "loss": 3.3443,
+      "step": 509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999960182956756e-06,
+      "loss": 3.4162,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999951821391105e-06,
+      "loss": 3.5466,
+      "step": 511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999942663491213e-06,
+      "loss": 3.4522,
+      "step": 512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999932709258544e-06,
+      "loss": 3.4683,
+      "step": 513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999921958694681e-06,
+      "loss": 3.4994,
+      "step": 514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999910411801335e-06,
+      "loss": 3.5499,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999898068580346e-06,
+      "loss": 3.4535,
+      "step": 516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99988492903368e-06,
+      "loss": 3.5073,
+      "step": 517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999870993163432e-06,
+      "loss": 3.429,
+      "step": 518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999856260971819e-06,
+      "loss": 3.4385,
+      "step": 519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999840732461185e-06,
+      "loss": 3.4583,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999824407634009e-06,
+      "loss": 3.4919,
+      "step": 521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999807286492887e-06,
+      "loss": 3.3693,
+      "step": 522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999789369040548e-06,
+      "loss": 3.4413,
+      "step": 523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999770655279843e-06,
+      "loss": 3.5326,
+      "step": 524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999751145213756e-06,
+      "loss": 3.3845,
+      "step": 525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999730838845392e-06,
+      "loss": 3.4513,
+      "step": 526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999709736177984e-06,
+      "loss": 3.4944,
+      "step": 527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999687837214895e-06,
+      "loss": 3.5027,
+      "step": 528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999665141959613e-06,
+      "loss": 3.417,
+      "step": 529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999641650415752e-06,
+      "loss": 3.3995,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999617362587056e-06,
+      "loss": 3.4835,
+      "step": 531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999592278477389e-06,
+      "loss": 3.5213,
+      "step": 532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999566398090745e-06,
+      "loss": 3.5634,
+      "step": 533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999539721431253e-06,
+      "loss": 3.6516,
+      "step": 534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999512248503155e-06,
+      "loss": 3.5607,
+      "step": 535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99948397931083e-06,
+      "loss": 3.4803,
+      "step": 536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999454913858779e-06,
+      "loss": 3.4693,
+      "step": 537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999425052151632e-06,
+      "loss": 3.5681,
+      "step": 538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999394394194146e-06,
+      "loss": 3.5093,
+      "step": 539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999362939991202e-06,
+      "loss": 3.4402,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999330689547811e-06,
+      "loss": 3.4983,
+      "step": 541
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999297642869106e-06,
+      "loss": 3.5484,
+      "step": 542
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999263799960355e-06,
+      "loss": 3.4673,
+      "step": 543
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999229160826947e-06,
+      "loss": 3.5846,
+      "step": 544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999193725474397e-06,
+      "loss": 3.5872,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999157493908349e-06,
+      "loss": 3.4185,
+      "step": 546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999120466134576e-06,
+      "loss": 3.4032,
+      "step": 547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999082642158972e-06,
+      "loss": 3.5427,
+      "step": 548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999044021987565e-06,
+      "loss": 3.562,
+      "step": 549
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999004605626502e-06,
+      "loss": 3.4885,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998964393082063e-06,
+      "loss": 3.5862,
+      "step": 551
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998923384360652e-06,
+      "loss": 3.5551,
+      "step": 552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998881579468802e-06,
+      "loss": 3.4032,
+      "step": 553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998838978413167e-06,
+      "loss": 3.3925,
+      "step": 554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998795581200538e-06,
+      "loss": 3.5038,
+      "step": 555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998751387837822e-06,
+      "loss": 3.462,
+      "step": 556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998706398332059e-06,
+      "loss": 3.5276,
+      "step": 557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998660612690417e-06,
+      "loss": 3.5552,
+      "step": 558
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998614030920183e-06,
+      "loss": 3.4833,
+      "step": 559
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99856665302878e-06,
+      "loss": 3.4392,
+      "step": 560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998518479023752e-06,
+      "loss": 3.4441,
+      "step": 561
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998469508912773e-06,
+      "loss": 3.4959,
+      "step": 562
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998419742703641e-06,
+      "loss": 3.4155,
+      "step": 563
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998369180404283e-06,
+      "loss": 3.5507,
+      "step": 564
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998317822022751e-06,
+      "loss": 3.4317,
+      "step": 565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998265667567226e-06,
+      "loss": 3.3735,
+      "step": 566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998212717046016e-06,
+      "loss": 3.4943,
+      "step": 567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99815897046755e-06,
+      "loss": 3.4624,
+      "step": 568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998104427840391e-06,
+      "loss": 3.4127,
+      "step": 569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998049089173227e-06,
+      "loss": 3.3605,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997992954474868e-06,
+      "loss": 3.4433,
+      "step": 571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997936023754258e-06,
+      "loss": 3.3913,
+      "step": 572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997878297020463e-06,
+      "loss": 3.5302,
+      "step": 573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997819774282676e-06,
+      "loss": 3.6367,
+      "step": 574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99776045555022e-06,
+      "loss": 3.4267,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99770034083254e-06,
+      "loss": 3.4221,
+      "step": 576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997639430139211e-06,
+      "loss": 3.3825,
+      "step": 577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997577723479939e-06,
+      "loss": 3.4185,
+      "step": 578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997515220864544e-06,
+      "loss": 3.4563,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997451922302987e-06,
+      "loss": 3.4466,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997387827805345e-06,
+      "loss": 3.5341,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997322937381829e-06,
+      "loss": 3.3628,
+      "step": 582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997257251042773e-06,
+      "loss": 3.4577,
+      "step": 583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997190768798639e-06,
+      "loss": 3.4834,
+      "step": 584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997123490660017e-06,
+      "loss": 3.5546,
+      "step": 585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997055416637617e-06,
+      "loss": 3.4919,
+      "step": 586
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996986546742289e-06,
+      "loss": 3.495,
+      "step": 587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996916880984995e-06,
+      "loss": 3.4825,
+      "step": 588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996846419376834e-06,
+      "loss": 3.4733,
+      "step": 589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996775161929027e-06,
+      "loss": 3.5733,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996703108652925e-06,
+      "loss": 3.3902,
+      "step": 591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996630259560001e-06,
+      "loss": 3.5258,
+      "step": 592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996556614661859e-06,
+      "loss": 3.524,
+      "step": 593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996482173970227e-06,
+      "loss": 3.4219,
+      "step": 594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996406937496965e-06,
+      "loss": 3.3701,
+      "step": 595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99633090525405e-06,
+      "loss": 3.5468,
+      "step": 596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996254077253598e-06,
+      "loss": 3.465,
+      "step": 597
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996176453507839e-06,
+      "loss": 3.426,
+      "step": 598
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99609803402914e-06,
+      "loss": 3.5204,
+      "step": 599
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996018818829989e-06,
+      "loss": 3.6021,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995938807923003e-06,
+      "loss": 3.4634,
+      "step": 601
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995858001320926e-06,
+      "loss": 3.4763,
+      "step": 602
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995776399036626e-06,
+      "loss": 3.5127,
+      "step": 603
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995694001083103e-06,
+      "loss": 3.5584,
+      "step": 604
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995610807473477e-06,
+      "loss": 3.5345,
+      "step": 605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995526818220998e-06,
+      "loss": 3.5603,
+      "step": 606
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995442033339046e-06,
+      "loss": 3.5189,
+      "step": 607
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995356452841122e-06,
+      "loss": 3.4112,
+      "step": 608
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995270076740857e-06,
+      "loss": 3.5445,
+      "step": 609
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995182905052008e-06,
+      "loss": 3.438,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995094937788459e-06,
+      "loss": 3.4947,
+      "step": 611
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99500617496422e-06,
+      "loss": 3.5491,
+      "step": 612
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994916616593428e-06,
+      "loss": 3.5032,
+      "step": 613
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994826262690347e-06,
+      "loss": 3.4676,
+      "step": 614
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99473511326937e-06,
+      "loss": 3.4853,
+      "step": 615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994643168345008e-06,
+      "loss": 3.343,
+      "step": 616
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99455042793191e-06,
+      "loss": 3.4827,
+      "step": 617
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994456892044846e-06,
+      "loss": 3.5069,
+      "step": 618
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99436256069871e-06,
+      "loss": 3.6305,
+      "step": 619
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994267433908533e-06,
+      "loss": 3.5468,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994171511689458e-06,
+      "loss": 3.4006,
+      "step": 621
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994074794056768e-06,
+      "loss": 3.4833,
+      "step": 622
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993977281025862e-06,
+      "loss": 3.3752,
+      "step": 623
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993878972612276e-06,
+      "loss": 3.5774,
+      "step": 624
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993779868831664e-06,
+      "loss": 3.4774,
+      "step": 625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99367996969981e-06,
+      "loss": 3.5111,
+      "step": 626
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993579275232627e-06,
+      "loss": 3.4913,
+      "step": 627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993477785446151e-06,
+      "loss": 3.3716,
+      "step": 628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993375500356546e-06,
+      "loss": 3.4684,
+      "step": 629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993272419980102e-06,
+      "loss": 3.4812,
+      "step": 630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99316854433324e-06,
+      "loss": 3.4223,
+      "step": 631
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9930638734325e-06,
+      "loss": 3.4437,
+      "step": 632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992958407294555e-06,
+      "loss": 3.5395,
+      "step": 633
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992852145936202e-06,
+      "loss": 3.4157,
+      "step": 634
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992745089374364e-06,
+      "loss": 3.5799,
+      "step": 635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992637237626092e-06,
+      "loss": 3.4083,
+      "step": 636
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992528590708566e-06,
+      "loss": 3.4253,
+      "step": 637
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992419148639088e-06,
+      "loss": 3.3974,
+      "step": 638
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992308911435088e-06,
+      "loss": 3.3973,
+      "step": 639
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992197879114124e-06,
+      "loss": 3.4675,
+      "step": 640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99208605169388e-06,
+      "loss": 3.4957,
+      "step": 641
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991973429192167e-06,
+      "loss": 3.5175,
+      "step": 642
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991860011626922e-06,
+      "loss": 3.5201,
+      "step": 643
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991745799016206e-06,
+      "loss": 3.6455,
+      "step": 644
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991630791378215e-06,
+      "loss": 3.4612,
+      "step": 645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991514988731262e-06,
+      "loss": 3.4791,
+      "step": 646
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99139839109379e-06,
+      "loss": 3.3964,
+      "step": 647
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991280998484374e-06,
+      "loss": 3.5192,
+      "step": 648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991162810921706e-06,
+      "loss": 3.5147,
+      "step": 649
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991043828424613e-06,
+      "loss": 3.4596,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990924051012043e-06,
+      "loss": 3.4086,
+      "step": 651
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990803478703073e-06,
+      "loss": 3.3938,
+      "step": 652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990682111516907e-06,
+      "loss": 3.4969,
+      "step": 653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990559949472875e-06,
+      "loss": 3.5307,
+      "step": 654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990436992590435e-06,
+      "loss": 3.5391,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990313240889167e-06,
+      "loss": 3.4427,
+      "step": 656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990188694388781e-06,
+      "loss": 3.6398,
+      "step": 657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990063353109116e-06,
+      "loss": 3.4755,
+      "step": 658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989937217070133e-06,
+      "loss": 3.3037,
+      "step": 659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989810286291923e-06,
+      "loss": 3.451,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9896825607947e-06,
+      "loss": 3.371,
+      "step": 661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989554040598808e-06,
+      "loss": 3.5279,
+      "step": 662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989424725724714e-06,
+      "loss": 3.4114,
+      "step": 663
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989294616193018e-06,
+      "loss": 3.4855,
+      "step": 664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989163712024438e-06,
+      "loss": 3.4774,
+      "step": 665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989032013239828e-06,
+      "loss": 3.557,
+      "step": 666
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988899519860158e-06,
+      "loss": 3.597,
+      "step": 667
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988766231906532e-06,
+      "loss": 3.5646,
+      "step": 668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98863214940018e-06,
+      "loss": 3.407,
+      "step": 669
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988497272362454e-06,
+      "loss": 3.4524,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98836160081484e-06,
+      "loss": 3.3834,
+      "step": 671
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988225134778941e-06,
+      "loss": 3.4522,
+      "step": 672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988087874276498e-06,
+      "loss": 3.5605,
+      "step": 673
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987949819329366e-06,
+      "loss": 3.5665,
+      "step": 674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987810969959537e-06,
+      "loss": 3.4703,
+      "step": 675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987671326189123e-06,
+      "loss": 3.3722,
+      "step": 676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987530888040366e-06,
+      "loss": 3.5142,
+      "step": 677
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987389655535634e-06,
+      "loss": 3.4832,
+      "step": 678
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987247628697418e-06,
+      "loss": 3.4163,
+      "step": 679
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987104807548341e-06,
+      "loss": 3.5198,
+      "step": 680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98696119211115e-06,
+      "loss": 3.462,
+      "step": 681
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986816782408717e-06,
+      "loss": 3.414,
+      "step": 682
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986671578464042e-06,
+      "loss": 3.5161,
+      "step": 683
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986525580300253e-06,
+      "loss": 3.5292,
+      "step": 684
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986378787940601e-06,
+      "loss": 3.5184,
+      "step": 685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986231201408467e-06,
+      "loss": 3.4149,
+      "step": 686
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986082820727356e-06,
+      "loss": 3.4874,
+      "step": 687
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9859336459209e-06,
+      "loss": 3.5627,
+      "step": 688
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985783677012857e-06,
+      "loss": 3.4003,
+      "step": 689
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985632914027116e-06,
+      "loss": 3.518,
+      "step": 690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985481356987685e-06,
+      "loss": 3.3548,
+      "step": 691
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985329005918702e-06,
+      "loss": 3.5286,
+      "step": 692
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985175860844437e-06,
+      "loss": 3.4916,
+      "step": 693
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985021921789274e-06,
+      "loss": 3.3779,
+      "step": 694
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984867188777736e-06,
+      "loss": 3.4956,
+      "step": 695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984711661834465e-06,
+      "loss": 3.4841,
+      "step": 696
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98455534098423e-06,
+      "loss": 3.409,
+      "step": 697
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98439822625193e-06,
+      "loss": 3.5565,
+      "step": 698
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98424031766259e-06,
+      "loss": 3.4638,
+      "step": 699
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984081615241356e-06,
+      "loss": 3.5008,
+      "step": 700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983922119013507e-06,
+      "loss": 3.4161,
+      "step": 701
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983761829004445e-06,
+      "loss": 3.4552,
+      "step": 702
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9836007452397e-06,
+      "loss": 3.372,
+      "step": 703
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983438867744923e-06,
+      "loss": 3.3468,
+      "step": 704
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983276196545902e-06,
+      "loss": 3.6066,
+      "step": 705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983112731668543e-06,
+      "loss": 3.4901,
+      "step": 706
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982948473138881e-06,
+      "loss": 3.499,
+      "step": 707
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982783420983075e-06,
+      "loss": 3.5446,
+      "step": 708
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982617575227416e-06,
+      "loss": 3.5094,
+      "step": 709
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982450935898316e-06,
+      "loss": 3.4166,
+      "step": 710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982283503022316e-06,
+      "loss": 3.4141,
+      "step": 711
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.982115276626083e-06,
+      "loss": 3.4531,
+      "step": 712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981946256736409e-06,
+      "loss": 3.5112,
+      "step": 713
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981776443380214e-06,
+      "loss": 3.4343,
+      "step": 714
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981605836584545e-06,
+      "loss": 3.3788,
+      "step": 715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981434436376572e-06,
+      "loss": 3.4344,
+      "step": 716
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981262242783597e-06,
+      "loss": 3.5668,
+      "step": 717
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981089255833042e-06,
+      "loss": 3.5524,
+      "step": 718
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.98091547555246e-06,
+      "loss": 3.5917,
+      "step": 719
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.980740901969527e-06,
+      "loss": 3.4098,
+      "step": 720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.98056553511205e-06,
+      "loss": 3.4803,
+      "step": 721
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.980389375007956e-06,
+      "loss": 3.6166,
+      "step": 722
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.980212421685303e-06,
+      "loss": 3.5719,
+      "step": 723
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.980034675172274e-06,
+      "loss": 3.5625,
+      "step": 724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97985613549718e-06,
+      "loss": 3.6134,
+      "step": 725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979676802688456e-06,
+      "loss": 3.4926,
+      "step": 726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979496676774663e-06,
+      "loss": 3.4169,
+      "step": 727
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97931575778449e-06,
+      "loss": 3.4258,
+      "step": 728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97913404574675e-06,
+      "loss": 3.5532,
+      "step": 729
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978951540690387e-06,
+      "loss": 3.4453,
+      "step": 730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978768242644466e-06,
+      "loss": 3.4636,
+      "step": 731
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978584151638182e-06,
+      "loss": 3.4107,
+      "step": 732
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978399267700855e-06,
+      "loss": 3.5201,
+      "step": 733
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97821359086193e-06,
+      "loss": 3.4455,
+      "step": 734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978027121150979e-06,
+      "loss": 3.5158,
+      "step": 735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977839858597703e-06,
+      "loss": 3.4996,
+      "step": 736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977651803231925e-06,
+      "loss": 3.4101,
+      "step": 737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977462955083596e-06,
+      "loss": 3.4584,
+      "step": 738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977273314182795e-06,
+      "loss": 3.3982,
+      "step": 739
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977082880559725e-06,
+      "loss": 3.4595,
+      "step": 740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976891654244718e-06,
+      "loss": 3.4597,
+      "step": 741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976699635268227e-06,
+      "loss": 3.5215,
+      "step": 742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976506823660836e-06,
+      "loss": 3.4048,
+      "step": 743
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976313219453255e-06,
+      "loss": 3.6094,
+      "step": 744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976118822676317e-06,
+      "loss": 3.4695,
+      "step": 745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975923633360985e-06,
+      "loss": 3.5538,
+      "step": 746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975727651538345e-06,
+      "loss": 3.4323,
+      "step": 747
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975530877239613e-06,
+      "loss": 3.564,
+      "step": 748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975333310496127e-06,
+      "loss": 3.487,
+      "step": 749
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975134951339351e-06,
+      "loss": 3.5288,
+      "step": 750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.974935799800881e-06,
+      "loss": 3.4811,
+      "step": 751
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.974735855912436e-06,
+      "loss": 3.4195,
+      "step": 752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.974535119705859e-06,
+      "loss": 3.442,
+      "step": 753
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97433359121312e-06,
+      "loss": 3.4686,
+      "step": 754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.974131270466317e-06,
+      "loss": 3.4643,
+      "step": 755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973928157497675e-06,
+      "loss": 3.5082,
+      "step": 756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973724252339541e-06,
+      "loss": 3.5096,
+      "step": 757
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97351955502439e-06,
+      "loss": 3.3489,
+      "step": 758
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973314065584828e-06,
+      "loss": 3.5042,
+      "step": 759
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97310778405358e-06,
+      "loss": 3.5078,
+      "step": 760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.972900710463498e-06,
+      "loss": 3.4537,
+      "step": 761
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.972692844847567e-06,
+      "loss": 3.4423,
+      "step": 762
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97248418723889e-06,
+      "loss": 3.5473,
+      "step": 763
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.972274737670702e-06,
+      "loss": 3.4717,
+      "step": 764
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97206449617636e-06,
+      "loss": 3.537,
+      "step": 765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97185346278935e-06,
+      "loss": 3.56,
+      "step": 766
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97164163754328e-06,
+      "loss": 3.471,
+      "step": 767
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97142902047189e-06,
+      "loss": 3.4942,
+      "step": 768
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.971215611609045e-06,
+      "loss": 3.3927,
+      "step": 769
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.971001410988729e-06,
+      "loss": 3.5457,
+      "step": 770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970786418645061e-06,
+      "loss": 3.4731,
+      "step": 771
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970570634612282e-06,
+      "loss": 3.4471,
+      "step": 772
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970354058924758e-06,
+      "loss": 3.3895,
+      "step": 773
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970136691616985e-06,
+      "loss": 3.507,
+      "step": 774
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969918532723581e-06,
+      "loss": 3.4764,
+      "step": 775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969699582279294e-06,
+      "loss": 3.4979,
+      "step": 776
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969479840318992e-06,
+      "loss": 3.4429,
+      "step": 777
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969259306877676e-06,
+      "loss": 3.4227,
+      "step": 778
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969037981990468e-06,
+      "loss": 3.378,
+      "step": 779
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968815865692622e-06,
+      "loss": 3.4449,
+      "step": 780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96859295801951e-06,
+      "loss": 3.5871,
+      "step": 781
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968369259006635e-06,
+      "loss": 3.4294,
+      "step": 782
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968144768689626e-06,
+      "loss": 3.5593,
+      "step": 783
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967919487104237e-06,
+      "loss": 3.5037,
+      "step": 784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967693414286347e-06,
+      "loss": 3.4378,
+      "step": 785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967466550271967e-06,
+      "loss": 3.5173,
+      "step": 786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967238895097223e-06,
+      "loss": 3.4688,
+      "step": 787
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967010448798376e-06,
+      "loss": 3.514,
+      "step": 788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966781211411811e-06,
+      "loss": 3.5525,
+      "step": 789
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966551182974038e-06,
+      "loss": 3.5216,
+      "step": 790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96632036352169e-06,
+      "loss": 3.4567,
+      "step": 791
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966088753091533e-06,
+      "loss": 3.4081,
+      "step": 792
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.965856351720457e-06,
+      "loss": 3.5282,
+      "step": 793
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.965623159445471e-06,
+      "loss": 3.4641,
+      "step": 794
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.965389176303717e-06,
+      "loss": 3.4237,
+      "step": 795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.965154402332464e-06,
+      "loss": 3.4983,
+      "step": 796
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.9649188375691e-06,
+      "loss": 3.4836,
+      "step": 797
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.964682482051145e-06,
+      "loss": 3.4157,
+      "step": 798
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.964445335816243e-06,
+      "loss": 3.478,
+      "step": 799
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.964207398902163e-06,
+      "loss": 3.4669,
+      "step": 800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.963968671346802e-06,
+      "loss": 3.4917,
+      "step": 801
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.963729153188182e-06,
+      "loss": 3.3976,
+      "step": 802
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.963488844464448e-06,
+      "loss": 3.6468,
+      "step": 803
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.963247745213876e-06,
+      "loss": 3.4228,
+      "step": 804
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.963005855474865e-06,
+      "loss": 3.4622,
+      "step": 805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.962763175285941e-06,
+      "loss": 3.4775,
+      "step": 806
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.962519704685754e-06,
+      "loss": 3.4096,
+      "step": 807
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.962275443713083e-06,
+      "loss": 3.4176,
+      "step": 808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.962030392406828e-06,
+      "loss": 3.462,
+      "step": 809
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.961784550806021e-06,
+      "loss": 3.5277,
+      "step": 810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.961537918949814e-06,
+      "loss": 3.3221,
+      "step": 811
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.961290496877492e-06,
+      "loss": 3.6116,
+      "step": 812
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.961042284628457e-06,
+      "loss": 3.4695,
+      "step": 813
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.960793282242243e-06,
+      "loss": 3.5959,
+      "step": 814
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.960543489758507e-06,
+      "loss": 3.3854,
+      "step": 815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.960292907217037e-06,
+      "loss": 3.5529,
+      "step": 816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.960041534657738e-06,
+      "loss": 3.5065,
+      "step": 817
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95978937212065e-06,
+      "loss": 3.3688,
+      "step": 818
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95953641964593e-06,
+      "loss": 3.4848,
+      "step": 819
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.959282677273869e-06,
+      "loss": 3.4058,
+      "step": 820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95902814504488e-06,
+      "loss": 3.3558,
+      "step": 821
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.958772822999499e-06,
+      "loss": 3.4579,
+      "step": 822
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.958516711178393e-06,
+      "loss": 3.404,
+      "step": 823
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.958259809622353e-06,
+      "loss": 3.4162,
+      "step": 824
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.958002118372293e-06,
+      "loss": 3.4794,
+      "step": 825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.957743637469256e-06,
+      "loss": 3.4177,
+      "step": 826
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.957484366954412e-06,
+      "loss": 3.517,
+      "step": 827
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.957224306869053e-06,
+      "loss": 3.3921,
+      "step": 828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.956963457254597e-06,
+      "loss": 3.3935,
+      "step": 829
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.956701818152592e-06,
+      "loss": 3.4314,
+      "step": 830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.956439389604705e-06,
+      "loss": 3.4528,
+      "step": 831
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.956176171652737e-06,
+      "loss": 3.4483,
+      "step": 832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.955912164338605e-06,
+      "loss": 3.4463,
+      "step": 833
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.955647367704362e-06,
+      "loss": 3.463,
+      "step": 834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95538178179218e-06,
+      "loss": 3.4075,
+      "step": 835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.955115406644357e-06,
+      "loss": 3.4801,
+      "step": 836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95484824230332e-06,
+      "loss": 3.4543,
+      "step": 837
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95458028881162e-06,
+      "loss": 3.3965,
+      "step": 838
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.954311546211931e-06,
+      "loss": 3.5194,
+      "step": 839
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.954042014547057e-06,
+      "loss": 3.4029,
+      "step": 840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.953771693859927e-06,
+      "loss": 3.5263,
+      "step": 841
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.953500584193593e-06,
+      "loss": 3.5208,
+      "step": 842
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.953228685591234e-06,
+      "loss": 3.4794,
+      "step": 843
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.952955998096155e-06,
+      "loss": 3.4512,
+      "step": 844
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.952682521751788e-06,
+      "loss": 3.4594,
+      "step": 845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.952408256601687e-06,
+      "loss": 3.5202,
+      "step": 846
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.952133202689536e-06,
+      "loss": 3.5577,
+      "step": 847
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.951857360059141e-06,
+      "loss": 3.5404,
+      "step": 848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.951580728754434e-06,
+      "loss": 3.3667,
+      "step": 849
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.951303308819475e-06,
+      "loss": 3.4285,
+      "step": 850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.951025100298449e-06,
+      "loss": 3.4482,
+      "step": 851
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.950746103235663e-06,
+      "loss": 3.5374,
+      "step": 852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.950466317675556e-06,
+      "loss": 3.5179,
+      "step": 853
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.950185743662685e-06,
+      "loss": 3.4672,
+      "step": 854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.949904381241742e-06,
+      "loss": 3.3577,
+      "step": 855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.949622230457533e-06,
+      "loss": 3.5543,
+      "step": 856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.949339291355e-06,
+      "loss": 3.4482,
+      "step": 857
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.949055563979205e-06,
+      "loss": 3.4756,
+      "step": 858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.948771048375337e-06,
+      "loss": 3.3437,
+      "step": 859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.948485744588709e-06,
+      "loss": 3.4393,
+      "step": 860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.948199652664762e-06,
+      "loss": 3.471,
+      "step": 861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.947912772649061e-06,
+      "loss": 3.4793,
+      "step": 862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.9476251045873e-06,
+      "loss": 3.454,
+      "step": 863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94733664852529e-06,
+      "loss": 3.3295,
+      "step": 864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.947047404508976e-06,
+      "loss": 3.5728,
+      "step": 865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.946757372584423e-06,
+      "loss": 3.454,
+      "step": 866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.946466552797829e-06,
+      "loss": 3.4356,
+      "step": 867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.946174945195508e-06,
+      "loss": 3.4585,
+      "step": 868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.945882549823906e-06,
+      "loss": 3.5121,
+      "step": 869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.945589366729591e-06,
+      "loss": 3.4769,
+      "step": 870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.945295395959259e-06,
+      "loss": 3.5238,
+      "step": 871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.945000637559728e-06,
+      "loss": 3.5393,
+      "step": 872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.944705091577945e-06,
+      "loss": 3.5134,
+      "step": 873
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.944408758060982e-06,
+      "loss": 3.5572,
+      "step": 874
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.944111637056035e-06,
+      "loss": 3.4423,
+      "step": 875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.943813728610428e-06,
+      "loss": 3.5128,
+      "step": 876
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.943515032771605e-06,
+      "loss": 3.4666,
+      "step": 877
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.943215549587139e-06,
+      "loss": 3.4523,
+      "step": 878
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94291527910473e-06,
+      "loss": 3.5372,
+      "step": 879
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.9426142213722e-06,
+      "loss": 3.3993,
+      "step": 880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.942312376437499e-06,
+      "loss": 3.4193,
+      "step": 881
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.942009744348704e-06,
+      "loss": 3.4355,
+      "step": 882
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.941706325154008e-06,
+      "loss": 3.5073,
+      "step": 883
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.941402118901743e-06,
+      "loss": 3.4649,
+      "step": 884
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.941097125640355e-06,
+      "loss": 3.5095,
+      "step": 885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.940791345418423e-06,
+      "loss": 3.5283,
+      "step": 886
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.940484778284645e-06,
+      "loss": 3.4155,
+      "step": 887
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.940177424287848e-06,
+      "loss": 3.4269,
+      "step": 888
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.939869283476987e-06,
+      "loss": 3.4535,
+      "step": 889
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.939560355901136e-06,
+      "loss": 3.4349,
+      "step": 890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.939250641609497e-06,
+      "loss": 3.5362,
+      "step": 891
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.938940140651398e-06,
+      "loss": 3.4203,
+      "step": 892
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.938628853076295e-06,
+      "loss": 3.3559,
+      "step": 893
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.938316778933763e-06,
+      "loss": 3.4429,
+      "step": 894
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.938003918273507e-06,
+      "loss": 3.5209,
+      "step": 895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.937690271145356e-06,
+      "loss": 3.4789,
+      "step": 896
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.937375837599263e-06,
+      "loss": 3.4743,
+      "step": 897
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.937060617685306e-06,
+      "loss": 3.3472,
+      "step": 898
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.936744611453694e-06,
+      "loss": 3.4743,
+      "step": 899
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.936427818954753e-06,
+      "loss": 3.3508,
+      "step": 900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.936110240238939e-06,
+      "loss": 3.5217,
+      "step": 901
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.935791875356832e-06,
+      "loss": 3.4384,
+      "step": 902
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.93547272435914e-06,
+      "loss": 3.2915,
+      "step": 903
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.935152787296689e-06,
+      "loss": 3.4375,
+      "step": 904
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.934832064220438e-06,
+      "loss": 3.4804,
+      "step": 905
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.934510555181469e-06,
+      "loss": 3.5847,
+      "step": 906
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.934188260230986e-06,
+      "loss": 3.4296,
+      "step": 907
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.93386517942032e-06,
+      "loss": 3.5466,
+      "step": 908
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.933541312800929e-06,
+      "loss": 3.504,
+      "step": 909
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.933216660424396e-06,
+      "loss": 3.4557,
+      "step": 910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.932891222342425e-06,
+      "loss": 3.4697,
+      "step": 911
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.932564998606849e-06,
+      "loss": 3.5082,
+      "step": 912
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.932237989269626e-06,
+      "loss": 3.4445,
+      "step": 913
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.931910194382836e-06,
+      "loss": 3.4797,
+      "step": 914
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.93158161399869e-06,
+      "loss": 3.4518,
+      "step": 915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.931252248169518e-06,
+      "loss": 3.4933,
+      "step": 916
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.930922096947776e-06,
+      "loss": 3.5335,
+      "step": 917
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.930591160386051e-06,
+      "loss": 3.4981,
+      "step": 918
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.93025943853705e-06,
+      "loss": 3.4734,
+      "step": 919
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.9299269314536e-06,
+      "loss": 3.5169,
+      "step": 920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.929593639188665e-06,
+      "loss": 3.4481,
+      "step": 921
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.929259561795325e-06,
+      "loss": 3.3403,
+      "step": 922
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.928924699326791e-06,
+      "loss": 3.4658,
+      "step": 923
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.928589051836392e-06,
+      "loss": 3.4374,
+      "step": 924
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.928252619377589e-06,
+      "loss": 3.5213,
+      "step": 925
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.927915402003964e-06,
+      "loss": 3.4457,
+      "step": 926
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.927577399769224e-06,
+      "loss": 3.4761,
+      "step": 927
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.927238612727204e-06,
+      "loss": 3.4774,
+      "step": 928
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.926899040931861e-06,
+      "loss": 3.5482,
+      "step": 929
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.926558684437278e-06,
+      "loss": 3.4301,
+      "step": 930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.926217543297664e-06,
+      "loss": 3.4714,
+      "step": 931
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92587561756735e-06,
+      "loss": 3.4711,
+      "step": 932
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.925532907300796e-06,
+      "loss": 3.3741,
+      "step": 933
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.925189412552585e-06,
+      "loss": 3.4999,
+      "step": 934
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.924845133377423e-06,
+      "loss": 3.3911,
+      "step": 935
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.924500069830145e-06,
+      "loss": 3.5205,
+      "step": 936
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.924154221965705e-06,
+      "loss": 3.489,
+      "step": 937
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.923807589839191e-06,
+      "loss": 3.4171,
+      "step": 938
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.923460173505805e-06,
+      "loss": 3.5201,
+      "step": 939
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.923111973020885e-06,
+      "loss": 3.4647,
+      "step": 940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922762988439884e-06,
+      "loss": 3.4385,
+      "step": 941
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922413219818386e-06,
+      "loss": 3.4547,
+      "step": 942
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922062667212095e-06,
+      "loss": 3.542,
+      "step": 943
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.921711330676848e-06,
+      "loss": 3.5076,
+      "step": 944
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.921359210268601e-06,
+      "loss": 3.5529,
+      "step": 945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.921006306043433e-06,
+      "loss": 3.5717,
+      "step": 946
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92065261805755e-06,
+      "loss": 3.4046,
+      "step": 947
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.920298146367287e-06,
+      "loss": 3.4273,
+      "step": 948
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.919942891029097e-06,
+      "loss": 3.5092,
+      "step": 949
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.919586852099562e-06,
+      "loss": 3.5697,
+      "step": 950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.919230029635388e-06,
+      "loss": 3.5238,
+      "step": 951
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.918872423693405e-06,
+      "loss": 3.3985,
+      "step": 952
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.918514034330569e-06,
+      "loss": 3.4357,
+      "step": 953
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.918154861603959e-06,
+      "loss": 3.398,
+      "step": 954
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.917794905570781e-06,
+      "loss": 3.4736,
+      "step": 955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.917434166288364e-06,
+      "loss": 3.4974,
+      "step": 956
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.917072643814163e-06,
+      "loss": 3.5792,
+      "step": 957
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.916710338205756e-06,
+      "loss": 3.4179,
+      "step": 958
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.916347249520848e-06,
+      "loss": 3.4263,
+      "step": 959
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.915983377817267e-06,
+      "loss": 3.3868,
+      "step": 960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.915618723152967e-06,
+      "loss": 3.4954,
+      "step": 961
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.915253285586024e-06,
+      "loss": 3.3834,
+      "step": 962
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.914887065174643e-06,
+      "loss": 3.4776,
+      "step": 963
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91452006197715e-06,
+      "loss": 3.6079,
+      "step": 964
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.914152276051996e-06,
+      "loss": 3.4181,
+      "step": 965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.913783707457759e-06,
+      "loss": 3.5244,
+      "step": 966
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91341435625314e-06,
+      "loss": 3.5095,
+      "step": 967
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.913044222496966e-06,
+      "loss": 3.4042,
+      "step": 968
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.912673306248188e-06,
+      "loss": 3.4722,
+      "step": 969
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.912301607565879e-06,
+      "loss": 3.5526,
+      "step": 970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91192912650924e-06,
+      "loss": 3.3936,
+      "step": 971
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.911555863137593e-06,
+      "loss": 3.4362,
+      "step": 972
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.911181817510392e-06,
+      "loss": 3.5358,
+      "step": 973
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.910806989687207e-06,
+      "loss": 3.5554,
+      "step": 974
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.910431379727737e-06,
+      "loss": 3.3768,
+      "step": 975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.910054987691805e-06,
+      "loss": 3.4689,
+      "step": 976
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.909677813639361e-06,
+      "loss": 3.4591,
+      "step": 977
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.90929985763047e-06,
+      "loss": 3.3859,
+      "step": 978
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.908921119725337e-06,
+      "loss": 3.3621,
+      "step": 979
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.908541599984276e-06,
+      "loss": 3.4625,
+      "step": 980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.908161298467735e-06,
+      "loss": 3.4722,
+      "step": 981
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.907780215236285e-06,
+      "loss": 3.3823,
+      "step": 982
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.90739835035062e-06,
+      "loss": 3.5198,
+      "step": 983
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.907015703871558e-06,
+      "loss": 3.5147,
+      "step": 984
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.906632275860044e-06,
+      "loss": 3.3958,
+      "step": 985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.906248066377144e-06,
+      "loss": 3.4333,
+      "step": 986
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.905863075484053e-06,
+      "loss": 3.2799,
+      "step": 987
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.905477303242085e-06,
+      "loss": 3.4195,
+      "step": 988
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.905090749712684e-06,
+      "loss": 3.5255,
+      "step": 989
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.904703414957412e-06,
+      "loss": 3.4226,
+      "step": 990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.904315299037966e-06,
+      "loss": 3.4576,
+      "step": 991
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.903926402016153e-06,
+      "loss": 3.4213,
+      "step": 992
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.903536723953916e-06,
+      "loss": 3.4038,
+      "step": 993
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.903146264913318e-06,
+      "loss": 3.4732,
+      "step": 994
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.902755024956548e-06,
+      "loss": 3.3742,
+      "step": 995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.902363004145914e-06,
+      "loss": 3.3639,
+      "step": 996
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.901970202543856e-06,
+      "loss": 3.3997,
+      "step": 997
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.901576620212934e-06,
+      "loss": 3.4971,
+      "step": 998
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.901182257215834e-06,
+      "loss": 3.5491,
+      "step": 999
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.900787113615365e-06,
+      "loss": 3.4641,
+      "step": 1000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.900391189474458e-06,
+      "loss": 3.4089,
+      "step": 1001
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.899994484856177e-06,
+      "loss": 3.4396,
+      "step": 1002
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8995969998237e-06,
+      "loss": 3.4319,
+      "step": 1003
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.899198734440335e-06,
+      "loss": 3.4,
+      "step": 1004
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.898799688769513e-06,
+      "loss": 3.4858,
+      "step": 1005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.89839986287479e-06,
+      "loss": 3.3717,
+      "step": 1006
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.897999256819846e-06,
+      "loss": 3.4261,
+      "step": 1007
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.897597870668483e-06,
+      "loss": 3.3855,
+      "step": 1008
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.897195704484633e-06,
+      "loss": 3.4645,
+      "step": 1009
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.896792758332343e-06,
+      "loss": 3.4903,
+      "step": 1010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.896389032275793e-06,
+      "loss": 3.5054,
+      "step": 1011
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.895984526379282e-06,
+      "loss": 3.5567,
+      "step": 1012
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.895579240707236e-06,
+      "loss": 3.5903,
+      "step": 1013
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.895173175324205e-06,
+      "loss": 3.581,
+      "step": 1014
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.89476633029486e-06,
+      "loss": 3.4707,
+      "step": 1015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.894358705684002e-06,
+      "loss": 3.384,
+      "step": 1016
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.89395030155655e-06,
+      "loss": 3.5091,
+      "step": 1017
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.893541117977552e-06,
+      "loss": 3.3967,
+      "step": 1018
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.893131155012176e-06,
+      "loss": 3.5666,
+      "step": 1019
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.892720412725717e-06,
+      "loss": 3.5125,
+      "step": 1020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.892308891183591e-06,
+      "loss": 3.59,
+      "step": 1021
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.891896590451345e-06,
+      "loss": 3.5116,
+      "step": 1022
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.891483510594642e-06,
+      "loss": 3.4008,
+      "step": 1023
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.891069651679273e-06,
+      "loss": 3.5148,
+      "step": 1024
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.890655013771153e-06,
+      "loss": 3.4908,
+      "step": 1025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.890239596936321e-06,
+      "loss": 3.4813,
+      "step": 1026
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.889823401240941e-06,
+      "loss": 3.522,
+      "step": 1027
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.889406426751296e-06,
+      "loss": 3.4753,
+      "step": 1028
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.888988673533801e-06,
+      "loss": 3.4975,
+      "step": 1029
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.88857014165499e-06,
+      "loss": 3.4368,
+      "step": 1030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.88815083118152e-06,
+      "loss": 3.4569,
+      "step": 1031
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.887730742180174e-06,
+      "loss": 3.38,
+      "step": 1032
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.88730987471786e-06,
+      "loss": 3.4996,
+      "step": 1033
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.886888228861608e-06,
+      "loss": 3.4621,
+      "step": 1034
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.886465804678576e-06,
+      "loss": 3.4414,
+      "step": 1035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.88604260223604e-06,
+      "loss": 3.4767,
+      "step": 1036
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.885618621601403e-06,
+      "loss": 3.5122,
+      "step": 1037
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.885193862842191e-06,
+      "loss": 3.416,
+      "step": 1038
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.884768326026056e-06,
+      "loss": 3.4803,
+      "step": 1039
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.88434201122077e-06,
+      "loss": 3.4866,
+      "step": 1040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.883914918494236e-06,
+      "loss": 3.4218,
+      "step": 1041
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.883487047914472e-06,
+      "loss": 3.4796,
+      "step": 1042
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.88305839954963e-06,
+      "loss": 3.5474,
+      "step": 1043
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.882628973467972e-06,
+      "loss": 3.4875,
+      "step": 1044
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.882198769737899e-06,
+      "loss": 3.468,
+      "step": 1045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.881767788427926e-06,
+      "loss": 3.4261,
+      "step": 1046
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.881336029606693e-06,
+      "loss": 3.4195,
+      "step": 1047
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.880903493342968e-06,
+      "loss": 3.5625,
+      "step": 1048
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.88047017970564e-06,
+      "loss": 3.4731,
+      "step": 1049
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.880036088763722e-06,
+      "loss": 3.4707,
+      "step": 1050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.87960122058635e-06,
+      "loss": 3.4711,
+      "step": 1051
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.879165575242788e-06,
+      "loss": 3.4752,
+      "step": 1052
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.878729152802416e-06,
+      "loss": 3.5142,
+      "step": 1053
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.878291953334744e-06,
+      "loss": 3.5471,
+      "step": 1054
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.877853976909406e-06,
+      "loss": 3.4879,
+      "step": 1055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.877415223596157e-06,
+      "loss": 3.444,
+      "step": 1056
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.876975693464873e-06,
+      "loss": 3.4685,
+      "step": 1057
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.876535386585561e-06,
+      "loss": 3.4171,
+      "step": 1058
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.876094303028348e-06,
+      "loss": 3.5028,
+      "step": 1059
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.875652442863483e-06,
+      "loss": 3.5104,
+      "step": 1060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.87520980616134e-06,
+      "loss": 3.4606,
+      "step": 1061
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.874766392992421e-06,
+      "loss": 3.3834,
+      "step": 1062
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.874322203427342e-06,
+      "loss": 3.3744,
+      "step": 1063
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.873877237536854e-06,
+      "loss": 3.5233,
+      "step": 1064
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.87343149539182e-06,
+      "loss": 3.4772,
+      "step": 1065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.872984977063236e-06,
+      "loss": 3.5058,
+      "step": 1066
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.872537682622221e-06,
+      "loss": 3.5084,
+      "step": 1067
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.87208961214001e-06,
+      "loss": 3.3274,
+      "step": 1068
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.871640765687967e-06,
+      "loss": 3.5329,
+      "step": 1069
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.871191143337582e-06,
+      "loss": 3.4379,
+      "step": 1070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.870740745160464e-06,
+      "loss": 3.4996,
+      "step": 1071
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.870289571228346e-06,
+      "loss": 3.5414,
+      "step": 1072
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.869837621613088e-06,
+      "loss": 3.4668,
+      "step": 1073
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.869384896386669e-06,
+      "loss": 3.4707,
+      "step": 1074
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.868931395621195e-06,
+      "loss": 3.4328,
+      "step": 1075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.868477119388897e-06,
+      "loss": 3.5059,
+      "step": 1076
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.868022067762122e-06,
+      "loss": 3.4937,
+      "step": 1077
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.867566240813346e-06,
+      "loss": 3.6021,
+      "step": 1078
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.867109638615173e-06,
+      "loss": 3.4643,
+      "step": 1079
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.86665226124032e-06,
+      "loss": 3.5556,
+      "step": 1080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.866194108761633e-06,
+      "loss": 3.4079,
+      "step": 1081
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.865735181252084e-06,
+      "loss": 3.4549,
+      "step": 1082
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.865275478784765e-06,
+      "loss": 3.444,
+      "step": 1083
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.86481500143289e-06,
+      "loss": 3.4051,
+      "step": 1084
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.8643537492698e-06,
+      "loss": 3.5362,
+      "step": 1085
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.863891722368958e-06,
+      "loss": 3.5495,
+      "step": 1086
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.86342892080395e-06,
+      "loss": 3.5257,
+      "step": 1087
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.862965344648485e-06,
+      "loss": 3.5048,
+      "step": 1088
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.862500993976398e-06,
+      "loss": 3.4729,
+      "step": 1089
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.862035868861644e-06,
+      "loss": 3.3867,
+      "step": 1090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.8615699693783e-06,
+      "loss": 3.6065,
+      "step": 1091
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.861103295600574e-06,
+      "loss": 3.4419,
+      "step": 1092
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.860635847602791e-06,
+      "loss": 3.5031,
+      "step": 1093
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.860167625459398e-06,
+      "loss": 3.4384,
+      "step": 1094
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.85969862924497e-06,
+      "loss": 3.4321,
+      "step": 1095
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.859228859034205e-06,
+      "loss": 3.4808,
+      "step": 1096
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.858758314901918e-06,
+      "loss": 3.3406,
+      "step": 1097
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.858286996923056e-06,
+      "loss": 3.4561,
+      "step": 1098
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.857814905172683e-06,
+      "loss": 3.5547,
+      "step": 1099
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.85734203972599e-06,
+      "loss": 3.4196,
+      "step": 1100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.856868400658288e-06,
+      "loss": 3.4058,
+      "step": 1101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.856393988045012e-06,
+      "loss": 3.3691,
+      "step": 1102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.855918801961722e-06,
+      "loss": 3.4494,
+      "step": 1103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.855442842484101e-06,
+      "loss": 3.5576,
+      "step": 1104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.854966109687953e-06,
+      "loss": 3.3953,
+      "step": 1105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.854488603649206e-06,
+      "loss": 3.4923,
+      "step": 1106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.854010324443914e-06,
+      "loss": 3.4072,
+      "step": 1107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.853531272148248e-06,
+      "loss": 3.4476,
+      "step": 1108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.853051446838511e-06,
+      "loss": 3.4735,
+      "step": 1109
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.85257084859112e-06,
+      "loss": 3.4533,
+      "step": 1110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.85208947748262e-06,
+      "loss": 3.4324,
+      "step": 1111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.851607333589677e-06,
+      "loss": 3.4993,
+      "step": 1112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.851124416989084e-06,
+      "loss": 3.4479,
+      "step": 1113
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.850640727757753e-06,
+      "loss": 3.4927,
+      "step": 1114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.850156265972722e-06,
+      "loss": 3.4507,
+      "step": 1115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.849671031711146e-06,
+      "loss": 3.3539,
+      "step": 1116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.849185025050311e-06,
+      "loss": 3.4424,
+      "step": 1117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.848698246067622e-06,
+      "loss": 3.3755,
+      "step": 1118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.848210694840609e-06,
+      "loss": 3.439,
+      "step": 1119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.847722371446921e-06,
+      "loss": 3.4153,
+      "step": 1120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.847233275964333e-06,
+      "loss": 3.5584,
+      "step": 1121
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.846743408470742e-06,
+      "loss": 3.5111,
+      "step": 1122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.84625276904417e-06,
+      "loss": 3.4549,
+      "step": 1123
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.84576135776276e-06,
+      "loss": 3.475,
+      "step": 1124
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.845269174704776e-06,
+      "loss": 3.5653,
+      "step": 1125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.84477621994861e-06,
+      "loss": 3.4528,
+      "step": 1126
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.844282493572774e-06,
+      "loss": 3.3605,
+      "step": 1127
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.843787995655903e-06,
+      "loss": 3.548,
+      "step": 1128
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.843292726276751e-06,
+      "loss": 3.4685,
+      "step": 1129
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.842796685514204e-06,
+      "loss": 3.4175,
+      "step": 1130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.842299873447263e-06,
+      "loss": 3.3905,
+      "step": 1131
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.841802290155054e-06,
+      "loss": 3.5266,
+      "step": 1132
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.841303935716828e-06,
+      "loss": 3.3965,
+      "step": 1133
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.840804810211954e-06,
+      "loss": 3.5639,
+      "step": 1134
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.840304913719931e-06,
+      "loss": 3.4266,
+      "step": 1135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.839804246320374e-06,
+      "loss": 3.4504,
+      "step": 1136
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.839302808093026e-06,
+      "loss": 3.5075,
+      "step": 1137
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.838800599117748e-06,
+      "loss": 3.4925,
+      "step": 1138
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.838297619474526e-06,
+      "loss": 3.495,
+      "step": 1139
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.837793869243468e-06,
+      "loss": 3.4343,
+      "step": 1140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.83728934850481e-06,
+      "loss": 3.4119,
+      "step": 1141
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.8367840573389e-06,
+      "loss": 3.4311,
+      "step": 1142
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.83627799582622e-06,
+      "loss": 3.3615,
+      "step": 1143
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.835771164047365e-06,
+      "loss": 3.2967,
+      "step": 1144
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.835263562083062e-06,
+      "loss": 3.4164,
+      "step": 1145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.834755190014154e-06,
+      "loss": 3.4065,
+      "step": 1146
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.834246047921608e-06,
+      "loss": 3.4737,
+      "step": 1147
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.833736135886513e-06,
+      "loss": 3.314,
+      "step": 1148
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.833225453990084e-06,
+      "loss": 3.5005,
+      "step": 1149
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.832714002313657e-06,
+      "loss": 3.3789,
+      "step": 1150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.832201780938688e-06,
+      "loss": 3.4689,
+      "step": 1151
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.831688789946758e-06,
+      "loss": 3.5652,
+      "step": 1152
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.831175029419573e-06,
+      "loss": 3.5996,
+      "step": 1153
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.830660499438955e-06,
+      "loss": 3.4612,
+      "step": 1154
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.830145200086855e-06,
+      "loss": 3.324,
+      "step": 1155
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.829629131445342e-06,
+      "loss": 3.4133,
+      "step": 1156
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.82911229359661e-06,
+      "loss": 3.5339,
+      "step": 1157
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.828594686622977e-06,
+      "loss": 3.3447,
+      "step": 1158
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.82807631060688e-06,
+      "loss": 3.414,
+      "step": 1159
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.827557165630879e-06,
+      "loss": 3.385,
+      "step": 1160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.827037251777657e-06,
+      "loss": 3.4713,
+      "step": 1161
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.826516569130023e-06,
+      "loss": 3.423,
+      "step": 1162
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.825995117770902e-06,
+      "loss": 3.5141,
+      "step": 1163
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.825472897783344e-06,
+      "loss": 3.4764,
+      "step": 1164
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.824949909250525e-06,
+      "loss": 3.5138,
+      "step": 1165
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.82442615225574e-06,
+      "loss": 3.5109,
+      "step": 1166
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.823901626882407e-06,
+      "loss": 3.491,
+      "step": 1167
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.823376333214065e-06,
+      "loss": 3.4418,
+      "step": 1168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.822850271334379e-06,
+      "loss": 3.4679,
+      "step": 1169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.82232344132713e-06,
+      "loss": 3.4438,
+      "step": 1170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.82179584327623e-06,
+      "loss": 3.3409,
+      "step": 1171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.821267477265705e-06,
+      "loss": 3.5066,
+      "step": 1172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.82073834337971e-06,
+      "loss": 3.2994,
+      "step": 1173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.820208441702516e-06,
+      "loss": 3.4259,
+      "step": 1174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.819677772318523e-06,
+      "loss": 3.4656,
+      "step": 1175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.819146335312249e-06,
+      "loss": 3.4551,
+      "step": 1176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.81861413076833e-06,
+      "loss": 3.5792,
+      "step": 1177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.818081158771538e-06,
+      "loss": 3.4527,
+      "step": 1178
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.817547419406755e-06,
+      "loss": 3.4602,
+      "step": 1179
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.817012912758986e-06,
+      "loss": 3.4223,
+      "step": 1180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.816477638913363e-06,
+      "loss": 3.391,
+      "step": 1181
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.815941597955138e-06,
+      "loss": 3.5305,
+      "step": 1182
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.815404789969686e-06,
+      "loss": 3.4621,
+      "step": 1183
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.814867215042503e-06,
+      "loss": 3.3249,
+      "step": 1184
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.814328873259208e-06,
+      "loss": 3.4756,
+      "step": 1185
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.813789764705543e-06,
+      "loss": 3.3926,
+      "step": 1186
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.813249889467369e-06,
+      "loss": 3.4373,
+      "step": 1187
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.812709247630671e-06,
+      "loss": 3.4329,
+      "step": 1188
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.81216783928156e-06,
+      "loss": 3.3941,
+      "step": 1189
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.811625664506259e-06,
+      "loss": 3.505,
+      "step": 1190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.811082723391123e-06,
+      "loss": 3.4662,
+      "step": 1191
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.810539016022625e-06,
+      "loss": 3.4497,
+      "step": 1192
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.80999454248736e-06,
+      "loss": 3.4801,
+      "step": 1193
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.809449302872045e-06,
+      "loss": 3.4962,
+      "step": 1194
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.808903297263521e-06,
+      "loss": 3.4737,
+      "step": 1195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.808356525748748e-06,
+      "loss": 3.4941,
+      "step": 1196
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.807808988414811e-06,
+      "loss": 3.5027,
+      "step": 1197
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.807260685348912e-06,
+      "loss": 3.5469,
+      "step": 1198
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.806711616638382e-06,
+      "loss": 3.4081,
+      "step": 1199
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.80616178237067e-06,
+      "loss": 3.3765,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.805611182633346e-06,
+      "loss": 3.4776,
+      "step": 1201
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.805059817514101e-06,
+      "loss": 3.547,
+      "step": 1202
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.804507687100753e-06,
+      "loss": 3.4701,
+      "step": 1203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.803954791481239e-06,
+      "loss": 3.4166,
+      "step": 1204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.803401130743618e-06,
+      "loss": 3.4505,
+      "step": 1205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.802846704976069e-06,
+      "loss": 3.3969,
+      "step": 1206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.802291514266892e-06,
+      "loss": 3.4642,
+      "step": 1207
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.801735558704516e-06,
+      "loss": 3.4383,
+      "step": 1208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.801178838377487e-06,
+      "loss": 3.4872,
+      "step": 1209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.800621353374471e-06,
+      "loss": 3.4952,
+      "step": 1210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.80006310378426e-06,
+      "loss": 3.4737,
+      "step": 1211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.799504089695762e-06,
+      "loss": 3.5438,
+      "step": 1212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.798944311198012e-06,
+      "loss": 3.5036,
+      "step": 1213
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.798383768380164e-06,
+      "loss": 3.2986,
+      "step": 1214
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.797822461331497e-06,
+      "loss": 3.5312,
+      "step": 1215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.79726039014141e-06,
+      "loss": 3.4312,
+      "step": 1216
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.79669755489942e-06,
+      "loss": 3.4966,
+      "step": 1217
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.796133955695168e-06,
+      "loss": 3.3879,
+      "step": 1218
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.795569592618424e-06,
+      "loss": 3.5307,
+      "step": 1219
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.795004465759067e-06,
+      "loss": 3.3882,
+      "step": 1220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.794438575207104e-06,
+      "loss": 3.5779,
+      "step": 1221
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.793871921052666e-06,
+      "loss": 3.4944,
+      "step": 1222
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.793304503386005e-06,
+      "loss": 3.4242,
+      "step": 1223
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.792736322297489e-06,
+      "loss": 3.4951,
+      "step": 1224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.792167377877612e-06,
+      "loss": 3.349,
+      "step": 1225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.79159767021699e-06,
+      "loss": 3.5395,
+      "step": 1226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.791027199406357e-06,
+      "loss": 3.402,
+      "step": 1227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.790455965536574e-06,
+      "loss": 3.5081,
+      "step": 1228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.78988396869862e-06,
+      "loss": 3.433,
+      "step": 1229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.789311208983592e-06,
+      "loss": 3.4525,
+      "step": 1230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.788737686482719e-06,
+      "loss": 3.4484,
+      "step": 1231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.78816340128734e-06,
+      "loss": 3.5221,
+      "step": 1232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.787588353488922e-06,
+      "loss": 3.4439,
+      "step": 1233
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.787012543179053e-06,
+      "loss": 3.4329,
+      "step": 1234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.78643597044944e-06,
+      "loss": 3.4512,
+      "step": 1235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.785858635391913e-06,
+      "loss": 3.4987,
+      "step": 1236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.785280538098425e-06,
+      "loss": 3.496,
+      "step": 1237
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.784701678661045e-06,
+      "loss": 3.3598,
+      "step": 1238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.78412205717197e-06,
+      "loss": 3.6085,
+      "step": 1239
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.783541673723514e-06,
+      "loss": 3.5387,
+      "step": 1240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.782960528408116e-06,
+      "loss": 3.4963,
+      "step": 1241
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.782378621318331e-06,
+      "loss": 3.5358,
+      "step": 1242
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.78179595254684e-06,
+      "loss": 3.4283,
+      "step": 1243
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.781212522186442e-06,
+      "loss": 3.4743,
+      "step": 1244
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.780628330330062e-06,
+      "loss": 3.4677,
+      "step": 1245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.78004337707074e-06,
+      "loss": 3.3956,
+      "step": 1246
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.779457662501643e-06,
+      "loss": 3.5546,
+      "step": 1247
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.778871186716058e-06,
+      "loss": 3.5568,
+      "step": 1248
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.778283949807388e-06,
+      "loss": 3.4725,
+      "step": 1249
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.777695951869164e-06,
+      "loss": 3.5389,
+      "step": 1250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.777107192995035e-06,
+      "loss": 3.5222,
+      "step": 1251
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.776517673278772e-06,
+      "loss": 3.5438,
+      "step": 1252
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.775927392814266e-06,
+      "loss": 3.4681,
+      "step": 1253
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.77533635169553e-06,
+      "loss": 3.368,
+      "step": 1254
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.774744550016699e-06,
+      "loss": 3.5153,
+      "step": 1255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.774151987872029e-06,
+      "loss": 3.5235,
+      "step": 1256
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.773558665355894e-06,
+      "loss": 3.5298,
+      "step": 1257
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.772964582562792e-06,
+      "loss": 3.5072,
+      "step": 1258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.772369739587345e-06,
+      "loss": 3.577,
+      "step": 1259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.771774136524287e-06,
+      "loss": 3.4813,
+      "step": 1260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.771177773468482e-06,
+      "loss": 3.3945,
+      "step": 1261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.770580650514913e-06,
+      "loss": 3.4968,
+      "step": 1262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.769982767758684e-06,
+      "loss": 3.3663,
+      "step": 1263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.769384125295012e-06,
+      "loss": 3.4918,
+      "step": 1264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.768784723219248e-06,
+      "loss": 3.4432,
+      "step": 1265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.768184561626854e-06,
+      "loss": 3.4555,
+      "step": 1266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.76758364061342e-06,
+      "loss": 3.4622,
+      "step": 1267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.766981960274653e-06,
+      "loss": 3.4035,
+      "step": 1268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.766379520706382e-06,
+      "loss": 3.567,
+      "step": 1269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.765776322004554e-06,
+      "loss": 3.403,
+      "step": 1270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.76517236426524e-06,
+      "loss": 3.5981,
+      "step": 1271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.764567647584633e-06,
+      "loss": 3.4707,
+      "step": 1272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.763962172059045e-06,
+      "loss": 3.5059,
+      "step": 1273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.763355937784908e-06,
+      "loss": 3.5358,
+      "step": 1274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.762748944858778e-06,
+      "loss": 3.4208,
+      "step": 1275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.762141193377329e-06,
+      "loss": 3.4472,
+      "step": 1276
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.761532683437355e-06,
+      "loss": 3.431,
+      "step": 1277
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.760923415135775e-06,
+      "loss": 3.5547,
+      "step": 1278
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.760313388569621e-06,
+      "loss": 3.5494,
+      "step": 1279
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.759702603836059e-06,
+      "loss": 3.5784,
+      "step": 1280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.759091061032359e-06,
+      "loss": 3.5364,
+      "step": 1281
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.75847876025593e-06,
+      "loss": 3.5072,
+      "step": 1282
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.757865701604283e-06,
+      "loss": 3.5053,
+      "step": 1283
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.757251885175063e-06,
+      "loss": 3.4115,
+      "step": 1284
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.756637311066033e-06,
+      "loss": 3.3499,
+      "step": 1285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.756021979375073e-06,
+      "loss": 3.4298,
+      "step": 1286
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.755405890200184e-06,
+      "loss": 3.5381,
+      "step": 1287
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.754789043639495e-06,
+      "loss": 3.4459,
+      "step": 1288
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.754171439791246e-06,
+      "loss": 3.44,
+      "step": 1289
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.753553078753802e-06,
+      "loss": 3.4986,
+      "step": 1290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.75293396062565e-06,
+      "loss": 3.5139,
+      "step": 1291
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.752314085505396e-06,
+      "loss": 3.3234,
+      "step": 1292
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.751693453491764e-06,
+      "loss": 3.4561,
+      "step": 1293
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.751072064683604e-06,
+      "loss": 3.5932,
+      "step": 1294
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.750449919179883e-06,
+      "loss": 3.3341,
+      "step": 1295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.749827017079687e-06,
+      "loss": 3.4772,
+      "step": 1296
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.749203358482227e-06,
+      "loss": 3.5187,
+      "step": 1297
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.74857894348683e-06,
+      "loss": 3.4302,
+      "step": 1298
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.747953772192947e-06,
+      "loss": 3.5162,
+      "step": 1299
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.747327844700147e-06,
+      "loss": 3.4902,
+      "step": 1300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.746701161108123e-06,
+      "loss": 3.4099,
+      "step": 1301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.746073721516684e-06,
+      "loss": 3.6057,
+      "step": 1302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.745445526025761e-06,
+      "loss": 3.3982,
+      "step": 1303
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.744816574735405e-06,
+      "loss": 3.4535,
+      "step": 1304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.74418686774579e-06,
+      "loss": 3.5797,
+      "step": 1305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.743556405157206e-06,
+      "loss": 3.336,
+      "step": 1306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.742925187070069e-06,
+      "loss": 3.4574,
+      "step": 1307
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.74229321358491e-06,
+      "loss": 3.3497,
+      "step": 1308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.741660484802383e-06,
+      "loss": 3.4447,
+      "step": 1309
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.74102700082326e-06,
+      "loss": 3.4348,
+      "step": 1310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.740392761748438e-06,
+      "loss": 3.4835,
+      "step": 1311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.739757767678928e-06,
+      "loss": 3.4532,
+      "step": 1312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.739122018715867e-06,
+      "loss": 3.5438,
+      "step": 1313
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.73848551496051e-06,
+      "loss": 3.4717,
+      "step": 1314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.73784825651423e-06,
+      "loss": 3.4372,
+      "step": 1315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.737210243478522e-06,
+      "loss": 3.5196,
+      "step": 1316
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.736571475955003e-06,
+      "loss": 3.4753,
+      "step": 1317
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.735931954045408e-06,
+      "loss": 3.505,
+      "step": 1318
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.73529167785159e-06,
+      "loss": 3.4696,
+      "step": 1319
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.73465064747553e-06,
+      "loss": 3.4956,
+      "step": 1320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.734008863019319e-06,
+      "loss": 3.3703,
+      "step": 1321
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.733366324585174e-06,
+      "loss": 3.4223,
+      "step": 1322
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.732723032275434e-06,
+      "loss": 3.4648,
+      "step": 1323
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.732078986192552e-06,
+      "loss": 3.4886,
+      "step": 1324
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.731434186439107e-06,
+      "loss": 3.4742,
+      "step": 1325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.73078863311779e-06,
+      "loss": 3.5732,
+      "step": 1326
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.730142326331423e-06,
+      "loss": 3.5109,
+      "step": 1327
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.72949526618294e-06,
+      "loss": 3.4258,
+      "step": 1328
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.728847452775397e-06,
+      "loss": 3.4331,
+      "step": 1329
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.728198886211967e-06,
+      "loss": 3.2845,
+      "step": 1330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.727549566595955e-06,
+      "loss": 3.524,
+      "step": 1331
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.726899494030768e-06,
+      "loss": 3.4621,
+      "step": 1332
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.726248668619946e-06,
+      "loss": 3.4981,
+      "step": 1333
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.725597090467145e-06,
+      "loss": 3.5147,
+      "step": 1334
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.72494475967614e-06,
+      "loss": 3.4962,
+      "step": 1335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.724291676350826e-06,
+      "loss": 3.512,
+      "step": 1336
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.723637840595221e-06,
+      "loss": 3.3545,
+      "step": 1337
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.722983252513458e-06,
+      "loss": 3.462,
+      "step": 1338
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.722327912209795e-06,
+      "loss": 3.4499,
+      "step": 1339
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.721671819788603e-06,
+      "loss": 3.4472,
+      "step": 1340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.721014975354379e-06,
+      "loss": 3.5158,
+      "step": 1341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.720357379011736e-06,
+      "loss": 3.4344,
+      "step": 1342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.71969903086541e-06,
+      "loss": 3.5248,
+      "step": 1343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.719039931020258e-06,
+      "loss": 3.3767,
+      "step": 1344
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.718380079581246e-06,
+      "loss": 3.6087,
+      "step": 1345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.717719476653475e-06,
+      "loss": 3.4711,
+      "step": 1346
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.717058122342153e-06,
+      "loss": 3.5206,
+      "step": 1347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.716396016752616e-06,
+      "loss": 3.4798,
+      "step": 1348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.715733159990317e-06,
+      "loss": 3.4715,
+      "step": 1349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.715069552160824e-06,
+      "loss": 3.4049,
+      "step": 1350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.714405193369831e-06,
+      "loss": 3.4242,
+      "step": 1351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.713740083723152e-06,
+      "loss": 3.4571,
+      "step": 1352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.713074223326712e-06,
+      "loss": 3.5274,
+      "step": 1353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.712407612286568e-06,
+      "loss": 3.5272,
+      "step": 1354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.711740250708888e-06,
+      "loss": 3.4382,
+      "step": 1355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.711072138699957e-06,
+      "loss": 3.4092,
+      "step": 1356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.71040327636619e-06,
+      "loss": 3.4631,
+      "step": 1357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.709733663814113e-06,
+      "loss": 3.3509,
+      "step": 1358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.709063301150376e-06,
+      "loss": 3.4154,
+      "step": 1359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.708392188481744e-06,
+      "loss": 3.4034,
+      "step": 1360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.707720325915105e-06,
+      "loss": 3.4018,
+      "step": 1361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.707047713557466e-06,
+      "loss": 3.4405,
+      "step": 1362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.706374351515952e-06,
+      "loss": 3.5111,
+      "step": 1363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.705700239897809e-06,
+      "loss": 3.3904,
+      "step": 1364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.705025378810401e-06,
+      "loss": 3.4897,
+      "step": 1365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.704349768361213e-06,
+      "loss": 3.39,
+      "step": 1366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.703673408657848e-06,
+      "loss": 3.3691,
+      "step": 1367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.702996299808029e-06,
+      "loss": 3.3893,
+      "step": 1368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.702318441919595e-06,
+      "loss": 3.5798,
+      "step": 1369
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.701639835100513e-06,
+      "loss": 3.501,
+      "step": 1370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.700960479458861e-06,
+      "loss": 3.5286,
+      "step": 1371
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.700280375102835e-06,
+      "loss": 3.6095,
+      "step": 1372
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.699599522140762e-06,
+      "loss": 3.4277,
+      "step": 1373
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.698917920681072e-06,
+      "loss": 3.4491,
+      "step": 1374
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.698235570832329e-06,
+      "loss": 3.4627,
+      "step": 1375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.697552472703205e-06,
+      "loss": 3.3699,
+      "step": 1376
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.696868626402499e-06,
+      "loss": 3.5081,
+      "step": 1377
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.696184032039126e-06,
+      "loss": 3.4692,
+      "step": 1378
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.695498689722119e-06,
+      "loss": 3.4693,
+      "step": 1379
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.694812599560632e-06,
+      "loss": 3.4171,
+      "step": 1380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.694125761663936e-06,
+      "loss": 3.3728,
+      "step": 1381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.693438176141425e-06,
+      "loss": 3.441,
+      "step": 1382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.69274984310261e-06,
+      "loss": 3.4727,
+      "step": 1383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.692060762657118e-06,
+      "loss": 3.3884,
+      "step": 1384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.691370934914697e-06,
+      "loss": 3.3621,
+      "step": 1385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.69068035998522e-06,
+      "loss": 3.4586,
+      "step": 1386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.689989037978669e-06,
+      "loss": 3.3563,
+      "step": 1387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.689296969005151e-06,
+      "loss": 3.3964,
+      "step": 1388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.68860415317489e-06,
+      "loss": 3.453,
+      "step": 1389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.687910590598232e-06,
+      "loss": 3.4173,
+      "step": 1390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.687216281385636e-06,
+      "loss": 3.432,
+      "step": 1391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.686521225647689e-06,
+      "loss": 3.515,
+      "step": 1392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.685825423495086e-06,
+      "loss": 3.4612,
+      "step": 1393
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.685128875038648e-06,
+      "loss": 3.4955,
+      "step": 1394
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.684431580389313e-06,
+      "loss": 3.3561,
+      "step": 1395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.68373353965814e-06,
+      "loss": 3.4617,
+      "step": 1396
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.683034752956303e-06,
+      "loss": 3.4464,
+      "step": 1397
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.682335220395096e-06,
+      "loss": 3.4105,
+      "step": 1398
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.681634942085933e-06,
+      "loss": 3.5462,
+      "step": 1399
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.680933918140348e-06,
+      "loss": 3.3968,
+      "step": 1400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.68023214866999e-06,
+      "loss": 3.4879,
+      "step": 1401
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.67952963378663e-06,
+      "loss": 3.4181,
+      "step": 1402
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.678826373602154e-06,
+      "loss": 3.5841,
+      "step": 1403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.678122368228571e-06,
+      "loss": 3.5198,
+      "step": 1404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.677417617778006e-06,
+      "loss": 3.4778,
+      "step": 1405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.676712122362706e-06,
+      "loss": 3.457,
+      "step": 1406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.67600588209503e-06,
+      "loss": 3.535,
+      "step": 1407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.675298897087461e-06,
+      "loss": 3.3421,
+      "step": 1408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.674591167452602e-06,
+      "loss": 3.5581,
+      "step": 1409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.67388269330317e-06,
+      "loss": 3.4156,
+      "step": 1410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.673173474752003e-06,
+      "loss": 3.5766,
+      "step": 1411
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.672463511912056e-06,
+      "loss": 3.4258,
+      "step": 1412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.671752804896404e-06,
+      "loss": 3.4984,
+      "step": 1413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.67104135381824e-06,
+      "loss": 3.3631,
+      "step": 1414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.670329158790877e-06,
+      "loss": 3.366,
+      "step": 1415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.669616219927746e-06,
+      "loss": 3.4845,
+      "step": 1416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.668902537342391e-06,
+      "loss": 3.4762,
+      "step": 1417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.668188111148484e-06,
+      "loss": 3.3924,
+      "step": 1418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.667472941459808e-06,
+      "loss": 3.5447,
+      "step": 1419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.666757028390267e-06,
+      "loss": 3.2755,
+      "step": 1420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.666040372053884e-06,
+      "loss": 3.4638,
+      "step": 1421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.6653229725648e-06,
+      "loss": 3.418,
+      "step": 1422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.664604830037272e-06,
+      "loss": 3.4184,
+      "step": 1423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.66388594458568e-06,
+      "loss": 3.5322,
+      "step": 1424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.66316631632452e-06,
+      "loss": 3.461,
+      "step": 1425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.662445945368401e-06,
+      "loss": 3.4277,
+      "step": 1426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.661724831832062e-06,
+      "loss": 3.5445,
+      "step": 1427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.66100297583035e-06,
+      "loss": 3.509,
+      "step": 1428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.660280377478232e-06,
+      "loss": 3.455,
+      "step": 1429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.659557036890801e-06,
+      "loss": 3.3971,
+      "step": 1430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.658832954183258e-06,
+      "loss": 3.3916,
+      "step": 1431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.658108129470926e-06,
+      "loss": 3.4723,
+      "step": 1432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.657382562869249e-06,
+      "loss": 3.5394,
+      "step": 1433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.656656254493783e-06,
+      "loss": 3.4438,
+      "step": 1434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.655929204460212e-06,
+      "loss": 3.4679,
+      "step": 1435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.655201412884328e-06,
+      "loss": 3.4107,
+      "step": 1436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.654472879882044e-06,
+      "loss": 3.4391,
+      "step": 1437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.653743605569397e-06,
+      "loss": 3.4739,
+      "step": 1438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.653013590062532e-06,
+      "loss": 3.3851,
+      "step": 1439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.652282833477721e-06,
+      "loss": 3.5248,
+      "step": 1440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.651551335931351e-06,
+      "loss": 3.4959,
+      "step": 1441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.650819097539923e-06,
+      "loss": 3.6018,
+      "step": 1442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.65008611842006e-06,
+      "loss": 3.3471,
+      "step": 1443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.649352398688506e-06,
+      "loss": 3.3894,
+      "step": 1444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.648617938462117e-06,
+      "loss": 3.475,
+      "step": 1445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.647882737857868e-06,
+      "loss": 3.3787,
+      "step": 1446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.647146796992855e-06,
+      "loss": 3.4665,
+      "step": 1447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.64641011598429e-06,
+      "loss": 3.4221,
+      "step": 1448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.645672694949501e-06,
+      "loss": 3.4946,
+      "step": 1449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.64493453400594e-06,
+      "loss": 3.3715,
+      "step": 1450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.644195633271167e-06,
+      "loss": 3.508,
+      "step": 1451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.64345599286287e-06,
+      "loss": 3.4326,
+      "step": 1452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.64271561289885e-06,
+      "loss": 3.5239,
+      "step": 1453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.641974493497024e-06,
+      "loss": 3.4438,
+      "step": 1454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.64123263477543e-06,
+      "loss": 3.4814,
+      "step": 1455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.640490036852223e-06,
+      "loss": 3.5291,
+      "step": 1456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.639746699845676e-06,
+      "loss": 3.4684,
+      "step": 1457
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.639002623874176e-06,
+      "loss": 3.4528,
+      "step": 1458
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.638257809056235e-06,
+      "loss": 3.4625,
+      "step": 1459
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.637512255510475e-06,
+      "loss": 3.3638,
+      "step": 1460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.63676596335564e-06,
+      "loss": 3.4397,
+      "step": 1461
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.63601893271059e-06,
+      "loss": 3.4098,
+      "step": 1462
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.635271163694309e-06,
+      "loss": 3.4535,
+      "step": 1463
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.634522656425885e-06,
+      "loss": 3.3929,
+      "step": 1464
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.633773411024535e-06,
+      "loss": 3.3438,
+      "step": 1465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.633023427609591e-06,
+      "loss": 3.5417,
+      "step": 1466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.632272706300501e-06,
+      "loss": 3.4383,
+      "step": 1467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.63152124721683e-06,
+      "loss": 3.4376,
+      "step": 1468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.630769050478264e-06,
+      "loss": 3.4232,
+      "step": 1469
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.6300161162046e-06,
+      "loss": 3.4034,
+      "step": 1470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.629262444515763e-06,
+      "loss": 3.4003,
+      "step": 1471
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.628508035531785e-06,
+      "loss": 3.3568,
+      "step": 1472
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.627752889372817e-06,
+      "loss": 3.4724,
+      "step": 1473
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.626997006159135e-06,
+      "loss": 3.3691,
+      "step": 1474
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.626240386011126e-06,
+      "loss": 3.4655,
+      "step": 1475
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.625483029049295e-06,
+      "loss": 3.4634,
+      "step": 1476
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.624724935394262e-06,
+      "loss": 3.398,
+      "step": 1477
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.623966105166772e-06,
+      "loss": 3.4755,
+      "step": 1478
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.62320653848768e-06,
+      "loss": 3.5278,
+      "step": 1479
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.622446235477962e-06,
+      "loss": 3.4951,
+      "step": 1480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.621685196258712e-06,
+      "loss": 3.5589,
+      "step": 1481
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.620923420951135e-06,
+      "loss": 3.4581,
+      "step": 1482
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.620160909676562e-06,
+      "loss": 3.4287,
+      "step": 1483
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.619397662556434e-06,
+      "loss": 3.4472,
+      "step": 1484
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.618633679712315e-06,
+      "loss": 3.2897,
+      "step": 1485
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.61786896126588e-06,
+      "loss": 3.4964,
+      "step": 1486
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.617103507338927e-06,
+      "loss": 3.4786,
+      "step": 1487
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.616337318053366e-06,
+      "loss": 3.3639,
+      "step": 1488
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.61557039353123e-06,
+      "loss": 3.3972,
+      "step": 1489
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.614802733894665e-06,
+      "loss": 3.3872,
+      "step": 1490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.614034339265933e-06,
+      "loss": 3.474,
+      "step": 1491
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.613265209767417e-06,
+      "loss": 3.4379,
+      "step": 1492
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.612495345521613e-06,
+      "loss": 3.4201,
+      "step": 1493
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.61172474665114e-06,
+      "loss": 3.5388,
+      "step": 1494
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.610953413278723e-06,
+      "loss": 3.3207,
+      "step": 1495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.610181345527217e-06,
+      "loss": 3.4912,
+      "step": 1496
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.609408543519587e-06,
+      "loss": 3.5299,
+      "step": 1497
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.608635007378915e-06,
+      "loss": 3.4817,
+      "step": 1498
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.607860737228401e-06,
+      "loss": 3.484,
+      "step": 1499
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.607085733191362e-06,
+      "loss": 3.4712,
+      "step": 1500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.60630999539123e-06,
+      "loss": 3.4907,
+      "step": 1501
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.605533523951557e-06,
+      "loss": 3.476,
+      "step": 1502
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.604756318996011e-06,
+      "loss": 3.4548,
+      "step": 1503
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.603978380648375e-06,
+      "loss": 3.5299,
+      "step": 1504
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.603199709032551e-06,
+      "loss": 3.5621,
+      "step": 1505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.602420304272557e-06,
+      "loss": 3.53,
+      "step": 1506
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.601640166492526e-06,
+      "loss": 3.5359,
+      "step": 1507
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.600859295816708e-06,
+      "loss": 3.526,
+      "step": 1508
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.600077692369476e-06,
+      "loss": 3.5246,
+      "step": 1509
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.599295356275311e-06,
+      "loss": 3.5509,
+      "step": 1510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.598512287658814e-06,
+      "loss": 3.6095,
+      "step": 1511
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.597728486644707e-06,
+      "loss": 3.455,
+      "step": 1512
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.596943953357817e-06,
+      "loss": 3.6381,
+      "step": 1513
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.596158687923105e-06,
+      "loss": 3.4068,
+      "step": 1514
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.595372690465632e-06,
+      "loss": 3.5516,
+      "step": 1515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.594585961110586e-06,
+      "loss": 3.4872,
+      "step": 1516
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.593798499983265e-06,
+      "loss": 3.4085,
+      "step": 1517
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.593010307209089e-06,
+      "loss": 3.405,
+      "step": 1518
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.592221382913592e-06,
+      "loss": 3.4865,
+      "step": 1519
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.591431727222425e-06,
+      "loss": 3.4143,
+      "step": 1520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.590641340261356e-06,
+      "loss": 3.4437,
+      "step": 1521
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.589850222156264e-06,
+      "loss": 3.5004,
+      "step": 1522
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.589058373033154e-06,
+      "loss": 3.4834,
+      "step": 1523
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.588265793018141e-06,
+      "loss": 3.4454,
+      "step": 1524
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.587472482237457e-06,
+      "loss": 3.43,
+      "step": 1525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.586678440817453e-06,
+      "loss": 3.4573,
+      "step": 1526
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.585883668884594e-06,
+      "loss": 3.4496,
+      "step": 1527
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.585088166565461e-06,
+      "loss": 3.4314,
+      "step": 1528
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.584291933986754e-06,
+      "loss": 3.4219,
+      "step": 1529
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.583494971275287e-06,
+      "loss": 3.5211,
+      "step": 1530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.58269727855799e-06,
+      "loss": 3.4895,
+      "step": 1531
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.581898855961911e-06,
+      "loss": 3.3599,
+      "step": 1532
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.581099703614216e-06,
+      "loss": 3.516,
+      "step": 1533
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.58029982164218e-06,
+      "loss": 3.3723,
+      "step": 1534
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.579499210173202e-06,
+      "loss": 3.5323,
+      "step": 1535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.578697869334793e-06,
+      "loss": 3.394,
+      "step": 1536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.577895799254582e-06,
+      "loss": 3.5287,
+      "step": 1537
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.577093000060313e-06,
+      "loss": 3.4911,
+      "step": 1538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.576289471879844e-06,
+      "loss": 3.4462,
+      "step": 1539
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.575485214841158e-06,
+      "loss": 3.4727,
+      "step": 1540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.574680229072341e-06,
+      "loss": 3.548,
+      "step": 1541
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.573874514701607e-06,
+      "loss": 3.4454,
+      "step": 1542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.573068071857277e-06,
+      "loss": 3.4928,
+      "step": 1543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.572260900667794e-06,
+      "loss": 3.4052,
+      "step": 1544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.571453001261713e-06,
+      "loss": 3.4987,
+      "step": 1545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.570644373767708e-06,
+      "loss": 3.3247,
+      "step": 1546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.569835018314569e-06,
+      "loss": 3.5058,
+      "step": 1547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.569024935031198e-06,
+      "loss": 3.5092,
+      "step": 1548
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.568214124046617e-06,
+      "loss": 3.4558,
+      "step": 1549
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.567402585489964e-06,
+      "loss": 3.4599,
+      "step": 1550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.56659031949049e-06,
+      "loss": 3.5,
+      "step": 1551
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.565777326177563e-06,
+      "loss": 3.3828,
+      "step": 1552
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.564963605680668e-06,
+      "loss": 3.4015,
+      "step": 1553
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.564149158129406e-06,
+      "loss": 3.5571,
+      "step": 1554
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.563333983653489e-06,
+      "loss": 3.5283,
+      "step": 1555
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.562518082382751e-06,
+      "loss": 3.4945,
+      "step": 1556
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.561701454447138e-06,
+      "loss": 3.4549,
+      "step": 1557
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.560884099976716e-06,
+      "loss": 3.4836,
+      "step": 1558
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.56006601910166e-06,
+      "loss": 3.4954,
+      "step": 1559
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.55924721195227e-06,
+      "loss": 3.3997,
+      "step": 1560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.558427678658949e-06,
+      "loss": 3.3769,
+      "step": 1561
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.557607419352226e-06,
+      "loss": 3.4107,
+      "step": 1562
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.556786434162744e-06,
+      "loss": 3.3169,
+      "step": 1563
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.555964723221258e-06,
+      "loss": 3.3941,
+      "step": 1564
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.55514228665864e-06,
+      "loss": 3.4335,
+      "step": 1565
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.55431912460588e-06,
+      "loss": 3.4836,
+      "step": 1566
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.553495237194081e-06,
+      "loss": 3.3974,
+      "step": 1567
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.55267062455446e-06,
+      "loss": 3.4881,
+      "step": 1568
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.551845286818355e-06,
+      "loss": 3.5117,
+      "step": 1569
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.551019224117216e-06,
+      "loss": 3.4821,
+      "step": 1570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.550192436582606e-06,
+      "loss": 3.5004,
+      "step": 1571
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.54936492434621e-06,
+      "loss": 3.4472,
+      "step": 1572
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.54853668753982e-06,
+      "loss": 3.3153,
+      "step": 1573
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.54770772629535e-06,
+      "loss": 3.3375,
+      "step": 1574
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.546878040744829e-06,
+      "loss": 3.5658,
+      "step": 1575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.546047631020398e-06,
+      "loss": 3.3642,
+      "step": 1576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.545216497254315e-06,
+      "loss": 3.5318,
+      "step": 1577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.544384639578952e-06,
+      "loss": 3.4057,
+      "step": 1578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.5435520581268e-06,
+      "loss": 3.4274,
+      "step": 1579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.542718753030463e-06,
+      "loss": 3.4356,
+      "step": 1580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.541884724422659e-06,
+      "loss": 3.4502,
+      "step": 1581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.541049972436222e-06,
+      "loss": 3.4853,
+      "step": 1582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.540214497204104e-06,
+      "loss": 3.5239,
+      "step": 1583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.539378298859365e-06,
+      "loss": 3.4669,
+      "step": 1584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.53854137753519e-06,
+      "loss": 3.4267,
+      "step": 1585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.537703733364871e-06,
+      "loss": 3.4785,
+      "step": 1586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.53686536648182e-06,
+      "loss": 3.3925,
+      "step": 1587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.536026277019562e-06,
+      "loss": 3.3279,
+      "step": 1588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.535186465111736e-06,
+      "loss": 3.5955,
+      "step": 1589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.534345930892097e-06,
+      "loss": 3.5124,
+      "step": 1590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.53350467449452e-06,
+      "loss": 3.3901,
+      "step": 1591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.532662696052986e-06,
+      "loss": 3.553,
+      "step": 1592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.531819995701598e-06,
+      "loss": 3.4749,
+      "step": 1593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.53097657357457e-06,
+      "loss": 3.5598,
+      "step": 1594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.530132429806233e-06,
+      "loss": 3.5222,
+      "step": 1595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.529287564531034e-06,
+      "loss": 3.4998,
+      "step": 1596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.528441977883531e-06,
+      "loss": 3.3684,
+      "step": 1597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.527595669998399e-06,
+      "loss": 3.4474,
+      "step": 1598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.52674864101043e-06,
+      "loss": 3.4631,
+      "step": 1599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.525900891054528e-06,
+      "loss": 3.3611,
+      "step": 1600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.525052420265714e-06,
+      "loss": 3.4565,
+      "step": 1601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.52420322877912e-06,
+      "loss": 3.4952,
+      "step": 1602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.523353316729997e-06,
+      "loss": 3.4174,
+      "step": 1603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.522502684253709e-06,
+      "loss": 3.3343,
+      "step": 1604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.521651331485736e-06,
+      "loss": 3.3548,
+      "step": 1605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.520799258561668e-06,
+      "loss": 3.471,
+      "step": 1606
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.519946465617217e-06,
+      "loss": 3.3434,
+      "step": 1607
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.519092952788206e-06,
+      "loss": 3.5006,
+      "step": 1608
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.518238720210569e-06,
+      "loss": 3.359,
+      "step": 1609
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.51738376802036e-06,
+      "loss": 3.3602,
+      "step": 1610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.516528096353748e-06,
+      "loss": 3.4995,
+      "step": 1611
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.515671705347012e-06,
+      "loss": 3.3935,
+      "step": 1612
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.51481459513655e-06,
+      "loss": 3.394,
+      "step": 1613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.51395676585887e-06,
+      "loss": 3.5472,
+      "step": 1614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.513098217650598e-06,
+      "loss": 3.4852,
+      "step": 1615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.512238950648474e-06,
+      "loss": 3.4442,
+      "step": 1616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.511378964989353e-06,
+      "loss": 3.495,
+      "step": 1617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.510518260810202e-06,
+      "loss": 3.4554,
+      "step": 1618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.509656838248105e-06,
+      "loss": 3.4343,
+      "step": 1619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.508794697440257e-06,
+      "loss": 3.4361,
+      "step": 1620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.507931838523973e-06,
+      "loss": 3.47,
+      "step": 1621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.50706826163668e-06,
+      "loss": 3.4801,
+      "step": 1622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.506203966915913e-06,
+      "loss": 3.4703,
+      "step": 1623
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.505338954499332e-06,
+      "loss": 3.5201,
+      "step": 1624
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.504473224524704e-06,
+      "loss": 3.3995,
+      "step": 1625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.503606777129913e-06,
+      "loss": 3.4369,
+      "step": 1626
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.502739612452954e-06,
+      "loss": 3.4852,
+      "step": 1627
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.501871730631944e-06,
+      "loss": 3.5216,
+      "step": 1628
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.501003131805104e-06,
+      "loss": 3.4237,
+      "step": 1629
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.500133816110779e-06,
+      "loss": 3.3798,
+      "step": 1630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.499263783687418e-06,
+      "loss": 3.422,
+      "step": 1631
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.498393034673597e-06,
+      "loss": 3.5283,
+      "step": 1632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.497521569207993e-06,
+      "loss": 3.349,
+      "step": 1633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.496649387429405e-06,
+      "loss": 3.3376,
+      "step": 1634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.49577648947674e-06,
+      "loss": 3.4415,
+      "step": 1635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.494902875489031e-06,
+      "loss": 3.3951,
+      "step": 1636
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.49402854560541e-06,
+      "loss": 3.353,
+      "step": 1637
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.493153499965135e-06,
+      "loss": 3.4273,
+      "step": 1638
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.49227773870757e-06,
+      "loss": 3.5449,
+      "step": 1639
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.491401261972194e-06,
+      "loss": 3.5778,
+      "step": 1640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.49052406989861e-06,
+      "loss": 3.4625,
+      "step": 1641
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.489646162626518e-06,
+      "loss": 3.2548,
+      "step": 1642
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.488767540295747e-06,
+      "loss": 3.4444,
+      "step": 1643
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.487888203046232e-06,
+      "loss": 3.4711,
+      "step": 1644
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.487008151018022e-06,
+      "loss": 3.3894,
+      "step": 1645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.486127384351282e-06,
+      "loss": 3.4796,
+      "step": 1646
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.48524590318629e-06,
+      "loss": 3.2984,
+      "step": 1647
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.484363707663443e-06,
+      "loss": 3.4473,
+      "step": 1648
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.483480797923241e-06,
+      "loss": 3.4429,
+      "step": 1649
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.482597174106306e-06,
+      "loss": 3.5645,
+      "step": 1650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.481712836353368e-06,
+      "loss": 3.4747,
+      "step": 1651
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.480827784805278e-06,
+      "loss": 3.493,
+      "step": 1652
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.479942019602997e-06,
+      "loss": 3.5521,
+      "step": 1653
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.479055540887599e-06,
+      "loss": 3.496,
+      "step": 1654
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.478168348800268e-06,
+      "loss": 3.4824,
+      "step": 1655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.477280443482311e-06,
+      "loss": 3.5235,
+      "step": 1656
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.47639182507514e-06,
+      "loss": 3.422,
+      "step": 1657
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.475502493720284e-06,
+      "loss": 3.4899,
+      "step": 1658
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.474612449559386e-06,
+      "loss": 3.5027,
+      "step": 1659
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.473721692734203e-06,
+      "loss": 3.3927,
+      "step": 1660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.472830223386604e-06,
+      "loss": 3.5554,
+      "step": 1661
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.47193804165857e-06,
+      "loss": 3.4937,
+      "step": 1662
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.4710451476922e-06,
+      "loss": 3.39,
+      "step": 1663
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.4701515416297e-06,
+      "loss": 3.4916,
+      "step": 1664
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.469257223613398e-06,
+      "loss": 3.5209,
+      "step": 1665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.468362193785726e-06,
+      "loss": 3.4727,
+      "step": 1666
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.467466452289238e-06,
+      "loss": 3.4022,
+      "step": 1667
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.466569999266595e-06,
+      "loss": 3.4757,
+      "step": 1668
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.465672834860574e-06,
+      "loss": 3.3906,
+      "step": 1669
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.464774959214062e-06,
+      "loss": 3.5133,
+      "step": 1670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.463876372470069e-06,
+      "loss": 3.3745,
+      "step": 1671
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.462977074771706e-06,
+      "loss": 3.497,
+      "step": 1672
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.462077066262206e-06,
+      "loss": 3.4046,
+      "step": 1673
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.461176347084909e-06,
+      "loss": 3.4735,
+      "step": 1674
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.46027491738327e-06,
+      "loss": 3.3751,
+      "step": 1675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.459372777300863e-06,
+      "loss": 3.4883,
+      "step": 1676
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.458469926981368e-06,
+      "loss": 3.4713,
+      "step": 1677
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.45756636656858e-06,
+      "loss": 3.4068,
+      "step": 1678
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.456662096206409e-06,
+      "loss": 3.4922,
+      "step": 1679
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.455757116038873e-06,
+      "loss": 3.4066,
+      "step": 1680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.454851426210112e-06,
+      "loss": 3.5047,
+      "step": 1681
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.45394502686437e-06,
+      "loss": 3.5107,
+      "step": 1682
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.45303791814601e-06,
+      "loss": 3.5373,
+      "step": 1683
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.452130100199504e-06,
+      "loss": 3.5028,
+      "step": 1684
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.45122157316944e-06,
+      "loss": 3.4652,
+      "step": 1685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.450312337200516e-06,
+      "loss": 3.6264,
+      "step": 1686
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.449402392437546e-06,
+      "loss": 3.4227,
+      "step": 1687
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.448491739025455e-06,
+      "loss": 3.4528,
+      "step": 1688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.447580377109281e-06,
+      "loss": 3.582,
+      "step": 1689
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.446668306834176e-06,
+      "loss": 3.4898,
+      "step": 1690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.445755528345402e-06,
+      "loss": 3.4846,
+      "step": 1691
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.44484204178834e-06,
+      "loss": 3.4385,
+      "step": 1692
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.443927847308474e-06,
+      "loss": 3.4421,
+      "step": 1693
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.44301294505141e-06,
+      "loss": 3.4782,
+      "step": 1694
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.442097335162862e-06,
+      "loss": 3.4554,
+      "step": 1695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.441181017788658e-06,
+      "loss": 3.3501,
+      "step": 1696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.440263993074736e-06,
+      "loss": 3.5909,
+      "step": 1697
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.439346261167153e-06,
+      "loss": 3.4802,
+      "step": 1698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.438427822212073e-06,
+      "loss": 3.423,
+      "step": 1699
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.437508676355774e-06,
+      "loss": 3.3736,
+      "step": 1700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.436588823744646e-06,
+      "loss": 3.517,
+      "step": 1701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.435668264525194e-06,
+      "loss": 3.4427,
+      "step": 1702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.43474699884403e-06,
+      "loss": 3.4575,
+      "step": 1703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.433825026847891e-06,
+      "loss": 3.5229,
+      "step": 1704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.432902348683611e-06,
+      "loss": 3.429,
+      "step": 1705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.431978964498143e-06,
+      "loss": 3.476,
+      "step": 1706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.431054874438559e-06,
+      "loss": 3.3885,
+      "step": 1707
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.43013007865203e-06,
+      "loss": 3.5045,
+      "step": 1708
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.429204577285853e-06,
+      "loss": 3.4917,
+      "step": 1709
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.428278370487428e-06,
+      "loss": 3.4566,
+      "step": 1710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.427351458404268e-06,
+      "loss": 3.4757,
+      "step": 1711
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.426423841184007e-06,
+      "loss": 3.4279,
+      "step": 1712
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.425495518974379e-06,
+      "loss": 3.4802,
+      "step": 1713
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.42456649192324e-06,
+      "loss": 3.3996,
+      "step": 1714
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.423636760178554e-06,
+      "loss": 3.4106,
+      "step": 1715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.422706323888398e-06,
+      "loss": 3.5078,
+      "step": 1716
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.421775183200962e-06,
+      "loss": 3.5088,
+      "step": 1717
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.420843338264543e-06,
+      "loss": 3.4849,
+      "step": 1718
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.419910789227558e-06,
+      "loss": 3.477,
+      "step": 1719
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.418977536238532e-06,
+      "loss": 3.5368,
+      "step": 1720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.418043579446103e-06,
+      "loss": 3.4411,
+      "step": 1721
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.41710891899902e-06,
+      "loss": 3.5097,
+      "step": 1722
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.416173555046146e-06,
+      "loss": 3.5108,
+      "step": 1723
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.415237487736452e-06,
+      "loss": 3.4217,
+      "step": 1724
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.41430071721903e-06,
+      "loss": 3.4218,
+      "step": 1725
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.413363243643072e-06,
+      "loss": 3.5008,
+      "step": 1726
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.412425067157888e-06,
+      "loss": 3.4637,
+      "step": 1727
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.411486187912904e-06,
+      "loss": 3.5167,
+      "step": 1728
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.41054660605765e-06,
+      "loss": 3.4902,
+      "step": 1729
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.409606321741776e-06,
+      "loss": 3.4924,
+      "step": 1730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.408665335115034e-06,
+      "loss": 3.4555,
+      "step": 1731
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.407723646327299e-06,
+      "loss": 3.4085,
+      "step": 1732
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.406781255528548e-06,
+      "loss": 3.4685,
+      "step": 1733
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.405838162868874e-06,
+      "loss": 3.4774,
+      "step": 1734
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.404894368498484e-06,
+      "loss": 3.4257,
+      "step": 1735
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.403949872567696e-06,
+      "loss": 3.5332,
+      "step": 1736
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.403004675226934e-06,
+      "loss": 3.5373,
+      "step": 1737
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.402058776626743e-06,
+      "loss": 3.4395,
+      "step": 1738
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.401112176917771e-06,
+      "loss": 3.4873,
+      "step": 1739
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.400164876250781e-06,
+      "loss": 3.4809,
+      "step": 1740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.39921687477665e-06,
+      "loss": 3.466,
+      "step": 1741
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.398268172646365e-06,
+      "loss": 3.5039,
+      "step": 1742
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.397318770011023e-06,
+      "loss": 3.3929,
+      "step": 1743
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.396368667021835e-06,
+      "loss": 3.4413,
+      "step": 1744
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.39541786383012e-06,
+      "loss": 3.4101,
+      "step": 1745
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.394466360587313e-06,
+      "loss": 3.3961,
+      "step": 1746
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.39351415744496e-06,
+      "loss": 3.414,
+      "step": 1747
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.392561254554712e-06,
+      "loss": 3.4597,
+      "step": 1748
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.39160765206834e-06,
+      "loss": 3.4509,
+      "step": 1749
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.390653350137724e-06,
+      "loss": 3.4183,
+      "step": 1750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.389698348914849e-06,
+      "loss": 3.5211,
+      "step": 1751
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.38874264855182e-06,
+      "loss": 3.4171,
+      "step": 1752
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.387786249200849e-06,
+      "loss": 3.393,
+      "step": 1753
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.386829151014262e-06,
+      "loss": 3.5615,
+      "step": 1754
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.385871354144494e-06,
+      "loss": 3.5536,
+      "step": 1755
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.384912858744088e-06,
+      "loss": 3.3548,
+      "step": 1756
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.383953664965706e-06,
+      "loss": 3.4362,
+      "step": 1757
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.382993772962114e-06,
+      "loss": 3.4775,
+      "step": 1758
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.382033182886197e-06,
+      "loss": 3.4226,
+      "step": 1759
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.381071894890942e-06,
+      "loss": 3.5648,
+      "step": 1760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.380109909129454e-06,
+      "loss": 3.3553,
+      "step": 1761
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.379147225754948e-06,
+      "loss": 3.4438,
+      "step": 1762
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.378183844920748e-06,
+      "loss": 3.6235,
+      "step": 1763
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.377219766780288e-06,
+      "loss": 3.4751,
+      "step": 1764
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.376254991487118e-06,
+      "loss": 3.3635,
+      "step": 1765
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.375289519194894e-06,
+      "loss": 3.3497,
+      "step": 1766
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.374323350057388e-06,
+      "loss": 3.5321,
+      "step": 1767
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.373356484228479e-06,
+      "loss": 3.4208,
+      "step": 1768
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.372388921862156e-06,
+      "loss": 3.3484,
+      "step": 1769
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.371420663112526e-06,
+      "loss": 3.4508,
+      "step": 1770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.370451708133795e-06,
+      "loss": 3.397,
+      "step": 1771
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.369482057080293e-06,
+      "loss": 3.4585,
+      "step": 1772
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.368511710106451e-06,
+      "loss": 3.4006,
+      "step": 1773
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.367540667366816e-06,
+      "loss": 3.4517,
+      "step": 1774
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.366568929016047e-06,
+      "loss": 3.4317,
+      "step": 1775
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.365596495208907e-06,
+      "loss": 3.4356,
+      "step": 1776
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.364623366100276e-06,
+      "loss": 3.4264,
+      "step": 1777
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.363649541845142e-06,
+      "loss": 3.3863,
+      "step": 1778
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.362675022598606e-06,
+      "loss": 3.4579,
+      "step": 1779
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.361699808515877e-06,
+      "loss": 3.4142,
+      "step": 1780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.360723899752272e-06,
+      "loss": 3.406,
+      "step": 1781
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.35974729646323e-06,
+      "loss": 3.4084,
+      "step": 1782
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.358769998804288e-06,
+      "loss": 3.4476,
+      "step": 1783
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.3577920069311e-06,
+      "loss": 3.4196,
+      "step": 1784
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.356813320999428e-06,
+      "loss": 3.4942,
+      "step": 1785
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.355833941165146e-06,
+      "loss": 3.4407,
+      "step": 1786
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.35485386758424e-06,
+      "loss": 3.4161,
+      "step": 1787
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.353873100412805e-06,
+      "loss": 3.5133,
+      "step": 1788
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.352891639807042e-06,
+      "loss": 3.5176,
+      "step": 1789
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.35190948592327e-06,
+      "loss": 3.5409,
+      "step": 1790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.350926638917916e-06,
+      "loss": 3.5602,
+      "step": 1791
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.349943098947513e-06,
+      "loss": 3.4934,
+      "step": 1792
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.34895886616871e-06,
+      "loss": 3.4793,
+      "step": 1793
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.347973940738266e-06,
+      "loss": 3.4349,
+      "step": 1794
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.346988322813046e-06,
+      "loss": 3.4569,
+      "step": 1795
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.346002012550027e-06,
+      "loss": 3.465,
+      "step": 1796
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.345015010106298e-06,
+      "loss": 3.4048,
+      "step": 1797
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.34402731563906e-06,
+      "loss": 3.3901,
+      "step": 1798
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.343038929305617e-06,
+      "loss": 3.4862,
+      "step": 1799
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.342049851263388e-06,
+      "loss": 3.534,
+      "step": 1800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.341060081669905e-06,
+      "loss": 3.4399,
+      "step": 1801
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.340069620682807e-06,
+      "loss": 3.5801,
+      "step": 1802
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.339078468459842e-06,
+      "loss": 3.3958,
+      "step": 1803
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.338086625158867e-06,
+      "loss": 3.3782,
+      "step": 1804
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.337094090937853e-06,
+      "loss": 3.4461,
+      "step": 1805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.336100865954881e-06,
+      "loss": 3.3463,
+      "step": 1806
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.335106950368137e-06,
+      "loss": 3.3948,
+      "step": 1807
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.334112344335924e-06,
+      "loss": 3.4763,
+      "step": 1808
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.333117048016648e-06,
+      "loss": 3.4082,
+      "step": 1809
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.33212106156883e-06,
+      "loss": 3.5696,
+      "step": 1810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.331124385151099e-06,
+      "loss": 3.4307,
+      "step": 1811
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.330127018922195e-06,
+      "loss": 3.2694,
+      "step": 1812
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.329128963040964e-06,
+      "loss": 3.4603,
+      "step": 1813
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.328130217666366e-06,
+      "loss": 3.4456,
+      "step": 1814
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.327130782957473e-06,
+      "loss": 3.459,
+      "step": 1815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.326130659073457e-06,
+      "loss": 3.4529,
+      "step": 1816
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.325129846173613e-06,
+      "loss": 3.4782,
+      "step": 1817
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.324128344417333e-06,
+      "loss": 3.3791,
+      "step": 1818
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.323126153964129e-06,
+      "loss": 3.4053,
+      "step": 1819
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.322123274973613e-06,
+      "loss": 3.5233,
+      "step": 1820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.321119707605517e-06,
+      "loss": 3.3746,
+      "step": 1821
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.320115452019676e-06,
+      "loss": 3.3373,
+      "step": 1822
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.319110508376035e-06,
+      "loss": 3.4778,
+      "step": 1823
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.318104876834652e-06,
+      "loss": 3.4718,
+      "step": 1824
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.31709855755569e-06,
+      "loss": 3.4056,
+      "step": 1825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.316091550699424e-06,
+      "loss": 3.4894,
+      "step": 1826
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.31508385642624e-06,
+      "loss": 3.463,
+      "step": 1827
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.314075474896631e-06,
+      "loss": 3.4247,
+      "step": 1828
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.3130664062712e-06,
+      "loss": 3.5452,
+      "step": 1829
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.312056650710661e-06,
+      "loss": 3.4511,
+      "step": 1830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.311046208375833e-06,
+      "loss": 3.5717,
+      "step": 1831
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.310035079427651e-06,
+      "loss": 3.4159,
+      "step": 1832
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.309023264027154e-06,
+      "loss": 3.4685,
+      "step": 1833
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.308010762335492e-06,
+      "loss": 3.5542,
+      "step": 1834
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.306997574513926e-06,
+      "loss": 3.4457,
+      "step": 1835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.305983700723824e-06,
+      "loss": 3.5766,
+      "step": 1836
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.304969141126665e-06,
+      "loss": 3.4575,
+      "step": 1837
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.303953895884033e-06,
+      "loss": 3.5158,
+      "step": 1838
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.302937965157628e-06,
+      "loss": 3.4064,
+      "step": 1839
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.301921349109255e-06,
+      "loss": 3.4492,
+      "step": 1840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.300904047900826e-06,
+      "loss": 3.4269,
+      "step": 1841
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.29988606169437e-06,
+      "loss": 3.4338,
+      "step": 1842
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.298867390652015e-06,
+      "loss": 3.4566,
+      "step": 1843
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.297848034936007e-06,
+      "loss": 3.3816,
+      "step": 1844
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.296827994708694e-06,
+      "loss": 3.5529,
+      "step": 1845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.295807270132538e-06,
+      "loss": 3.5293,
+      "step": 1846
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.294785861370107e-06,
+      "loss": 3.3638,
+      "step": 1847
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.29376376858408e-06,
+      "loss": 3.4757,
+      "step": 1848
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.292740991937243e-06,
+      "loss": 3.485,
+      "step": 1849
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.291717531592495e-06,
+      "loss": 3.3731,
+      "step": 1850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.290693387712836e-06,
+      "loss": 3.5283,
+      "step": 1851
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.289668560461385e-06,
+      "loss": 3.4347,
+      "step": 1852
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.288643050001362e-06,
+      "loss": 3.4216,
+      "step": 1853
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.287616856496097e-06,
+      "loss": 3.477,
+      "step": 1854
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.286589980109032e-06,
+      "loss": 3.5188,
+      "step": 1855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.285562421003716e-06,
+      "loss": 3.5334,
+      "step": 1856
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.284534179343804e-06,
+      "loss": 3.4506,
+      "step": 1857
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.283505255293066e-06,
+      "loss": 3.4442,
+      "step": 1858
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.282475649015377e-06,
+      "loss": 3.4047,
+      "step": 1859
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.281445360674717e-06,
+      "loss": 3.5739,
+      "step": 1860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.280414390435182e-06,
+      "loss": 3.4585,
+      "step": 1861
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.279382738460971e-06,
+      "loss": 3.5686,
+      "step": 1862
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.278350404916397e-06,
+      "loss": 3.4232,
+      "step": 1863
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.277317389965871e-06,
+      "loss": 3.3437,
+      "step": 1864
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.276283693773926e-06,
+      "loss": 3.486,
+      "step": 1865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.275249316505195e-06,
+      "loss": 3.5298,
+      "step": 1866
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.274214258324424e-06,
+      "loss": 3.4545,
+      "step": 1867
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.273178519396459e-06,
+      "loss": 3.4365,
+      "step": 1868
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.272142099886266e-06,
+      "loss": 3.4547,
+      "step": 1869
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.271104999958913e-06,
+      "loss": 3.4041,
+      "step": 1870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.270067219779574e-06,
+      "loss": 3.4644,
+      "step": 1871
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.269028759513538e-06,
+      "loss": 3.4402,
+      "step": 1872
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.267989619326196e-06,
+      "loss": 3.4785,
+      "step": 1873
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.266949799383053e-06,
+      "loss": 3.4803,
+      "step": 1874
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.265909299849721e-06,
+      "loss": 3.5251,
+      "step": 1875
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.264868120891913e-06,
+      "loss": 3.4441,
+      "step": 1876
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.26382626267546e-06,
+      "loss": 3.5023,
+      "step": 1877
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.262783725366296e-06,
+      "loss": 3.4949,
+      "step": 1878
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.261740509130463e-06,
+      "loss": 3.4277,
+      "step": 1879
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.260696614134115e-06,
+      "loss": 3.411,
+      "step": 1880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.259652040543509e-06,
+      "loss": 3.4236,
+      "step": 1881
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.258606788525013e-06,
+      "loss": 3.4142,
+      "step": 1882
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.257560858245105e-06,
+      "loss": 3.4725,
+      "step": 1883
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.256514249870366e-06,
+      "loss": 3.5779,
+      "step": 1884
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.255466963567488e-06,
+      "loss": 3.3892,
+      "step": 1885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.254418999503272e-06,
+      "loss": 3.4859,
+      "step": 1886
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.253370357844622e-06,
+      "loss": 3.466,
+      "step": 1887
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.252321038758558e-06,
+      "loss": 3.4309,
+      "step": 1888
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.251271042412201e-06,
+      "loss": 3.385,
+      "step": 1889
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.250220368972784e-06,
+      "loss": 3.5143,
+      "step": 1890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.249169018607643e-06,
+      "loss": 3.4674,
+      "step": 1891
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.24811699148423e-06,
+      "loss": 3.3586,
+      "step": 1892
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.24706428777009e-06,
+      "loss": 3.4363,
+      "step": 1893
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.246010907632894e-06,
+      "loss": 3.4694,
+      "step": 1894
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.244956851240412e-06,
+      "loss": 3.4339,
+      "step": 1895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.243902118760518e-06,
+      "loss": 3.361,
+      "step": 1896
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.2428467103612e-06,
+      "loss": 3.5157,
+      "step": 1897
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.24179062621055e-06,
+      "loss": 3.4707,
+      "step": 1898
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.240733866476768e-06,
+      "loss": 3.415,
+      "step": 1899
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.239676431328164e-06,
+      "loss": 3.5588,
+      "step": 1900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.238618320933153e-06,
+      "loss": 3.4274,
+      "step": 1901
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.23755953546026e-06,
+      "loss": 3.5258,
+      "step": 1902
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.236500075078114e-06,
+      "loss": 3.4685,
+      "step": 1903
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.235439939955458e-06,
+      "loss": 3.4666,
+      "step": 1904
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.234379130261131e-06,
+      "loss": 3.3785,
+      "step": 1905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.23331764616409e-06,
+      "loss": 3.5411,
+      "step": 1906
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.2322554878334e-06,
+      "loss": 3.3428,
+      "step": 1907
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.231192655438222e-06,
+      "loss": 3.4591,
+      "step": 1908
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.230129149147836e-06,
+      "loss": 3.3695,
+      "step": 1909
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.229064969131621e-06,
+      "loss": 3.4407,
+      "step": 1910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.228000115559072e-06,
+      "loss": 3.4765,
+      "step": 1911
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.226934588599786e-06,
+      "loss": 3.4715,
+      "step": 1912
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.225868388423463e-06,
+      "loss": 3.3966,
+      "step": 1913
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.22480151519992e-06,
+      "loss": 3.3828,
+      "step": 1914
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.223733969099075e-06,
+      "loss": 3.4875,
+      "step": 1915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.222665750290953e-06,
+      "loss": 3.4997,
+      "step": 1916
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.221596858945687e-06,
+      "loss": 3.5242,
+      "step": 1917
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.22052729523352e-06,
+      "loss": 3.4739,
+      "step": 1918
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.2194570593248e-06,
+      "loss": 3.4344,
+      "step": 1919
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.218386151389978e-06,
+      "loss": 3.4387,
+      "step": 1920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.217314571599617e-06,
+      "loss": 3.4067,
+      "step": 1921
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.21624232012439e-06,
+      "loss": 3.4288,
+      "step": 1922
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.215169397135066e-06,
+      "loss": 3.5117,
+      "step": 1923
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.214095802802533e-06,
+      "loss": 3.4263,
+      "step": 1924
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.21302153729778e-06,
+      "loss": 3.3966,
+      "step": 1925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.2119466007919e-06,
+      "loss": 3.544,
+      "step": 1926
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.210870993456102e-06,
+      "loss": 3.3748,
+      "step": 1927
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.209794715461691e-06,
+      "loss": 3.4761,
+      "step": 1928
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.208717766980086e-06,
+      "loss": 3.3868,
+      "step": 1929
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.207640148182811e-06,
+      "loss": 3.5482,
+      "step": 1930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.206561859241497e-06,
+      "loss": 3.4692,
+      "step": 1931
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.20548290032788e-06,
+      "loss": 3.3007,
+      "step": 1932
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.204403271613804e-06,
+      "loss": 3.4704,
+      "step": 1933
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.203322973271223e-06,
+      "loss": 3.4626,
+      "step": 1934
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.20224200547219e-06,
+      "loss": 3.4303,
+      "step": 1935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.201160368388872e-06,
+      "loss": 3.4052,
+      "step": 1936
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.20007806219354e-06,
+      "loss": 3.4236,
+      "step": 1937
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.198995087058567e-06,
+      "loss": 3.572,
+      "step": 1938
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.197911443156442e-06,
+      "loss": 3.4839,
+      "step": 1939
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.196827130659752e-06,
+      "loss": 3.5234,
+      "step": 1940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.195742149741192e-06,
+      "loss": 3.4829,
+      "step": 1941
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.19465650057357e-06,
+      "loss": 3.399,
+      "step": 1942
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.193570183329793e-06,
+      "loss": 3.417,
+      "step": 1943
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.192483198182876e-06,
+      "loss": 3.4913,
+      "step": 1944
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.191395545305942e-06,
+      "loss": 3.499,
+      "step": 1945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.190307224872221e-06,
+      "loss": 3.368,
+      "step": 1946
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.189218237055046e-06,
+      "loss": 3.4503,
+      "step": 1947
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.18812858202786e-06,
+      "loss": 3.426,
+      "step": 1948
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.187038259964211e-06,
+      "loss": 3.4272,
+      "step": 1949
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.185947271037751e-06,
+      "loss": 3.4375,
+      "step": 1950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.184855615422243e-06,
+      "loss": 3.3222,
+      "step": 1951
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.18376329329155e-06,
+      "loss": 3.3385,
+      "step": 1952
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.182670304819645e-06,
+      "loss": 3.4152,
+      "step": 1953
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.181576650180606e-06,
+      "loss": 3.4497,
+      "step": 1954
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.180482329548621e-06,
+      "loss": 3.4382,
+      "step": 1955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.179387343097978e-06,
+      "loss": 3.603,
+      "step": 1956
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.178291691003073e-06,
+      "loss": 3.4745,
+      "step": 1957
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.17719537343841e-06,
+      "loss": 3.3332,
+      "step": 1958
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.176098390578598e-06,
+      "loss": 3.4368,
+      "step": 1959
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.175000742598351e-06,
+      "loss": 3.3957,
+      "step": 1960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.17390242967249e-06,
+      "loss": 3.4806,
+      "step": 1961
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.172803451975942e-06,
+      "loss": 3.4346,
+      "step": 1962
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.171703809683738e-06,
+      "loss": 3.3913,
+      "step": 1963
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.170603502971017e-06,
+      "loss": 3.3647,
+      "step": 1964
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.169502532013023e-06,
+      "loss": 3.4695,
+      "step": 1965
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.168400896985106e-06,
+      "loss": 3.3995,
+      "step": 1966
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.16729859806272e-06,
+      "loss": 3.4087,
+      "step": 1967
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.166195635421432e-06,
+      "loss": 3.4034,
+      "step": 1968
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.165092009236902e-06,
+      "loss": 3.406,
+      "step": 1969
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.163987719684908e-06,
+      "loss": 3.4854,
+      "step": 1970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.162882766941324e-06,
+      "loss": 3.3051,
+      "step": 1971
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.161777151182137e-06,
+      "loss": 3.4705,
+      "step": 1972
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.160670872583435e-06,
+      "loss": 3.4224,
+      "step": 1973
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.159563931321416e-06,
+      "loss": 3.5323,
+      "step": 1974
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.158456327572378e-06,
+      "loss": 3.4706,
+      "step": 1975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.157348061512728e-06,
+      "loss": 3.4931,
+      "step": 1976
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.156239133318977e-06,
+      "loss": 3.542,
+      "step": 1977
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.155129543167744e-06,
+      "loss": 3.4376,
+      "step": 1978
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.15401929123575e-06,
+      "loss": 3.4848,
+      "step": 1979
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.152908377699823e-06,
+      "loss": 3.4072,
+      "step": 1980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.151796802736899e-06,
+      "loss": 3.3623,
+      "step": 1981
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.150684566524012e-06,
+      "loss": 3.3872,
+      "step": 1982
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.149571669238311e-06,
+      "loss": 3.5051,
+      "step": 1983
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.148458111057043e-06,
+      "loss": 3.4587,
+      "step": 1984
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.147343892157563e-06,
+      "loss": 3.4592,
+      "step": 1985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.14622901271733e-06,
+      "loss": 3.449,
+      "step": 1986
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.145113472913911e-06,
+      "loss": 3.4943,
+      "step": 1987
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.143997272924974e-06,
+      "loss": 3.4933,
+      "step": 1988
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.142880412928295e-06,
+      "loss": 3.3583,
+      "step": 1989
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.141762893101755e-06,
+      "loss": 3.3173,
+      "step": 1990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.14064471362334e-06,
+      "loss": 3.6244,
+      "step": 1991
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.139525874671139e-06,
+      "loss": 3.4444,
+      "step": 1992
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.13840637642335e-06,
+      "loss": 3.3589,
+      "step": 1993
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.13728621905827e-06,
+      "loss": 3.427,
+      "step": 1994
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.136165402754309e-06,
+      "loss": 3.4023,
+      "step": 1995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.135043927689975e-06,
+      "loss": 3.493,
+      "step": 1996
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.133921794043886e-06,
+      "loss": 3.5089,
+      "step": 1997
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.132799001994756e-06,
+      "loss": 3.4442,
+      "step": 1998
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.131675551721418e-06,
+      "loss": 3.3943,
+      "step": 1999
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.1305514434028e-06,
+      "loss": 3.3765,
+      "step": 2000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.129426677217934e-06,
+      "loss": 3.4983,
+      "step": 2001
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.128301253345962e-06,
+      "loss": 3.3509,
+      "step": 2002
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.127175171966128e-06,
+      "loss": 3.5457,
+      "step": 2003
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.12604843325778e-06,
+      "loss": 3.4392,
+      "step": 2004
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.124921037400374e-06,
+      "loss": 3.4599,
+      "step": 2005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.123792984573468e-06,
+      "loss": 3.4418,
+      "step": 2006
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.122664274956723e-06,
+      "loss": 3.4679,
+      "step": 2007
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.12153490872991e-06,
+      "loss": 3.4904,
+      "step": 2008
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.120404886072898e-06,
+      "loss": 3.4048,
+      "step": 2009
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.119274207165666e-06,
+      "loss": 3.3675,
+      "step": 2010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.118142872188298e-06,
+      "loss": 3.4657,
+      "step": 2011
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.117010881320973e-06,
+      "loss": 3.4775,
+      "step": 2012
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.115878234743987e-06,
+      "loss": 3.3831,
+      "step": 2013
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.114744932637734e-06,
+      "loss": 3.5562,
+      "step": 2014
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.11361097518271e-06,
+      "loss": 3.4742,
+      "step": 2015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.112476362559521e-06,
+      "loss": 3.3984,
+      "step": 2016
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.111341094948876e-06,
+      "loss": 3.4827,
+      "step": 2017
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.110205172531585e-06,
+      "loss": 3.4677,
+      "step": 2018
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.109068595488566e-06,
+      "loss": 3.3897,
+      "step": 2019
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.10793136400084e-06,
+      "loss": 3.4265,
+      "step": 2020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.10679347824953e-06,
+      "loss": 3.4106,
+      "step": 2021
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.105654938415868e-06,
+      "loss": 3.3259,
+      "step": 2022
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.104515744681184e-06,
+      "loss": 3.4742,
+      "step": 2023
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.103375897226919e-06,
+      "loss": 3.4218,
+      "step": 2024
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.102235396234613e-06,
+      "loss": 3.4765,
+      "step": 2025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.10109424188591e-06,
+      "loss": 3.4775,
+      "step": 2026
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.099952434362564e-06,
+      "loss": 3.3859,
+      "step": 2027
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.098809973846425e-06,
+      "loss": 3.3434,
+      "step": 2028
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.097666860519451e-06,
+      "loss": 3.5325,
+      "step": 2029
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.096523094563708e-06,
+      "loss": 3.4811,
+      "step": 2030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.095378676161358e-06,
+      "loss": 3.4397,
+      "step": 2031
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.09423360549467e-06,
+      "loss": 3.3177,
+      "step": 2032
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.093087882746022e-06,
+      "loss": 3.5234,
+      "step": 2033
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.091941508097886e-06,
+      "loss": 3.4387,
+      "step": 2034
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.090794481732848e-06,
+      "loss": 3.4656,
+      "step": 2035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.089646803833589e-06,
+      "loss": 3.4867,
+      "step": 2036
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.088498474582901e-06,
+      "loss": 3.46,
+      "step": 2037
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.087349494163673e-06,
+      "loss": 3.4426,
+      "step": 2038
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.086199862758905e-06,
+      "loss": 3.4229,
+      "step": 2039
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.085049580551696e-06,
+      "loss": 3.5287,
+      "step": 2040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.083898647725247e-06,
+      "loss": 3.3391,
+      "step": 2041
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.082747064462869e-06,
+      "loss": 3.4655,
+      "step": 2042
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.081594830947968e-06,
+      "loss": 3.4418,
+      "step": 2043
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.080441947364065e-06,
+      "loss": 3.4799,
+      "step": 2044
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.07928841389477e-06,
+      "loss": 3.3853,
+      "step": 2045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.078134230723813e-06,
+      "loss": 3.5279,
+      "step": 2046
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.076979398035013e-06,
+      "loss": 3.4882,
+      "step": 2047
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.075823916012298e-06,
+      "loss": 3.4099,
+      "step": 2048
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.074667784839703e-06,
+      "loss": 3.489,
+      "step": 2049
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.073511004701362e-06,
+      "loss": 3.4193,
+      "step": 2050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.072353575781513e-06,
+      "loss": 3.3391,
+      "step": 2051
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.071195498264497e-06,
+      "loss": 3.318,
+      "step": 2052
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.070036772334761e-06,
+      "loss": 3.4339,
+      "step": 2053
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.068877398176854e-06,
+      "loss": 3.4536,
+      "step": 2054
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.067717375975424e-06,
+      "loss": 3.5258,
+      "step": 2055
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.066556705915229e-06,
+      "loss": 3.3486,
+      "step": 2056
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.065395388181126e-06,
+      "loss": 3.309,
+      "step": 2057
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.064233422958078e-06,
+      "loss": 3.449,
+      "step": 2058
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.063070810431145e-06,
+      "loss": 3.509,
+      "step": 2059
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.061907550785498e-06,
+      "loss": 3.4113,
+      "step": 2060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.060743644206407e-06,
+      "loss": 3.5162,
+      "step": 2061
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.059579090879245e-06,
+      "loss": 3.4368,
+      "step": 2062
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.058413890989488e-06,
+      "loss": 3.5431,
+      "step": 2063
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.057248044722718e-06,
+      "loss": 3.5805,
+      "step": 2064
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.056081552264614e-06,
+      "loss": 3.3498,
+      "step": 2065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.054914413800962e-06,
+      "loss": 3.5,
+      "step": 2066
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.053746629517653e-06,
+      "loss": 3.4243,
+      "step": 2067
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.052578199600675e-06,
+      "loss": 3.4241,
+      "step": 2068
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.051409124236122e-06,
+      "loss": 3.5065,
+      "step": 2069
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.050239403610195e-06,
+      "loss": 3.443,
+      "step": 2070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.049069037909188e-06,
+      "loss": 3.3862,
+      "step": 2071
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.047898027319508e-06,
+      "loss": 3.4497,
+      "step": 2072
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.046726372027655e-06,
+      "loss": 3.5605,
+      "step": 2073
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.045554072220241e-06,
+      "loss": 3.3723,
+      "step": 2074
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.044381128083974e-06,
+      "loss": 3.4558,
+      "step": 2075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.04320753980567e-06,
+      "loss": 3.4918,
+      "step": 2076
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.04203330757224e-06,
+      "loss": 3.3544,
+      "step": 2077
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.040858431570703e-06,
+      "loss": 3.3928,
+      "step": 2078
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.039682911988182e-06,
+      "loss": 3.5301,
+      "step": 2079
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.038506749011899e-06,
+      "loss": 3.5573,
+      "step": 2080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.037329942829179e-06,
+      "loss": 3.4504,
+      "step": 2081
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.036152493627451e-06,
+      "loss": 3.4107,
+      "step": 2082
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.034974401594242e-06,
+      "loss": 3.4762,
+      "step": 2083
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.033795666917191e-06,
+      "loss": 3.4235,
+      "step": 2084
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.032616289784028e-06,
+      "loss": 3.4801,
+      "step": 2085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.031436270382595e-06,
+      "loss": 3.656,
+      "step": 2086
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.030255608900826e-06,
+      "loss": 3.4759,
+      "step": 2087
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.029074305526768e-06,
+      "loss": 3.4621,
+      "step": 2088
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.027892360448562e-06,
+      "loss": 3.3937,
+      "step": 2089
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.026709773854458e-06,
+      "loss": 3.4404,
+      "step": 2090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.0255265459328e-06,
+      "loss": 3.4383,
+      "step": 2091
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.024342676872044e-06,
+      "loss": 3.4865,
+      "step": 2092
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.023158166860737e-06,
+      "loss": 3.5003,
+      "step": 2093
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.02197301608754e-06,
+      "loss": 3.4866,
+      "step": 2094
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.020787224741207e-06,
+      "loss": 3.5459,
+      "step": 2095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.019600793010596e-06,
+      "loss": 3.4638,
+      "step": 2096
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.01841372108467e-06,
+      "loss": 3.5141,
+      "step": 2097
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.017226009152491e-06,
+      "loss": 3.5091,
+      "step": 2098
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.016037657403225e-06,
+      "loss": 3.5011,
+      "step": 2099
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.014848666026138e-06,
+      "loss": 3.3614,
+      "step": 2100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.0136590352106e-06,
+      "loss": 3.4984,
+      "step": 2101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.012468765146079e-06,
+      "loss": 3.5986,
+      "step": 2102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.01127785602215e-06,
+      "loss": 3.5142,
+      "step": 2103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.010086308028487e-06,
+      "loss": 3.4665,
+      "step": 2104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.008894121354862e-06,
+      "loss": 3.3699,
+      "step": 2105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.007701296191158e-06,
+      "loss": 3.5405,
+      "step": 2106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.006507832727352e-06,
+      "loss": 3.4984,
+      "step": 2107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.005313731153525e-06,
+      "loss": 3.3598,
+      "step": 2108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.004118991659859e-06,
+      "loss": 3.4578,
+      "step": 2109
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.002923614436639e-06,
+      "loss": 3.4741,
+      "step": 2110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.001727599674252e-06,
+      "loss": 3.4723,
+      "step": 2111
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.000530947563181e-06,
+      "loss": 3.4754,
+      "step": 2112
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.99933365829402e-06,
+      "loss": 3.3583,
+      "step": 2113
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.998135732057458e-06,
+      "loss": 3.4703,
+      "step": 2114
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.996937169044284e-06,
+      "loss": 3.4625,
+      "step": 2115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.995737969445395e-06,
+      "loss": 3.3462,
+      "step": 2116
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.994538133451783e-06,
+      "loss": 3.5066,
+      "step": 2117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.993337661254544e-06,
+      "loss": 3.4593,
+      "step": 2118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.992136553044877e-06,
+      "loss": 3.4724,
+      "step": 2119
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.990934809014079e-06,
+      "loss": 3.4351,
+      "step": 2120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.989732429353548e-06,
+      "loss": 3.5532,
+      "step": 2121
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.98852941425479e-06,
+      "loss": 3.4671,
+      "step": 2122
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.987325763909404e-06,
+      "loss": 3.404,
+      "step": 2123
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.986121478509096e-06,
+      "loss": 3.4123,
+      "step": 2124
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.984916558245666e-06,
+      "loss": 3.4703,
+      "step": 2125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.983711003311025e-06,
+      "loss": 3.4627,
+      "step": 2126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.982504813897173e-06,
+      "loss": 3.5015,
+      "step": 2127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.981297990196225e-06,
+      "loss": 3.4221,
+      "step": 2128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.980090532400387e-06,
+      "loss": 3.4147,
+      "step": 2129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.978882440701968e-06,
+      "loss": 3.4923,
+      "step": 2130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.97767371529338e-06,
+      "loss": 3.4309,
+      "step": 2131
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.976464356367133e-06,
+      "loss": 3.4788,
+      "step": 2132
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.975254364115841e-06,
+      "loss": 3.4579,
+      "step": 2133
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.97404373873222e-06,
+      "loss": 3.4363,
+      "step": 2134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.972832480409079e-06,
+      "loss": 3.4008,
+      "step": 2135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.971620589339336e-06,
+      "loss": 3.3993,
+      "step": 2136
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.97040806571601e-06,
+      "loss": 3.4372,
+      "step": 2137
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.96919490973221e-06,
+      "loss": 3.4209,
+      "step": 2138
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.96798112158116e-06,
+      "loss": 3.469,
+      "step": 2139
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.966766701456177e-06,
+      "loss": 3.4933,
+      "step": 2140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.965551649550677e-06,
+      "loss": 3.4668,
+      "step": 2141
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.964335966058182e-06,
+      "loss": 3.401,
+      "step": 2142
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.963119651172312e-06,
+      "loss": 3.4199,
+      "step": 2143
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.961902705086785e-06,
+      "loss": 3.3617,
+      "step": 2144
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.960685127995423e-06,
+      "loss": 3.4677,
+      "step": 2145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.95946692009215e-06,
+      "loss": 3.354,
+      "step": 2146
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.958248081570984e-06,
+      "loss": 3.3977,
+      "step": 2147
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.957028612626051e-06,
+      "loss": 3.5005,
+      "step": 2148
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.95580851345157e-06,
+      "loss": 3.5262,
+      "step": 2149
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.95458778424187e-06,
+      "loss": 3.4438,
+      "step": 2150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.95336642519137e-06,
+      "loss": 3.4088,
+      "step": 2151
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.952144436494594e-06,
+      "loss": 3.4228,
+      "step": 2152
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.950921818346169e-06,
+      "loss": 3.3857,
+      "step": 2153
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.949698570940816e-06,
+      "loss": 3.4133,
+      "step": 2154
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.948474694473362e-06,
+      "loss": 3.349,
+      "step": 2155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.947250189138732e-06,
+      "loss": 3.3672,
+      "step": 2156
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.946025055131949e-06,
+      "loss": 3.5327,
+      "step": 2157
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.944799292648137e-06,
+      "loss": 3.4712,
+      "step": 2158
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.943572901882527e-06,
+      "loss": 3.4167,
+      "step": 2159
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.942345883030439e-06,
+      "loss": 3.5282,
+      "step": 2160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.941118236287299e-06,
+      "loss": 3.3809,
+      "step": 2161
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.939889961848635e-06,
+      "loss": 3.5288,
+      "step": 2162
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.938661059910069e-06,
+      "loss": 3.4728,
+      "step": 2163
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.937431530667329e-06,
+      "loss": 3.5736,
+      "step": 2164
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.936201374316238e-06,
+      "loss": 3.5439,
+      "step": 2165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.934970591052722e-06,
+      "loss": 3.407,
+      "step": 2166
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.933739181072807e-06,
+      "loss": 3.4349,
+      "step": 2167
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.932507144572616e-06,
+      "loss": 3.535,
+      "step": 2168
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.931274481748374e-06,
+      "loss": 3.404,
+      "step": 2169
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.930041192796404e-06,
+      "loss": 3.4289,
+      "step": 2170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.928807277913132e-06,
+      "loss": 3.4659,
+      "step": 2171
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.927572737295081e-06,
+      "loss": 3.571,
+      "step": 2172
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.926337571138872e-06,
+      "loss": 3.4944,
+      "step": 2173
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.925101779641232e-06,
+      "loss": 3.4902,
+      "step": 2174
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.92386536299898e-06,
+      "loss": 3.4739,
+      "step": 2175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.922628321409042e-06,
+      "loss": 3.5266,
+      "step": 2176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.921390655068434e-06,
+      "loss": 3.5587,
+      "step": 2177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.920152364174282e-06,
+      "loss": 3.4865,
+      "step": 2178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.918913448923804e-06,
+      "loss": 3.4365,
+      "step": 2179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.917673909514321e-06,
+      "loss": 3.425,
+      "step": 2180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.916433746143253e-06,
+      "loss": 3.502,
+      "step": 2181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.915192959008116e-06,
+      "loss": 3.4544,
+      "step": 2182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.913951548306532e-06,
+      "loss": 3.3385,
+      "step": 2183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.912709514236218e-06,
+      "loss": 3.5069,
+      "step": 2184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.911466856994986e-06,
+      "loss": 3.4779,
+      "step": 2185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.91022357678076e-06,
+      "loss": 3.4658,
+      "step": 2186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.908979673791546e-06,
+      "loss": 3.4187,
+      "step": 2187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.907735148225465e-06,
+      "loss": 3.3868,
+      "step": 2188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.90649000028073e-06,
+      "loss": 3.4185,
+      "step": 2189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.905244230155651e-06,
+      "loss": 3.4338,
+      "step": 2190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.903997838048641e-06,
+      "loss": 3.3832,
+      "step": 2191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.902750824158213e-06,
+      "loss": 3.4692,
+      "step": 2192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.901503188682972e-06,
+      "loss": 3.3892,
+      "step": 2193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.900254931821632e-06,
+      "loss": 3.3345,
+      "step": 2194
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.899006053772998e-06,
+      "loss": 3.342,
+      "step": 2195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.897756554735976e-06,
+      "loss": 3.5489,
+      "step": 2196
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.896506434909575e-06,
+      "loss": 3.4072,
+      "step": 2197
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.895255694492898e-06,
+      "loss": 3.3841,
+      "step": 2198
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.894004333685147e-06,
+      "loss": 3.3976,
+      "step": 2199
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.892752352685625e-06,
+      "loss": 3.4331,
+      "step": 2200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.891499751693734e-06,
+      "loss": 3.5237,
+      "step": 2201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.890246530908973e-06,
+      "loss": 3.3911,
+      "step": 2202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.888992690530941e-06,
+      "loss": 3.4913,
+      "step": 2203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.887738230759334e-06,
+      "loss": 3.3183,
+      "step": 2204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.88648315179395e-06,
+      "loss": 3.3325,
+      "step": 2205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.885227453834678e-06,
+      "loss": 3.3534,
+      "step": 2206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.883971137081517e-06,
+      "loss": 3.5067,
+      "step": 2207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.882714201734556e-06,
+      "loss": 3.5488,
+      "step": 2208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.881456647993987e-06,
+      "loss": 3.4751,
+      "step": 2209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.880198476060095e-06,
+      "loss": 3.3982,
+      "step": 2210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.87893968613327e-06,
+      "loss": 3.6183,
+      "step": 2211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.877680278413995e-06,
+      "loss": 3.52,
+      "step": 2212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.876420253102857e-06,
+      "loss": 3.469,
+      "step": 2213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.875159610400535e-06,
+      "loss": 3.4147,
+      "step": 2214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.873898350507812e-06,
+      "loss": 3.4889,
+      "step": 2215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.872636473625564e-06,
+      "loss": 3.5832,
+      "step": 2216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.871373979954772e-06,
+      "loss": 3.4265,
+      "step": 2217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.870110869696507e-06,
+      "loss": 3.4622,
+      "step": 2218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.868847143051946e-06,
+      "loss": 3.4558,
+      "step": 2219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.867582800222358e-06,
+      "loss": 3.4402,
+      "step": 2220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.866317841409116e-06,
+      "loss": 3.3936,
+      "step": 2221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.865052266813686e-06,
+      "loss": 3.5577,
+      "step": 2222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.863786076637633e-06,
+      "loss": 3.4656,
+      "step": 2223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.862519271082624e-06,
+      "loss": 3.4509,
+      "step": 2224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.861251850350418e-06,
+      "loss": 3.5508,
+      "step": 2225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.859983814642877e-06,
+      "loss": 3.3189,
+      "step": 2226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.858715164161956e-06,
+      "loss": 3.3984,
+      "step": 2227
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.857445899109716e-06,
+      "loss": 3.3397,
+      "step": 2228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.856176019688308e-06,
+      "loss": 3.403,
+      "step": 2229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.854905526099981e-06,
+      "loss": 3.5471,
+      "step": 2230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.85363441854709e-06,
+      "loss": 3.4449,
+      "step": 2231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.852362697232077e-06,
+      "loss": 3.522,
+      "step": 2232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.851090362357491e-06,
+      "loss": 3.4371,
+      "step": 2233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.849817414125973e-06,
+      "loss": 3.4564,
+      "step": 2234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.848543852740263e-06,
+      "loss": 3.4418,
+      "step": 2235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.847269678403202e-06,
+      "loss": 3.3616,
+      "step": 2236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.84599489131772e-06,
+      "loss": 3.4469,
+      "step": 2237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.844719491686855e-06,
+      "loss": 3.4508,
+      "step": 2238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.843443479713735e-06,
+      "loss": 3.416,
+      "step": 2239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.84216685560159e-06,
+      "loss": 3.5362,
+      "step": 2240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.840889619553747e-06,
+      "loss": 3.4294,
+      "step": 2241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.839611771773627e-06,
+      "loss": 3.3366,
+      "step": 2242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.838333312464754e-06,
+      "loss": 3.4586,
+      "step": 2243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.83705424183074e-06,
+      "loss": 3.3385,
+      "step": 2244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.835774560075305e-06,
+      "loss": 3.4219,
+      "step": 2245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.834494267402262e-06,
+      "loss": 3.4655,
+      "step": 2246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.833213364015522e-06,
+      "loss": 3.4173,
+      "step": 2247
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.831931850119089e-06,
+      "loss": 3.5223,
+      "step": 2248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.83064972591707e-06,
+      "loss": 3.5024,
+      "step": 2249
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.829366991613666e-06,
+      "loss": 3.3119,
+      "step": 2250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.828083647413177e-06,
+      "loss": 3.4226,
+      "step": 2251
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.826799693519996e-06,
+      "loss": 3.4379,
+      "step": 2252
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.82551513013862e-06,
+      "loss": 3.512,
+      "step": 2253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.824229957473638e-06,
+      "loss": 3.611,
+      "step": 2254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.822944175729738e-06,
+      "loss": 3.3667,
+      "step": 2255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.821657785111701e-06,
+      "loss": 3.4293,
+      "step": 2256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.820370785824414e-06,
+      "loss": 3.3773,
+      "step": 2257
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.819083178072851e-06,
+      "loss": 3.3557,
+      "step": 2258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.817794962062089e-06,
+      "loss": 3.465,
+      "step": 2259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.8165061379973e-06,
+      "loss": 3.5328,
+      "step": 2260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.815216706083753e-06,
+      "loss": 3.4586,
+      "step": 2261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.813926666526813e-06,
+      "loss": 3.3771,
+      "step": 2262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.812636019531942e-06,
+      "loss": 3.4078,
+      "step": 2263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.811344765304698e-06,
+      "loss": 3.2869,
+      "step": 2264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.810052904050744e-06,
+      "loss": 3.4129,
+      "step": 2265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.808760435975825e-06,
+      "loss": 3.3251,
+      "step": 2266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.807467361285793e-06,
+      "loss": 3.5113,
+      "step": 2267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.806173680186594e-06,
+      "loss": 3.4798,
+      "step": 2268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.804879392884271e-06,
+      "loss": 3.6467,
+      "step": 2269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.803584499584961e-06,
+      "loss": 3.4323,
+      "step": 2270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.802289000494901e-06,
+      "loss": 3.4037,
+      "step": 2271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.800992895820425e-06,
+      "loss": 3.3493,
+      "step": 2272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.799696185767958e-06,
+      "loss": 3.3971,
+      "step": 2273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.798398870544027e-06,
+      "loss": 3.4667,
+      "step": 2274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.797100950355254e-06,
+      "loss": 3.3702,
+      "step": 2275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.795802425408352e-06,
+      "loss": 3.5369,
+      "step": 2276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.794503295910142e-06,
+      "loss": 3.4429,
+      "step": 2277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.793203562067529e-06,
+      "loss": 3.431,
+      "step": 2278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.79190322408752e-06,
+      "loss": 3.4581,
+      "step": 2279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.790602282177221e-06,
+      "loss": 3.4584,
+      "step": 2280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.789300736543829e-06,
+      "loss": 3.4769,
+      "step": 2281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.787998587394637e-06,
+      "loss": 3.3905,
+      "step": 2282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.786695834937039e-06,
+      "loss": 3.5007,
+      "step": 2283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.785392479378522e-06,
+      "loss": 3.4955,
+      "step": 2284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.784088520926669e-06,
+      "loss": 3.3831,
+      "step": 2285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.782783959789157e-06,
+      "loss": 3.4355,
+      "step": 2286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.781478796173764e-06,
+      "loss": 3.5537,
+      "step": 2287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.780173030288359e-06,
+      "loss": 3.4897,
+      "step": 2288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.778866662340913e-06,
+      "loss": 3.5389,
+      "step": 2289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.777559692539487e-06,
+      "loss": 3.3814,
+      "step": 2290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.77625212109224e-06,
+      "loss": 3.4564,
+      "step": 2291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.774943948207427e-06,
+      "loss": 3.4279,
+      "step": 2292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.773635174093395e-06,
+      "loss": 3.3207,
+      "step": 2293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.772325798958597e-06,
+      "loss": 3.5759,
+      "step": 2294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.771015823011571e-06,
+      "loss": 3.5471,
+      "step": 2295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.769705246460955e-06,
+      "loss": 3.5481,
+      "step": 2296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.768394069515485e-06,
+      "loss": 3.2794,
+      "step": 2297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.767082292383985e-06,
+      "loss": 3.431,
+      "step": 2298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.765769915275383e-06,
+      "loss": 3.4789,
+      "step": 2299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.7644569383987e-06,
+      "loss": 3.5049,
+      "step": 2300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.763143361963049e-06,
+      "loss": 3.3791,
+      "step": 2301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.761829186177644e-06,
+      "loss": 3.3513,
+      "step": 2302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.760514411251789e-06,
+      "loss": 3.3711,
+      "step": 2303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.759199037394888e-06,
+      "loss": 3.4556,
+      "step": 2304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.757883064816436e-06,
+      "loss": 3.3021,
+      "step": 2305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.75656649372603e-06,
+      "loss": 3.4354,
+      "step": 2306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.755249324333355e-06,
+      "loss": 3.5187,
+      "step": 2307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.753931556848195e-06,
+      "loss": 3.4818,
+      "step": 2308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.75261319148043e-06,
+      "loss": 3.4418,
+      "step": 2309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.75129422844003e-06,
+      "loss": 3.4725,
+      "step": 2310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.74997466793707e-06,
+      "loss": 3.4757,
+      "step": 2311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.748654510181709e-06,
+      "loss": 3.4374,
+      "step": 2312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.74733375538421e-06,
+      "loss": 3.3371,
+      "step": 2313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.746012403754924e-06,
+      "loss": 3.5566,
+      "step": 2314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.744690455504306e-06,
+      "loss": 3.485,
+      "step": 2315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.743367910842895e-06,
+      "loss": 3.4855,
+      "step": 2316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.742044769981332e-06,
+      "loss": 3.393,
+      "step": 2317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.740721033130352e-06,
+      "loss": 3.4668,
+      "step": 2318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.739396700500786e-06,
+      "loss": 3.5706,
+      "step": 2319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.738071772303556e-06,
+      "loss": 3.3247,
+      "step": 2320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.736746248749683e-06,
+      "loss": 3.4005,
+      "step": 2321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.73542013005028e-06,
+      "loss": 3.3952,
+      "step": 2322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.734093416416556e-06,
+      "loss": 3.4395,
+      "step": 2323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.732766108059814e-06,
+      "loss": 3.4139,
+      "step": 2324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.731438205191453e-06,
+      "loss": 3.5009,
+      "step": 2325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.730109708022964e-06,
+      "loss": 3.4195,
+      "step": 2326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.728780616765938e-06,
+      "loss": 3.529,
+      "step": 2327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.727450931632056e-06,
+      "loss": 3.3832,
+      "step": 2328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.726120652833093e-06,
+      "loss": 3.5196,
+      "step": 2329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.724789780580922e-06,
+      "loss": 3.502,
+      "step": 2330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.723458315087507e-06,
+      "loss": 3.4995,
+      "step": 2331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.722126256564911e-06,
+      "loss": 3.3622,
+      "step": 2332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.720793605225289e-06,
+      "loss": 3.5096,
+      "step": 2333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.719460361280888e-06,
+      "loss": 3.2843,
+      "step": 2334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.718126524944051e-06,
+      "loss": 3.4239,
+      "step": 2335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.716792096427217e-06,
+      "loss": 3.4398,
+      "step": 2336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.71545707594292e-06,
+      "loss": 3.4285,
+      "step": 2337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.714121463703785e-06,
+      "loss": 3.4806,
+      "step": 2338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.712785259922534e-06,
+      "loss": 3.4449,
+      "step": 2339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.711448464811978e-06,
+      "loss": 3.5056,
+      "step": 2340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.710111078585033e-06,
+      "loss": 3.4223,
+      "step": 2341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.708773101454698e-06,
+      "loss": 3.357,
+      "step": 2342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.707434533634069e-06,
+      "loss": 3.4381,
+      "step": 2343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.70609537533634e-06,
+      "loss": 3.3485,
+      "step": 2344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.704755626774796e-06,
+      "loss": 3.4427,
+      "step": 2345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.703415288162817e-06,
+      "loss": 3.4288,
+      "step": 2346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.702074359713878e-06,
+      "loss": 3.4813,
+      "step": 2347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.700732841641542e-06,
+      "loss": 3.5676,
+      "step": 2348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.699390734159474e-06,
+      "loss": 3.4143,
+      "step": 2349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.69804803748143e-06,
+      "loss": 3.5134,
+      "step": 2350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.696704751821257e-06,
+      "loss": 3.4819,
+      "step": 2351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.695360877392897e-06,
+      "loss": 3.42,
+      "step": 2352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.69401641441039e-06,
+      "loss": 3.5641,
+      "step": 2353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.692671363087864e-06,
+      "loss": 3.429,
+      "step": 2354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.691325723639543e-06,
+      "loss": 3.4261,
+      "step": 2355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.689979496279747e-06,
+      "loss": 3.5592,
+      "step": 2356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.688632681222886e-06,
+      "loss": 3.4043,
+      "step": 2357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.687285278683463e-06,
+      "loss": 3.377,
+      "step": 2358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.685937288876081e-06,
+      "loss": 3.5356,
+      "step": 2359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.68458871201543e-06,
+      "loss": 3.4366,
+      "step": 2360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.683239548316296e-06,
+      "loss": 3.5785,
+      "step": 2361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.681889797993557e-06,
+      "loss": 3.4089,
+      "step": 2362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.680539461262186e-06,
+      "loss": 3.4884,
+      "step": 2363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.679188538337248e-06,
+      "loss": 3.4827,
+      "step": 2364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.677837029433905e-06,
+      "loss": 3.3725,
+      "step": 2365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.676484934767409e-06,
+      "loss": 3.4014,
+      "step": 2366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.675132254553103e-06,
+      "loss": 3.3676,
+      "step": 2367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.67377898900643e-06,
+      "loss": 3.4596,
+      "step": 2368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.672425138342918e-06,
+      "loss": 3.4816,
+      "step": 2369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.671070702778198e-06,
+      "loss": 3.3984,
+      "step": 2370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.669715682527985e-06,
+      "loss": 3.5212,
+      "step": 2371
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.668360077808093e-06,
+      "loss": 3.5482,
+      "step": 2372
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.667003888834425e-06,
+      "loss": 3.4289,
+      "step": 2373
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.665647115822981e-06,
+      "loss": 3.4573,
+      "step": 2374
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.66428975898985e-06,
+      "loss": 3.4455,
+      "step": 2375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.662931818551217e-06,
+      "loss": 3.4481,
+      "step": 2376
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.66157329472336e-06,
+      "loss": 3.4458,
+      "step": 2377
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.660214187722646e-06,
+      "loss": 3.4469,
+      "step": 2378
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.65885449776554e-06,
+      "loss": 3.5614,
+      "step": 2379
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.657494225068599e-06,
+      "loss": 3.3621,
+      "step": 2380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.656133369848468e-06,
+      "loss": 3.5242,
+      "step": 2381
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.65477193232189e-06,
+      "loss": 3.6007,
+      "step": 2382
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.6534099127057e-06,
+      "loss": 3.4726,
+      "step": 2383
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.652047311216823e-06,
+      "loss": 3.4546,
+      "step": 2384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.650684128072277e-06,
+      "loss": 3.4578,
+      "step": 2385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.649320363489178e-06,
+      "loss": 3.4562,
+      "step": 2386
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.647956017684729e-06,
+      "loss": 3.4375,
+      "step": 2387
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.646591090876225e-06,
+      "loss": 3.3679,
+      "step": 2388
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.645225583281058e-06,
+      "loss": 3.4837,
+      "step": 2389
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.64385949511671e-06,
+      "loss": 3.4528,
+      "step": 2390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.642492826600755e-06,
+      "loss": 3.4867,
+      "step": 2391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.641125577950857e-06,
+      "loss": 3.4447,
+      "step": 2392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.639757749384782e-06,
+      "loss": 3.568,
+      "step": 2393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.638389341120378e-06,
+      "loss": 3.3971,
+      "step": 2394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.63702035337559e-06,
+      "loss": 3.5141,
+      "step": 2395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.635650786368452e-06,
+      "loss": 3.4724,
+      "step": 2396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.634280640317096e-06,
+      "loss": 3.411,
+      "step": 2397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.632909915439743e-06,
+      "loss": 3.4879,
+      "step": 2398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.631538611954703e-06,
+      "loss": 3.5246,
+      "step": 2399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.630166730080385e-06,
+      "loss": 3.4488,
+      "step": 2400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.628794270035284e-06,
+      "loss": 3.5595,
+      "step": 2401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.627421232037988e-06,
+      "loss": 3.5624,
+      "step": 2402
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.626047616307183e-06,
+      "loss": 3.4642,
+      "step": 2403
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.62467342306164e-06,
+      "loss": 3.4165,
+      "step": 2404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.623298652520224e-06,
+      "loss": 3.5827,
+      "step": 2405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.621923304901894e-06,
+      "loss": 3.5386,
+      "step": 2406
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.620547380425699e-06,
+      "loss": 3.5472,
+      "step": 2407
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.61917087931078e-06,
+      "loss": 3.3493,
+      "step": 2408
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.617793801776368e-06,
+      "loss": 3.4319,
+      "step": 2409
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.61641614804179e-06,
+      "loss": 3.3899,
+      "step": 2410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.615037918326465e-06,
+      "loss": 3.5415,
+      "step": 2411
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.613659112849898e-06,
+      "loss": 3.3926,
+      "step": 2412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.612279731831689e-06,
+      "loss": 3.4446,
+      "step": 2413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.610899775491531e-06,
+      "loss": 3.4658,
+      "step": 2414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.609519244049208e-06,
+      "loss": 3.454,
+      "step": 2415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.608138137724593e-06,
+      "loss": 3.4549,
+      "step": 2416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.606756456737656e-06,
+      "loss": 3.4601,
+      "step": 2417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.605374201308452e-06,
+      "loss": 3.4228,
+      "step": 2418
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.603991371657132e-06,
+      "loss": 3.3975,
+      "step": 2419
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.602607968003935e-06,
+      "loss": 3.5152,
+      "step": 2420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.601223990569197e-06,
+      "loss": 3.3729,
+      "step": 2421
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.599839439573339e-06,
+      "loss": 3.3539,
+      "step": 2422
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.598454315236877e-06,
+      "loss": 3.5018,
+      "step": 2423
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.597068617780419e-06,
+      "loss": 3.4281,
+      "step": 2424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.59568234742466e-06,
+      "loss": 3.4169,
+      "step": 2425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.594295504390391e-06,
+      "loss": 3.5362,
+      "step": 2426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.59290808889849e-06,
+      "loss": 3.4437,
+      "step": 2427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.591520101169932e-06,
+      "loss": 3.5955,
+      "step": 2428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.590131541425778e-06,
+      "loss": 3.4207,
+      "step": 2429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.588742409887181e-06,
+      "loss": 3.4643,
+      "step": 2430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.587352706775384e-06,
+      "loss": 3.4307,
+      "step": 2431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.585962432311728e-06,
+      "loss": 3.3607,
+      "step": 2432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.584571586717635e-06,
+      "loss": 3.5327,
+      "step": 2433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.583180170214625e-06,
+      "loss": 3.3656,
+      "step": 2434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.581788183024305e-06,
+      "loss": 3.527,
+      "step": 2435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.580395625368377e-06,
+      "loss": 3.2905,
+      "step": 2436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.579002497468627e-06,
+      "loss": 3.2849,
+      "step": 2437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.577608799546943e-06,
+      "loss": 3.4992,
+      "step": 2438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.576214531825292e-06,
+      "loss": 3.2662,
+      "step": 2439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.574819694525736e-06,
+      "loss": 3.5862,
+      "step": 2440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.573424287870431e-06,
+      "loss": 3.4474,
+      "step": 2441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.572028312081622e-06,
+      "loss": 3.4823,
+      "step": 2442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.57063176738164e-06,
+      "loss": 3.5343,
+      "step": 2443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.569234653992916e-06,
+      "loss": 3.4467,
+      "step": 2444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.567836972137962e-06,
+      "loss": 3.4204,
+      "step": 2445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.566438722039384e-06,
+      "loss": 3.5273,
+      "step": 2446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.56503990391988e-06,
+      "loss": 3.491,
+      "step": 2447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.563640518002242e-06,
+      "loss": 3.4789,
+      "step": 2448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.562240564509338e-06,
+      "loss": 3.416,
+      "step": 2449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.560840043664144e-06,
+      "loss": 3.4543,
+      "step": 2450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.559438955689719e-06,
+      "loss": 3.3656,
+      "step": 2451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.558037300809209e-06,
+      "loss": 3.4704,
+      "step": 2452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.556635079245852e-06,
+      "loss": 3.4304,
+      "step": 2453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.555232291222982e-06,
+      "loss": 3.5629,
+      "step": 2454
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.553828936964017e-06,
+      "loss": 3.4497,
+      "step": 2455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.552425016692464e-06,
+      "loss": 3.3634,
+      "step": 2456
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.551020530631928e-06,
+      "loss": 3.4864,
+      "step": 2457
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.549615479006095e-06,
+      "loss": 3.3475,
+      "step": 2458
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.548209862038746e-06,
+      "loss": 3.438,
+      "step": 2459
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.546803679953754e-06,
+      "loss": 3.3435,
+      "step": 2460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.545396932975079e-06,
+      "loss": 3.4048,
+      "step": 2461
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.543989621326768e-06,
+      "loss": 3.3467,
+      "step": 2462
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.542581745232963e-06,
+      "loss": 3.4699,
+      "step": 2463
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.541173304917895e-06,
+      "loss": 3.5509,
+      "step": 2464
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.539764300605885e-06,
+      "loss": 3.4182,
+      "step": 2465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.538354732521342e-06,
+      "loss": 3.5492,
+      "step": 2466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.53694460088876e-06,
+      "loss": 3.3062,
+      "step": 2467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.535533905932739e-06,
+      "loss": 3.4737,
+      "step": 2468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.53412264787795e-06,
+      "loss": 3.52,
+      "step": 2469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.532710826949164e-06,
+      "loss": 3.3485,
+      "step": 2470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.53129844337124e-06,
+      "loss": 3.3734,
+      "step": 2471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.529885497369126e-06,
+      "loss": 3.4282,
+      "step": 2472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.528471989167858e-06,
+      "loss": 3.4999,
+      "step": 2473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.527057918992564e-06,
+      "loss": 3.4307,
+      "step": 2474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.525643287068464e-06,
+      "loss": 3.3678,
+      "step": 2475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.524228093620858e-06,
+      "loss": 3.4229,
+      "step": 2476
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.522812338875148e-06,
+      "loss": 3.4279,
+      "step": 2477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.521396023056813e-06,
+      "loss": 3.4211,
+      "step": 2478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.519979146391433e-06,
+      "loss": 3.4502,
+      "step": 2479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.518561709104667e-06,
+      "loss": 3.5386,
+      "step": 2480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.51714371142227e-06,
+      "loss": 3.4609,
+      "step": 2481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.515725153570084e-06,
+      "loss": 3.5835,
+      "step": 2482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.51430603577404e-06,
+      "loss": 3.4201,
+      "step": 2483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.512886358260162e-06,
+      "loss": 3.3636,
+      "step": 2484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.511466121254552e-06,
+      "loss": 3.3797,
+      "step": 2485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.510045324983417e-06,
+      "loss": 3.4563,
+      "step": 2486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.50862396967304e-06,
+      "loss": 3.4267,
+      "step": 2487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.507202055549799e-06,
+      "loss": 3.3941,
+      "step": 2488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.50577958284016e-06,
+      "loss": 3.4682,
+      "step": 2489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.504356551770681e-06,
+      "loss": 3.4996,
+      "step": 2490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.502932962568001e-06,
+      "loss": 3.4098,
+      "step": 2491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.501508815458856e-06,
+      "loss": 3.3744,
+      "step": 2492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.500084110670065e-06,
+      "loss": 3.4205,
+      "step": 2493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.498658848428541e-06,
+      "loss": 3.5304,
+      "step": 2494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.497233028961282e-06,
+      "loss": 3.5701,
+      "step": 2495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.495806652495377e-06,
+      "loss": 3.3647,
+      "step": 2496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.494379719258e-06,
+      "loss": 3.3886,
+      "step": 2497
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.492952229476421e-06,
+      "loss": 3.3573,
+      "step": 2498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.491524183377989e-06,
+      "loss": 3.4296,
+      "step": 2499
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.490095581190149e-06,
+      "loss": 3.4758,
+      "step": 2500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.488666423140433e-06,
+      "loss": 3.5814,
+      "step": 2501
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.487236709456457e-06,
+      "loss": 3.564,
+      "step": 2502
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.485806440365935e-06,
+      "loss": 3.3612,
+      "step": 2503
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.484375616096658e-06,
+      "loss": 3.5005,
+      "step": 2504
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.482944236876514e-06,
+      "loss": 3.5178,
+      "step": 2505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.481512302933476e-06,
+      "loss": 3.4442,
+      "step": 2506
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.480079814495608e-06,
+      "loss": 3.3485,
+      "step": 2507
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.478646771791054e-06,
+      "loss": 3.4735,
+      "step": 2508
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.477213175048059e-06,
+      "loss": 3.5447,
+      "step": 2509
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.475779024494946e-06,
+      "loss": 3.4797,
+      "step": 2510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.47434432036013e-06,
+      "loss": 3.4259,
+      "step": 2511
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.472909062872116e-06,
+      "loss": 3.4284,
+      "step": 2512
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.471473252259493e-06,
+      "loss": 3.487,
+      "step": 2513
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.470036888750941e-06,
+      "loss": 3.388,
+      "step": 2514
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.468599972575229e-06,
+      "loss": 3.5398,
+      "step": 2515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.467162503961209e-06,
+      "loss": 3.4103,
+      "step": 2516
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.465724483137827e-06,
+      "loss": 3.4648,
+      "step": 2517
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.464285910334113e-06,
+      "loss": 3.5638,
+      "step": 2518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.462846785779186e-06,
+      "loss": 3.2996,
+      "step": 2519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.461407109702253e-06,
+      "loss": 3.535,
+      "step": 2520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.45996688233261e-06,
+      "loss": 3.4287,
+      "step": 2521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.45852610389964e-06,
+      "loss": 3.4794,
+      "step": 2522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.457084774632811e-06,
+      "loss": 3.443,
+      "step": 2523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.455642894761684e-06,
+      "loss": 3.4624,
+      "step": 2524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.454200464515902e-06,
+      "loss": 3.5621,
+      "step": 2525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.4527574841252e-06,
+      "loss": 3.4563,
+      "step": 2526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.451313953819401e-06,
+      "loss": 3.5529,
+      "step": 2527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.449869873828411e-06,
+      "loss": 3.3832,
+      "step": 2528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.448425244382228e-06,
+      "loss": 3.4802,
+      "step": 2529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.446980065710933e-06,
+      "loss": 3.3447,
+      "step": 2530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.4455343380447e-06,
+      "loss": 3.3928,
+      "step": 2531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.444088061613788e-06,
+      "loss": 3.4505,
+      "step": 2532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.442641236648542e-06,
+      "loss": 3.4517,
+      "step": 2533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.441193863379396e-06,
+      "loss": 3.3683,
+      "step": 2534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.43974594203687e-06,
+      "loss": 3.4437,
+      "step": 2535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.43829747285157e-06,
+      "loss": 3.5108,
+      "step": 2536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.436848456054196e-06,
+      "loss": 3.3161,
+      "step": 2537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.435398891875529e-06,
+      "loss": 3.3967,
+      "step": 2538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.433948780546436e-06,
+      "loss": 3.4411,
+      "step": 2539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.432498122297879e-06,
+      "loss": 3.3632,
+      "step": 2540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.431046917360897e-06,
+      "loss": 3.4584,
+      "step": 2541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.429595165966624e-06,
+      "loss": 3.3261,
+      "step": 2542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.428142868346276e-06,
+      "loss": 3.5932,
+      "step": 2543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.426690024731161e-06,
+      "loss": 3.488,
+      "step": 2544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.425236635352667e-06,
+      "loss": 3.3896,
+      "step": 2545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.423782700442277e-06,
+      "loss": 3.3383,
+      "step": 2546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.422328220231555e-06,
+      "loss": 3.5168,
+      "step": 2547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.420873194952153e-06,
+      "loss": 3.4165,
+      "step": 2548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.419417624835811e-06,
+      "loss": 3.4158,
+      "step": 2549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.417961510114357e-06,
+      "loss": 3.3813,
+      "step": 2550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.416504851019702e-06,
+      "loss": 3.4685,
+      "step": 2551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.415047647783847e-06,
+      "loss": 3.4719,
+      "step": 2552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.413589900638876e-06,
+      "loss": 3.4525,
+      "step": 2553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.412131609816968e-06,
+      "loss": 3.3951,
+      "step": 2554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.410672775550376e-06,
+      "loss": 3.3293,
+      "step": 2555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.409213398071448e-06,
+      "loss": 3.4875,
+      "step": 2556
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.407753477612616e-06,
+      "loss": 3.5465,
+      "step": 2557
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.406293014406404e-06,
+      "loss": 3.5274,
+      "step": 2558
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.404832008685414e-06,
+      "loss": 3.4603,
+      "step": 2559
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.403370460682336e-06,
+      "loss": 3.4064,
+      "step": 2560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.401908370629949e-06,
+      "loss": 3.3646,
+      "step": 2561
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.400445738761123e-06,
+      "loss": 3.4253,
+      "step": 2562
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.398982565308803e-06,
+      "loss": 3.514,
+      "step": 2563
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.39751885050603e-06,
+      "loss": 3.4903,
+      "step": 2564
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.396054594585926e-06,
+      "loss": 3.3768,
+      "step": 2565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.394589797781699e-06,
+      "loss": 3.5416,
+      "step": 2566
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.393124460326647e-06,
+      "loss": 3.4383,
+      "step": 2567
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.391658582454152e-06,
+      "loss": 3.4728,
+      "step": 2568
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.390192164397682e-06,
+      "loss": 3.4679,
+      "step": 2569
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.388725206390789e-06,
+      "loss": 3.4833,
+      "step": 2570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.387257708667116e-06,
+      "loss": 3.4293,
+      "step": 2571
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.385789671460387e-06,
+      "loss": 3.4163,
+      "step": 2572
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.384321095004413e-06,
+      "loss": 3.3032,
+      "step": 2573
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.382851979533095e-06,
+      "loss": 3.4399,
+      "step": 2574
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.381382325280415e-06,
+      "loss": 3.331,
+      "step": 2575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.379912132480441e-06,
+      "loss": 3.4051,
+      "step": 2576
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.37844140136733e-06,
+      "loss": 3.4561,
+      "step": 2577
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.376970132175323e-06,
+      "loss": 3.4287,
+      "step": 2578
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.375498325138746e-06,
+      "loss": 3.3885,
+      "step": 2579
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.37402598049201e-06,
+      "loss": 3.499,
+      "step": 2580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.372553098469616e-06,
+      "loss": 3.4522,
+      "step": 2581
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.371079679306147e-06,
+      "loss": 3.4136,
+      "step": 2582
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.369605723236269e-06,
+      "loss": 3.5759,
+      "step": 2583
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.36813123049474e-06,
+      "loss": 3.5103,
+      "step": 2584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.366656201316396e-06,
+      "loss": 3.4234,
+      "step": 2585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.365180635936168e-06,
+      "loss": 3.657,
+      "step": 2586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.363704534589065e-06,
+      "loss": 3.4283,
+      "step": 2587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.36222789751018e-06,
+      "loss": 3.4887,
+      "step": 2588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.3607507249347e-06,
+      "loss": 3.5512,
+      "step": 2589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.359273017097884e-06,
+      "loss": 3.4062,
+      "step": 2590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.357794774235094e-06,
+      "loss": 3.4992,
+      "step": 2591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.356315996581758e-06,
+      "loss": 3.3951,
+      "step": 2592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.354836684373406e-06,
+      "loss": 3.3739,
+      "step": 2593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.353356837845642e-06,
+      "loss": 3.4616,
+      "step": 2594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.35187645723416e-06,
+      "loss": 3.4827,
+      "step": 2595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.350395542774737e-06,
+      "loss": 3.4156,
+      "step": 2596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.348914094703232e-06,
+      "loss": 3.4191,
+      "step": 2597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.347432113255601e-06,
+      "loss": 3.4474,
+      "step": 2598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.345949598667872e-06,
+      "loss": 3.4537,
+      "step": 2599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.344466551176163e-06,
+      "loss": 3.4142,
+      "step": 2600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.342982971016677e-06,
+      "loss": 3.597,
+      "step": 2601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.341498858425702e-06,
+      "loss": 3.5178,
+      "step": 2602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.340014213639609e-06,
+      "loss": 3.3752,
+      "step": 2603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.338529036894855e-06,
+      "loss": 3.4338,
+      "step": 2604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.337043328427983e-06,
+      "loss": 3.4952,
+      "step": 2605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.335557088475618e-06,
+      "loss": 3.4443,
+      "step": 2606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.334070317274473e-06,
+      "loss": 3.4293,
+      "step": 2607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.33258301506134e-06,
+      "loss": 3.4984,
+      "step": 2608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.331095182073103e-06,
+      "loss": 3.4907,
+      "step": 2609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.329606818546725e-06,
+      "loss": 3.4682,
+      "step": 2610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.328117924719254e-06,
+      "loss": 3.4699,
+      "step": 2611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.326628500827826e-06,
+      "loss": 3.4396,
+      "step": 2612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.32513854710966e-06,
+      "loss": 3.5078,
+      "step": 2613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.323648063802056e-06,
+      "loss": 3.4416,
+      "step": 2614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.322157051142401e-06,
+      "loss": 3.3953,
+      "step": 2615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.32066550936817e-06,
+      "loss": 3.4177,
+      "step": 2616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.319173438716911e-06,
+      "loss": 3.4196,
+      "step": 2617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.31768083942627e-06,
+      "loss": 3.4278,
+      "step": 2618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.316187711733969e-06,
+      "loss": 3.4319,
+      "step": 2619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.314694055877814e-06,
+      "loss": 3.5315,
+      "step": 2620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.313199872095701e-06,
+      "loss": 3.433,
+      "step": 2621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.311705160625605e-06,
+      "loss": 3.4053,
+      "step": 2622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.310209921705583e-06,
+      "loss": 3.5266,
+      "step": 2623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.308714155573785e-06,
+      "loss": 3.571,
+      "step": 2624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.307217862468433e-06,
+      "loss": 3.4774,
+      "step": 2625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.305721042627844e-06,
+      "loss": 3.5093,
+      "step": 2626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.304223696290412e-06,
+      "loss": 3.3181,
+      "step": 2627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.302725823694619e-06,
+      "loss": 3.3198,
+      "step": 2628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.301227425079026e-06,
+      "loss": 3.3887,
+      "step": 2629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.29972850068228e-06,
+      "loss": 3.4842,
+      "step": 2630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.298229050743115e-06,
+      "loss": 3.3138,
+      "step": 2631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.296729075500345e-06,
+      "loss": 3.5251,
+      "step": 2632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.29522857519287e-06,
+      "loss": 3.51,
+      "step": 2633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.293727550059668e-06,
+      "loss": 3.5454,
+      "step": 2634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.292226000339807e-06,
+      "loss": 3.5813,
+      "step": 2635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.290723926272439e-06,
+      "loss": 3.4567,
+      "step": 2636
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.289221328096795e-06,
+      "loss": 3.3884,
+      "step": 2637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.28771820605219e-06,
+      "loss": 3.4007,
+      "step": 2638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.286214560378025e-06,
+      "loss": 3.424,
+      "step": 2639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.284710391313783e-06,
+      "loss": 3.4297,
+      "step": 2640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.283205699099034e-06,
+      "loss": 3.4216,
+      "step": 2641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.281700483973422e-06,
+      "loss": 3.5944,
+      "step": 2642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.280194746176683e-06,
+      "loss": 3.4627,
+      "step": 2643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.278688485948634e-06,
+      "loss": 3.4134,
+      "step": 2644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.277181703529173e-06,
+      "loss": 3.4371,
+      "step": 2645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.275674399158284e-06,
+      "loss": 3.4635,
+      "step": 2646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.274166573076032e-06,
+      "loss": 3.4562,
+      "step": 2647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.27265822552257e-06,
+      "loss": 3.4086,
+      "step": 2648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.271149356738124e-06,
+      "loss": 3.375,
+      "step": 2649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.269639966963012e-06,
+      "loss": 3.5091,
+      "step": 2650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.268130056437633e-06,
+      "loss": 3.4001,
+      "step": 2651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.266619625402465e-06,
+      "loss": 3.4077,
+      "step": 2652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.265108674098074e-06,
+      "loss": 3.4446,
+      "step": 2653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.263597202765109e-06,
+      "loss": 3.4251,
+      "step": 2654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.262085211644295e-06,
+      "loss": 3.5024,
+      "step": 2655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.260572700976448e-06,
+      "loss": 3.3562,
+      "step": 2656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.259059671002462e-06,
+      "loss": 3.3814,
+      "step": 2657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.257546121963313e-06,
+      "loss": 3.3743,
+      "step": 2658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.256032054100063e-06,
+      "loss": 3.4563,
+      "step": 2659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.254517467653858e-06,
+      "loss": 3.3385,
+      "step": 2660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.253002362865919e-06,
+      "loss": 3.3979,
+      "step": 2661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.251486739977555e-06,
+      "loss": 3.3768,
+      "step": 2662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.249970599230159e-06,
+      "loss": 3.4651,
+      "step": 2663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.248453940865204e-06,
+      "loss": 3.5095,
+      "step": 2664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.246936765124245e-06,
+      "loss": 3.4707,
+      "step": 2665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.24541907224892e-06,
+      "loss": 3.5121,
+      "step": 2666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.243900862480949e-06,
+      "loss": 3.4824,
+      "step": 2667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.242382136062135e-06,
+      "loss": 3.528,
+      "step": 2668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.240862893234365e-06,
+      "loss": 3.4028,
+      "step": 2669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.239343134239603e-06,
+      "loss": 3.4038,
+      "step": 2670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.237822859319902e-06,
+      "loss": 3.2446,
+      "step": 2671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.236302068717393e-06,
+      "loss": 3.4904,
+      "step": 2672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.234780762674288e-06,
+      "loss": 3.33,
+      "step": 2673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.233258941432883e-06,
+      "loss": 3.5879,
+      "step": 2674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.231736605235559e-06,
+      "loss": 3.4459,
+      "step": 2675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.230213754324773e-06,
+      "loss": 3.3918,
+      "step": 2676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.228690388943068e-06,
+      "loss": 3.3841,
+      "step": 2677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.22716650933307e-06,
+      "loss": 3.5588,
+      "step": 2678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.225642115737479e-06,
+      "loss": 3.346,
+      "step": 2679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.224117208399088e-06,
+      "loss": 3.4111,
+      "step": 2680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.222591787560767e-06,
+      "loss": 3.4117,
+      "step": 2681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.221065853465464e-06,
+      "loss": 3.3913,
+      "step": 2682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.219539406356211e-06,
+      "loss": 3.586,
+      "step": 2683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.218012446476128e-06,
+      "loss": 3.4335,
+      "step": 2684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.216484974068409e-06,
+      "loss": 3.5102,
+      "step": 2685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.21495698937633e-06,
+      "loss": 3.4308,
+      "step": 2686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.213428492643253e-06,
+      "loss": 3.517,
+      "step": 2687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.211899484112618e-06,
+      "loss": 3.397,
+      "step": 2688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.210369964027949e-06,
+      "loss": 3.3911,
+      "step": 2689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.20883993263285e-06,
+      "loss": 3.4295,
+      "step": 2690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.207309390171006e-06,
+      "loss": 3.5081,
+      "step": 2691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.205778336886182e-06,
+      "loss": 3.3922,
+      "step": 2692
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.20424677302223e-06,
+      "loss": 3.497,
+      "step": 2693
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.202714698823078e-06,
+      "loss": 3.4586,
+      "step": 2694
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.201182114532737e-06,
+      "loss": 3.485,
+      "step": 2695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.199649020395299e-06,
+      "loss": 3.4296,
+      "step": 2696
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.198115416654937e-06,
+      "loss": 3.5247,
+      "step": 2697
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.196581303555909e-06,
+      "loss": 3.5227,
+      "step": 2698
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.195046681342545e-06,
+      "loss": 3.4155,
+      "step": 2699
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.193511550259268e-06,
+      "loss": 3.4549,
+      "step": 2700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.19197591055057e-06,
+      "loss": 3.5259,
+      "step": 2701
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.190439762461033e-06,
+      "loss": 3.4563,
+      "step": 2702
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.188903106235318e-06,
+      "loss": 3.4424,
+      "step": 2703
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.187365942118162e-06,
+      "loss": 3.5859,
+      "step": 2704
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.185828270354391e-06,
+      "loss": 3.4654,
+      "step": 2705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.184290091188904e-06,
+      "loss": 3.4036,
+      "step": 2706
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.182751404866685e-06,
+      "loss": 3.3718,
+      "step": 2707
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.1812122116328e-06,
+      "loss": 3.4777,
+      "step": 2708
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.179672511732391e-06,
+      "loss": 3.3126,
+      "step": 2709
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.178132305410687e-06,
+      "loss": 3.5282,
+      "step": 2710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.17659159291299e-06,
+      "loss": 3.5779,
+      "step": 2711
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.17505037448469e-06,
+      "loss": 3.4192,
+      "step": 2712
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.173508650371254e-06,
+      "loss": 3.3715,
+      "step": 2713
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.171966420818227e-06,
+      "loss": 3.4802,
+      "step": 2714
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.170423686071243e-06,
+      "loss": 3.4306,
+      "step": 2715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.168880446376003e-06,
+      "loss": 3.4754,
+      "step": 2716
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.167336701978305e-06,
+      "loss": 3.3487,
+      "step": 2717
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.165792453124013e-06,
+      "loss": 3.3758,
+      "step": 2718
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.164247700059078e-06,
+      "loss": 3.4747,
+      "step": 2719
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.16270244302953e-06,
+      "loss": 3.4797,
+      "step": 2720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.161156682281482e-06,
+      "loss": 3.3954,
+      "step": 2721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.159610418061124e-06,
+      "loss": 3.3293,
+      "step": 2722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.158063650614724e-06,
+      "loss": 3.3688,
+      "step": 2723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.156516380188635e-06,
+      "loss": 3.4493,
+      "step": 2724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.15496860702929e-06,
+      "loss": 3.4085,
+      "step": 2725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.1534203313832e-06,
+      "loss": 3.5487,
+      "step": 2726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.151871553496952e-06,
+      "loss": 3.4789,
+      "step": 2727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.150322273617223e-06,
+      "loss": 3.4766,
+      "step": 2728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.148772491990762e-06,
+      "loss": 3.5374,
+      "step": 2729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.1472222088644e-06,
+      "loss": 3.386,
+      "step": 2730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.145671424485047e-06,
+      "loss": 3.3206,
+      "step": 2731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.144120139099697e-06,
+      "loss": 3.4737,
+      "step": 2732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.142568352955419e-06,
+      "loss": 3.3645,
+      "step": 2733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.14101606629936e-06,
+      "loss": 3.452,
+      "step": 2734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.139463279378757e-06,
+      "loss": 3.4969,
+      "step": 2735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.137909992440915e-06,
+      "loss": 3.409,
+      "step": 2736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.136356205733224e-06,
+      "loss": 3.4145,
+      "step": 2737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.134801919503155e-06,
+      "loss": 3.5058,
+      "step": 2738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.133247133998255e-06,
+      "loss": 3.5092,
+      "step": 2739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.131691849466154e-06,
+      "loss": 3.4213,
+      "step": 2740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.130136066154558e-06,
+      "loss": 3.4646,
+      "step": 2741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.128579784311255e-06,
+      "loss": 3.4863,
+      "step": 2742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.127023004184109e-06,
+      "loss": 3.4483,
+      "step": 2743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.125465726021068e-06,
+      "loss": 3.4059,
+      "step": 2744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.12390795007016e-06,
+      "loss": 3.404,
+      "step": 2745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.122349676579484e-06,
+      "loss": 3.4835,
+      "step": 2746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.120790905797226e-06,
+      "loss": 3.3651,
+      "step": 2747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.119231637971651e-06,
+      "loss": 3.4333,
+      "step": 2748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.117671873351099e-06,
+      "loss": 3.4251,
+      "step": 2749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.11611161218399e-06,
+      "loss": 3.3733,
+      "step": 2750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.114550854718826e-06,
+      "loss": 3.5777,
+      "step": 2751
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.112989601204186e-06,
+      "loss": 3.3859,
+      "step": 2752
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.111427851888728e-06,
+      "loss": 3.3629,
+      "step": 2753
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.10986560702119e-06,
+      "loss": 3.4353,
+      "step": 2754
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.108302866850388e-06,
+      "loss": 3.4048,
+      "step": 2755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.106739631625216e-06,
+      "loss": 3.4306,
+      "step": 2756
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.10517590159465e-06,
+      "loss": 3.3874,
+      "step": 2757
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.10361167700774e-06,
+      "loss": 3.4881,
+      "step": 2758
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.102046958113621e-06,
+      "loss": 3.5105,
+      "step": 2759
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.100481745161499e-06,
+      "loss": 3.4721,
+      "step": 2760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.098916038400667e-06,
+      "loss": 3.4803,
+      "step": 2761
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.09734983808049e-06,
+      "loss": 3.5651,
+      "step": 2762
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.095783144450417e-06,
+      "loss": 3.5548,
+      "step": 2763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.09421595775997e-06,
+      "loss": 3.4055,
+      "step": 2764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.092648278258752e-06,
+      "loss": 3.4185,
+      "step": 2765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.091080106196446e-06,
+      "loss": 3.4343,
+      "step": 2766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.089511441822813e-06,
+      "loss": 3.3572,
+      "step": 2767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.08794228538769e-06,
+      "loss": 3.5846,
+      "step": 2768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.086372637140993e-06,
+      "loss": 3.4028,
+      "step": 2769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.084802497332721e-06,
+      "loss": 3.4674,
+      "step": 2770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.083231866212947e-06,
+      "loss": 3.5141,
+      "step": 2771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.081660744031818e-06,
+      "loss": 3.4032,
+      "step": 2772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.08008913103957e-06,
+      "loss": 3.4589,
+      "step": 2773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.07851702748651e-06,
+      "loss": 3.4619,
+      "step": 2774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.076944433623023e-06,
+      "loss": 3.4224,
+      "step": 2775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.075371349699573e-06,
+      "loss": 3.4716,
+      "step": 2776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.073797775966702e-06,
+      "loss": 3.4702,
+      "step": 2777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.072223712675033e-06,
+      "loss": 3.4171,
+      "step": 2778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.070649160075264e-06,
+      "loss": 3.5224,
+      "step": 2779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.06907411841817e-06,
+      "loss": 3.468,
+      "step": 2780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.067498587954606e-06,
+      "loss": 3.4286,
+      "step": 2781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.065922568935505e-06,
+      "loss": 3.4186,
+      "step": 2782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.064346061611876e-06,
+      "loss": 3.542,
+      "step": 2783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.062769066234807e-06,
+      "loss": 3.4417,
+      "step": 2784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.061191583055465e-06,
+      "loss": 3.5863,
+      "step": 2785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.05961361232509e-06,
+      "loss": 3.4317,
+      "step": 2786
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.058035154295005e-06,
+      "loss": 3.3816,
+      "step": 2787
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.056456209216609e-06,
+      "loss": 3.4157,
+      "step": 2788
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.054876777341376e-06,
+      "loss": 3.508,
+      "step": 2789
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.053296858920862e-06,
+      "loss": 3.4296,
+      "step": 2790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.051716454206696e-06,
+      "loss": 3.4507,
+      "step": 2791
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.050135563450587e-06,
+      "loss": 3.4868,
+      "step": 2792
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.048554186904322e-06,
+      "loss": 3.4746,
+      "step": 2793
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.046972324819762e-06,
+      "loss": 3.4259,
+      "step": 2794
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.045389977448851e-06,
+      "loss": 3.4407,
+      "step": 2795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.043807145043604e-06,
+      "loss": 3.2381,
+      "step": 2796
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.042223827856117e-06,
+      "loss": 3.4464,
+      "step": 2797
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.040640026138563e-06,
+      "loss": 3.3653,
+      "step": 2798
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.03905574014319e-06,
+      "loss": 3.4541,
+      "step": 2799
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.037470970122327e-06,
+      "loss": 3.4891,
+      "step": 2800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.035885716328375e-06,
+      "loss": 3.3384,
+      "step": 2801
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.034299979013819e-06,
+      "loss": 3.5027,
+      "step": 2802
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.032713758431211e-06,
+      "loss": 3.4788,
+      "step": 2803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.031127054833192e-06,
+      "loss": 3.4064,
+      "step": 2804
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.029539868472468e-06,
+      "loss": 3.5315,
+      "step": 2805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.027952199601832e-06,
+      "loss": 3.4048,
+      "step": 2806
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.026364048474145e-06,
+      "loss": 3.4923,
+      "step": 2807
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.024775415342355e-06,
+      "loss": 3.4786,
+      "step": 2808
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.023186300459476e-06,
+      "loss": 3.4906,
+      "step": 2809
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.021596704078604e-06,
+      "loss": 3.3743,
+      "step": 2810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.020006626452916e-06,
+      "loss": 3.4818,
+      "step": 2811
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.01841606783566e-06,
+      "loss": 3.4392,
+      "step": 2812
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.016825028480156e-06,
+      "loss": 3.4673,
+      "step": 2813
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.015233508639811e-06,
+      "loss": 3.4596,
+      "step": 2814
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.013641508568103e-06,
+      "loss": 3.3485,
+      "step": 2815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.01204902851859e-06,
+      "loss": 3.36,
+      "step": 2816
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.010456068744898e-06,
+      "loss": 3.4016,
+      "step": 2817
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.008862629500739e-06,
+      "loss": 3.4924,
+      "step": 2818
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.007268711039898e-06,
+      "loss": 3.3574,
+      "step": 2819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.005674313616231e-06,
+      "loss": 3.419,
+      "step": 2820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.00407943748368e-06,
+      "loss": 3.4926,
+      "step": 2821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.002484082896258e-06,
+      "loss": 3.4551,
+      "step": 2822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.000888250108054e-06,
+      "loss": 3.4482,
+      "step": 2823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.999291939373232e-06,
+      "loss": 3.4974,
+      "step": 2824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.997695150946034e-06,
+      "loss": 3.3503,
+      "step": 2825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.996097885080781e-06,
+      "loss": 3.3844,
+      "step": 2826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.994500142031863e-06,
+      "loss": 3.425,
+      "step": 2827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.992901922053751e-06,
+      "loss": 3.4216,
+      "step": 2828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.991303225400994e-06,
+      "loss": 3.5891,
+      "step": 2829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.989704052328212e-06,
+      "loss": 3.3841,
+      "step": 2830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.988104403090098e-06,
+      "loss": 3.4373,
+      "step": 2831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.986504277941432e-06,
+      "loss": 3.3922,
+      "step": 2832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.984903677137061e-06,
+      "loss": 3.3711,
+      "step": 2833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.983302600931911e-06,
+      "loss": 3.4159,
+      "step": 2834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.981701049580982e-06,
+      "loss": 3.4361,
+      "step": 2835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.98009902333935e-06,
+      "loss": 3.4351,
+      "step": 2836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.978496522462167e-06,
+      "loss": 3.4777,
+      "step": 2837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.976893547204662e-06,
+      "loss": 3.3508,
+      "step": 2838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.975290097822138e-06,
+      "loss": 3.4474,
+      "step": 2839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.973686174569971e-06,
+      "loss": 3.4085,
+      "step": 2840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.972081777703622e-06,
+      "loss": 3.4319,
+      "step": 2841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.970476907478615e-06,
+      "loss": 3.4477,
+      "step": 2842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.968871564150555e-06,
+      "loss": 3.4445,
+      "step": 2843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.967265747975124e-06,
+      "loss": 3.5272,
+      "step": 2844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.965659459208077e-06,
+      "loss": 3.424,
+      "step": 2845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.964052698105247e-06,
+      "loss": 3.3606,
+      "step": 2846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.962445464922539e-06,
+      "loss": 3.4509,
+      "step": 2847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.960837759915934e-06,
+      "loss": 3.4824,
+      "step": 2848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.959229583341488e-06,
+      "loss": 3.4103,
+      "step": 2849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.957620935455334e-06,
+      "loss": 3.4386,
+      "step": 2850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.956011816513679e-06,
+      "loss": 3.4797,
+      "step": 2851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.954402226772804e-06,
+      "loss": 3.4555,
+      "step": 2852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.952792166489065e-06,
+      "loss": 3.4024,
+      "step": 2853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.951181635918895e-06,
+      "loss": 3.4858,
+      "step": 2854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.9495706353188e-06,
+      "loss": 3.4334,
+      "step": 2855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.947959164945362e-06,
+      "loss": 3.345,
+      "step": 2856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.946347225055237e-06,
+      "loss": 3.3978,
+      "step": 2857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.944734815905156e-06,
+      "loss": 3.4571,
+      "step": 2858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.943121937751924e-06,
+      "loss": 3.4777,
+      "step": 2859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.941508590852422e-06,
+      "loss": 3.4669,
+      "step": 2860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.939894775463605e-06,
+      "loss": 3.4676,
+      "step": 2861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.938280491842505e-06,
+      "loss": 3.4051,
+      "step": 2862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.936665740246224e-06,
+      "loss": 3.5384,
+      "step": 2863
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.93505052093194e-06,
+      "loss": 3.3978,
+      "step": 2864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.933434834156909e-06,
+      "loss": 3.4146,
+      "step": 2865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.931818680178456e-06,
+      "loss": 3.5427,
+      "step": 2866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.930202059253986e-06,
+      "loss": 3.3726,
+      "step": 2867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.928584971640974e-06,
+      "loss": 3.4778,
+      "step": 2868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.92696741759697e-06,
+      "loss": 3.3696,
+      "step": 2869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.925349397379604e-06,
+      "loss": 3.5017,
+      "step": 2870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.92373091124657e-06,
+      "loss": 3.3535,
+      "step": 2871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.922111959455644e-06,
+      "loss": 3.4233,
+      "step": 2872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.920492542264673e-06,
+      "loss": 3.409,
+      "step": 2873
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.91887265993158e-06,
+      "loss": 3.3428,
+      "step": 2874
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.91725231271436e-06,
+      "loss": 3.4756,
+      "step": 2875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.915631500871084e-06,
+      "loss": 3.4244,
+      "step": 2876
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.914010224659895e-06,
+      "loss": 3.2695,
+      "step": 2877
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.912388484339012e-06,
+      "loss": 3.3729,
+      "step": 2878
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.910766280166726e-06,
+      "loss": 3.5258,
+      "step": 2879
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.909143612401403e-06,
+      "loss": 3.4634,
+      "step": 2880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.907520481301485e-06,
+      "loss": 3.5866,
+      "step": 2881
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.905896887125482e-06,
+      "loss": 3.5099,
+      "step": 2882
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.904272830131984e-06,
+      "loss": 3.3575,
+      "step": 2883
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.90264831057965e-06,
+      "loss": 3.5128,
+      "step": 2884
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.901023328727216e-06,
+      "loss": 3.5267,
+      "step": 2885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.899397884833488e-06,
+      "loss": 3.3976,
+      "step": 2886
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.897771979157351e-06,
+      "loss": 3.4802,
+      "step": 2887
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.896145611957759e-06,
+      "loss": 3.5347,
+      "step": 2888
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.894518783493741e-06,
+      "loss": 3.5308,
+      "step": 2889
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.892891494024398e-06,
+      "loss": 3.46,
+      "step": 2890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.891263743808905e-06,
+      "loss": 3.4461,
+      "step": 2891
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.889635533106515e-06,
+      "loss": 3.3931,
+      "step": 2892
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.888006862176547e-06,
+      "loss": 3.446,
+      "step": 2893
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.8863777312784e-06,
+      "loss": 3.45,
+      "step": 2894
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.884748140671541e-06,
+      "loss": 3.456,
+      "step": 2895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.883118090615512e-06,
+      "loss": 3.3745,
+      "step": 2896
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.881487581369927e-06,
+      "loss": 3.4324,
+      "step": 2897
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.87985661319448e-06,
+      "loss": 3.5203,
+      "step": 2898
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.878225186348926e-06,
+      "loss": 3.3564,
+      "step": 2899
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.876593301093104e-06,
+      "loss": 3.4862,
+      "step": 2900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.87496095768692e-06,
+      "loss": 3.4389,
+      "step": 2901
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.873328156390356e-06,
+      "loss": 3.5223,
+      "step": 2902
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.871694897463464e-06,
+      "loss": 3.5326,
+      "step": 2903
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.870061181166372e-06,
+      "loss": 3.4416,
+      "step": 2904
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.868427007759279e-06,
+      "loss": 3.4063,
+      "step": 2905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.866792377502456e-06,
+      "loss": 3.3358,
+      "step": 2906
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.865157290656252e-06,
+      "loss": 3.437,
+      "step": 2907
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.863521747481078e-06,
+      "loss": 3.4136,
+      "step": 2908
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.861885748237427e-06,
+      "loss": 3.3773,
+      "step": 2909
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.860249293185864e-06,
+      "loss": 3.4186,
+      "step": 2910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.858612382587026e-06,
+      "loss": 3.364,
+      "step": 2911
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.856975016701616e-06,
+      "loss": 3.3788,
+      "step": 2912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.855337195790416e-06,
+      "loss": 3.503,
+      "step": 2913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.853698920114281e-06,
+      "loss": 3.355,
+      "step": 2914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.852060189934136e-06,
+      "loss": 3.4016,
+      "step": 2915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.850421005510977e-06,
+      "loss": 3.298,
+      "step": 2916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.848781367105875e-06,
+      "loss": 3.4759,
+      "step": 2917
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.847141274979977e-06,
+      "loss": 3.4056,
+      "step": 2918
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.845500729394492e-06,
+      "loss": 3.4708,
+      "step": 2919
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.843859730610709e-06,
+      "loss": 3.305,
+      "step": 2920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.842218278889987e-06,
+      "loss": 3.503,
+      "step": 2921
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.840576374493761e-06,
+      "loss": 3.4886,
+      "step": 2922
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.838934017683526e-06,
+      "loss": 3.5512,
+      "step": 2923
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.837291208720867e-06,
+      "loss": 3.4281,
+      "step": 2924
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.835647947867424e-06,
+      "loss": 3.3735,
+      "step": 2925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.834004235384922e-06,
+      "loss": 3.5173,
+      "step": 2926
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.83236007153515e-06,
+      "loss": 3.4571,
+      "step": 2927
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.83071545657997e-06,
+      "loss": 3.5461,
+      "step": 2928
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.829070390781319e-06,
+      "loss": 3.3903,
+      "step": 2929
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.827424874401203e-06,
+      "loss": 3.5465,
+      "step": 2930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.825778907701704e-06,
+      "loss": 3.3675,
+      "step": 2931
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.824132490944968e-06,
+      "loss": 3.4482,
+      "step": 2932
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.82248562439322e-06,
+      "loss": 3.4477,
+      "step": 2933
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.820838308308751e-06,
+      "loss": 3.4028,
+      "step": 2934
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.819190542953929e-06,
+      "loss": 3.5439,
+      "step": 2935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.81754232859119e-06,
+      "loss": 3.4483,
+      "step": 2936
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.815893665483043e-06,
+      "loss": 3.5055,
+      "step": 2937
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.814244553892065e-06,
+      "loss": 3.5109,
+      "step": 2938
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.812594994080912e-06,
+      "loss": 3.4988,
+      "step": 2939
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.810944986312303e-06,
+      "loss": 3.316,
+      "step": 2940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.809294530849035e-06,
+      "loss": 3.4968,
+      "step": 2941
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.80764362795397e-06,
+      "loss": 3.4196,
+      "step": 2942
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.805992277890047e-06,
+      "loss": 3.4988,
+      "step": 2943
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.804340480920274e-06,
+      "loss": 3.6003,
+      "step": 2944
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.80268823730773e-06,
+      "loss": 3.4254,
+      "step": 2945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.801035547315563e-06,
+      "loss": 3.3957,
+      "step": 2946
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.799382411206995e-06,
+      "loss": 3.3881,
+      "step": 2947
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.797728829245321e-06,
+      "loss": 3.4043,
+      "step": 2948
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.796074801693903e-06,
+      "loss": 3.5592,
+      "step": 2949
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.794420328816172e-06,
+      "loss": 3.4698,
+      "step": 2950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.792765410875637e-06,
+      "loss": 3.461,
+      "step": 2951
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.791110048135874e-06,
+      "loss": 3.3538,
+      "step": 2952
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.789454240860528e-06,
+      "loss": 3.3364,
+      "step": 2953
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.787797989313317e-06,
+      "loss": 3.5468,
+      "step": 2954
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.786141293758031e-06,
+      "loss": 3.5833,
+      "step": 2955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.784484154458529e-06,
+      "loss": 3.418,
+      "step": 2956
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.78282657167874e-06,
+      "loss": 3.5115,
+      "step": 2957
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.781168545682665e-06,
+      "loss": 3.4312,
+      "step": 2958
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.779510076734376e-06,
+      "loss": 3.4909,
+      "step": 2959
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.777851165098012e-06,
+      "loss": 3.4496,
+      "step": 2960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.776191811037788e-06,
+      "loss": 3.4223,
+      "step": 2961
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.774532014817983e-06,
+      "loss": 3.5107,
+      "step": 2962
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.772871776702955e-06,
+      "loss": 3.4724,
+      "step": 2963
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.771211096957125e-06,
+      "loss": 3.5088,
+      "step": 2964
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.769549975844986e-06,
+      "loss": 3.4029,
+      "step": 2965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.767888413631102e-06,
+      "loss": 3.388,
+      "step": 2966
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.766226410580108e-06,
+      "loss": 3.3653,
+      "step": 2967
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.764563966956711e-06,
+      "loss": 3.4147,
+      "step": 2968
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.76290108302568e-06,
+      "loss": 3.4619,
+      "step": 2969
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.761237759051866e-06,
+      "loss": 3.4754,
+      "step": 2970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.75957399530018e-06,
+      "loss": 3.5151,
+      "step": 2971
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.757909792035608e-06,
+      "loss": 3.3575,
+      "step": 2972
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.756245149523206e-06,
+      "loss": 3.4735,
+      "step": 2973
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.754580068028096e-06,
+      "loss": 3.5702,
+      "step": 2974
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.752914547815477e-06,
+      "loss": 3.4538,
+      "step": 2975
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.751248589150611e-06,
+      "loss": 3.4593,
+      "step": 2976
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.749582192298832e-06,
+      "loss": 3.4684,
+      "step": 2977
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.747915357525546e-06,
+      "loss": 3.4658,
+      "step": 2978
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.746248085096227e-06,
+      "loss": 3.3214,
+      "step": 2979
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.744580375276416e-06,
+      "loss": 3.3618,
+      "step": 2980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.742912228331732e-06,
+      "loss": 3.5605,
+      "step": 2981
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.741243644527853e-06,
+      "loss": 3.4989,
+      "step": 2982
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.739574624130532e-06,
+      "loss": 3.4725,
+      "step": 2983
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.737905167405596e-06,
+      "loss": 3.4704,
+      "step": 2984
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.73623527461893e-06,
+      "loss": 3.4164,
+      "step": 2985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.7345649460365e-06,
+      "loss": 3.379,
+      "step": 2986
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.732894181924336e-06,
+      "loss": 3.4256,
+      "step": 2987
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.731222982548534e-06,
+      "loss": 3.5063,
+      "step": 2988
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.729551348175269e-06,
+      "loss": 3.495,
+      "step": 2989
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.727879279070773e-06,
+      "loss": 3.5101,
+      "step": 2990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.72620677550136e-06,
+      "loss": 3.4952,
+      "step": 2991
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.724533837733404e-06,
+      "loss": 3.4884,
+      "step": 2992
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.722860466033349e-06,
+      "loss": 3.4564,
+      "step": 2993
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.721186660667715e-06,
+      "loss": 3.4769,
+      "step": 2994
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.719512421903083e-06,
+      "loss": 3.5223,
+      "step": 2995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.717837750006106e-06,
+      "loss": 3.3743,
+      "step": 2996
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.716162645243509e-06,
+      "loss": 3.4167,
+      "step": 2997
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.71448710788208e-06,
+      "loss": 3.3041,
+      "step": 2998
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.712811138188682e-06,
+      "loss": 3.5045,
+      "step": 2999
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.711134736430244e-06,
+      "loss": 3.5088,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.709457902873759e-06,
+      "loss": 3.37,
+      "step": 3001
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.7077806377863e-06,
+      "loss": 3.433,
+      "step": 3002
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.706102941434999e-06,
+      "loss": 3.4298,
+      "step": 3003
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.704424814087056e-06,
+      "loss": 3.4904,
+      "step": 3004
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.702746256009753e-06,
+      "loss": 3.3342,
+      "step": 3005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.701067267470424e-06,
+      "loss": 3.4331,
+      "step": 3006
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.699387848736482e-06,
+      "loss": 3.4846,
+      "step": 3007
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.697708000075404e-06,
+      "loss": 3.4926,
+      "step": 3008
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.696027721754737e-06,
+      "loss": 3.5892,
+      "step": 3009
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.694347014042096e-06,
+      "loss": 3.4137,
+      "step": 3010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.692665877205165e-06,
+      "loss": 3.478,
+      "step": 3011
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.690984311511695e-06,
+      "loss": 3.3954,
+      "step": 3012
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.68930231722951e-06,
+      "loss": 3.3796,
+      "step": 3013
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.687619894626493e-06,
+      "loss": 3.3916,
+      "step": 3014
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.685937043970603e-06,
+      "loss": 3.4577,
+      "step": 3015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.684253765529865e-06,
+      "loss": 3.4891,
+      "step": 3016
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.682570059572374e-06,
+      "loss": 3.4584,
+      "step": 3017
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.680885926366289e-06,
+      "loss": 3.4576,
+      "step": 3018
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.679201366179838e-06,
+      "loss": 3.4757,
+      "step": 3019
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.67751637928132e-06,
+      "loss": 3.4965,
+      "step": 3020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.675830965939101e-06,
+      "loss": 3.4357,
+      "step": 3021
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.674145126421614e-06,
+      "loss": 3.4405,
+      "step": 3022
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.672458860997358e-06,
+      "loss": 3.467,
+      "step": 3023
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.670772169934902e-06,
+      "loss": 3.4349,
+      "step": 3024
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.669085053502883e-06,
+      "loss": 3.5187,
+      "step": 3025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.667397511970006e-06,
+      "loss": 3.4229,
+      "step": 3026
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.66570954560504e-06,
+      "loss": 3.4896,
+      "step": 3027
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.664021154676828e-06,
+      "loss": 3.5614,
+      "step": 3028
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.662332339454278e-06,
+      "loss": 3.412,
+      "step": 3029
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.660643100206364e-06,
+      "loss": 3.4597,
+      "step": 3030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.658953437202126e-06,
+      "loss": 3.5008,
+      "step": 3031
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.657263350710676e-06,
+      "loss": 3.3571,
+      "step": 3032
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.65557284100119e-06,
+      "loss": 3.4278,
+      "step": 3033
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.653881908342916e-06,
+      "loss": 3.5024,
+      "step": 3034
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.652190553005162e-06,
+      "loss": 3.3993,
+      "step": 3035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.650498775257308e-06,
+      "loss": 3.3314,
+      "step": 3036
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.648806575368805e-06,
+      "loss": 3.4607,
+      "step": 3037
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.647113953609163e-06,
+      "loss": 3.4522,
+      "step": 3038
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.645420910247962e-06,
+      "loss": 3.41,
+      "step": 3039
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.643727445554855e-06,
+      "loss": 3.4072,
+      "step": 3040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.642033559799553e-06,
+      "loss": 3.4268,
+      "step": 3041
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.64033925325184e-06,
+      "loss": 3.5095,
+      "step": 3042
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.638644526181567e-06,
+      "loss": 3.3328,
+      "step": 3043
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.636949378858647e-06,
+      "loss": 3.4672,
+      "step": 3044
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.635253811553067e-06,
+      "loss": 3.3676,
+      "step": 3045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.633557824534875e-06,
+      "loss": 3.4235,
+      "step": 3046
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.63186141807419e-06,
+      "loss": 3.4055,
+      "step": 3047
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.630164592441191e-06,
+      "loss": 3.4635,
+      "step": 3048
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6284673479061365e-06,
+      "loss": 3.4538,
+      "step": 3049
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.626769684739337e-06,
+      "loss": 3.4152,
+      "step": 3050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.62507160321118e-06,
+      "loss": 3.4201,
+      "step": 3051
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.623373103592117e-06,
+      "loss": 3.453,
+      "step": 3052
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.621674186152661e-06,
+      "loss": 3.4249,
+      "step": 3053
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6199748511634005e-06,
+      "loss": 3.3873,
+      "step": 3054
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.618275098894983e-06,
+      "loss": 3.3687,
+      "step": 3055
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.616574929618126e-06,
+      "loss": 3.4791,
+      "step": 3056
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.6148743436036155e-06,
+      "loss": 3.2863,
+      "step": 3057
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.6131733411222954e-06,
+      "loss": 3.5169,
+      "step": 3058
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.6114719224450865e-06,
+      "loss": 3.4447,
+      "step": 3059
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.609770087842969e-06,
+      "loss": 3.4456,
+      "step": 3060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.608067837586991e-06,
+      "loss": 3.4461,
+      "step": 3061
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.606365171948267e-06,
+      "loss": 3.3993,
+      "step": 3062
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.60466209119798e-06,
+      "loss": 3.4468,
+      "step": 3063
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.602958595607375e-06,
+      "loss": 3.4161,
+      "step": 3064
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.601254685447764e-06,
+      "loss": 3.5436,
+      "step": 3065
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.599550360990531e-06,
+      "loss": 3.3867,
+      "step": 3066
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.597845622507114e-06,
+      "loss": 3.4431,
+      "step": 3067
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.596140470269029e-06,
+      "loss": 3.4521,
+      "step": 3068
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.594434904547852e-06,
+      "loss": 3.2625,
+      "step": 3069
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.592728925615223e-06,
+      "loss": 3.4764,
+      "step": 3070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.591022533742852e-06,
+      "loss": 3.4172,
+      "step": 3071
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5893157292025155e-06,
+      "loss": 3.2995,
+      "step": 3072
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.587608512266052e-06,
+      "loss": 3.3442,
+      "step": 3073
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5859008832053636e-06,
+      "loss": 3.4991,
+      "step": 3074
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.584192842292428e-06,
+      "loss": 3.3969,
+      "step": 3075
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.582484389799279e-06,
+      "loss": 3.5147,
+      "step": 3076
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.580775525998016e-06,
+      "loss": 3.5396,
+      "step": 3077
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.579066251160814e-06,
+      "loss": 3.4945,
+      "step": 3078
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5773565655599e-06,
+      "loss": 3.5208,
+      "step": 3079
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.575646469467576e-06,
+      "loss": 3.4733,
+      "step": 3080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.573935963156206e-06,
+      "loss": 3.3762,
+      "step": 3081
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.572225046898218e-06,
+      "loss": 3.4275,
+      "step": 3082
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.570513720966108e-06,
+      "loss": 3.4399,
+      "step": 3083
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.568801985632439e-06,
+      "loss": 3.4529,
+      "step": 3084
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.56708984116983e-06,
+      "loss": 3.46,
+      "step": 3085
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.565377287850977e-06,
+      "loss": 3.3425,
+      "step": 3086
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.563664325948633e-06,
+      "loss": 3.3486,
+      "step": 3087
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.56195095573562e-06,
+      "loss": 3.5199,
+      "step": 3088
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.56023717748482e-06,
+      "loss": 3.4879,
+      "step": 3089
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.558522991469188e-06,
+      "loss": 3.4762,
+      "step": 3090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.556808397961739e-06,
+      "loss": 3.4256,
+      "step": 3091
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.555093397235553e-06,
+      "loss": 3.4621,
+      "step": 3092
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.553377989563775e-06,
+      "loss": 3.5817,
+      "step": 3093
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.551662175219614e-06,
+      "loss": 3.4529,
+      "step": 3094
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.549945954476347e-06,
+      "loss": 3.3987,
+      "step": 3095
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5482293276073135e-06,
+      "loss": 3.493,
+      "step": 3096
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.546512294885917e-06,
+      "loss": 3.4435,
+      "step": 3097
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.544794856585627e-06,
+      "loss": 3.4466,
+      "step": 3098
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.543077012979978e-06,
+      "loss": 3.2844,
+      "step": 3099
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.541358764342566e-06,
+      "loss": 3.4417,
+      "step": 3100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.539640110947054e-06,
+      "loss": 3.3706,
+      "step": 3101
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.537921053067171e-06,
+      "loss": 3.6437,
+      "step": 3102
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.536201590976709e-06,
+      "loss": 3.5134,
+      "step": 3103
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5344817249495195e-06,
+      "loss": 3.523,
+      "step": 3104
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.532761455259528e-06,
+      "loss": 3.4107,
+      "step": 3105
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5310407821807165e-06,
+      "loss": 3.4342,
+      "step": 3106
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.529319705987135e-06,
+      "loss": 3.395,
+      "step": 3107
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.527598226952895e-06,
+      "loss": 3.3899,
+      "step": 3108
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5258763453521746e-06,
+      "loss": 3.4713,
+      "step": 3109
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.524154061459215e-06,
+      "loss": 3.4658,
+      "step": 3110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.522431375548321e-06,
+      "loss": 3.4099,
+      "step": 3111
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.520708287893863e-06,
+      "loss": 3.4485,
+      "step": 3112
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.518984798770273e-06,
+      "loss": 3.394,
+      "step": 3113
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.51726090845205e-06,
+      "loss": 3.423,
+      "step": 3114
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.515536617213754e-06,
+      "loss": 3.4258,
+      "step": 3115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.51381192533001e-06,
+      "loss": 3.4255,
+      "step": 3116
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.512086833075509e-06,
+      "loss": 3.5025,
+      "step": 3117
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.510361340725001e-06,
+      "loss": 3.4914,
+      "step": 3118
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.508635448553303e-06,
+      "loss": 3.4165,
+      "step": 3119
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.506909156835297e-06,
+      "loss": 3.4259,
+      "step": 3120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.505182465845924e-06,
+      "loss": 3.4824,
+      "step": 3121
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.503455375860192e-06,
+      "loss": 3.5449,
+      "step": 3122
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.501727887153174e-06,
+      "loss": 3.4804,
+      "step": 3123
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 3.4196,
+      "step": 3124
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.498271714675873e-06,
+      "loss": 3.4116,
+      "step": 3125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.49654303145605e-06,
+      "loss": 3.4123,
+      "step": 3126
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.494813950615858e-06,
+      "loss": 3.4745,
+      "step": 3127
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.493084472430683e-06,
+      "loss": 3.3662,
+      "step": 3128
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.491354597175978e-06,
+      "loss": 3.4584,
+      "step": 3129
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.489624325127257e-06,
+      "loss": 3.4141,
+      "step": 3130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.487893656560097e-06,
+      "loss": 3.4662,
+      "step": 3131
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.486162591750139e-06,
+      "loss": 3.5443,
+      "step": 3132
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.4844311309730875e-06,
+      "loss": 3.4376,
+      "step": 3133
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.482699274504708e-06,
+      "loss": 3.4587,
+      "step": 3134
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.480967022620833e-06,
+      "loss": 3.5143,
+      "step": 3135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.479234375597353e-06,
+      "loss": 3.5,
+      "step": 3136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.477501333710224e-06,
+      "loss": 3.3445,
+      "step": 3137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.475767897235467e-06,
+      "loss": 3.3535,
+      "step": 3138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.474034066449161e-06,
+      "loss": 3.4215,
+      "step": 3139
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.472299841627452e-06,
+      "loss": 3.4481,
+      "step": 3140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.470565223046546e-06,
+      "loss": 3.4649,
+      "step": 3141
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4688302109827146e-06,
+      "loss": 3.3338,
+      "step": 3142
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.467094805712289e-06,
+      "loss": 3.3286,
+      "step": 3143
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.465359007511667e-06,
+      "loss": 3.506,
+      "step": 3144
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.463622816657302e-06,
+      "loss": 3.4146,
+      "step": 3145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.461886233425717e-06,
+      "loss": 3.4142,
+      "step": 3146
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.460149258093495e-06,
+      "loss": 3.5135,
+      "step": 3147
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.45841189093728e-06,
+      "loss": 3.4158,
+      "step": 3148
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.45667413223378e-06,
+      "loss": 3.3753,
+      "step": 3149
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.454935982259765e-06,
+      "loss": 3.4045,
+      "step": 3150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.453197441292068e-06,
+      "loss": 3.5269,
+      "step": 3151
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.451458509607583e-06,
+      "loss": 3.4143,
+      "step": 3152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.449719187483267e-06,
+      "loss": 3.485,
+      "step": 3153
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.44797947519614e-06,
+      "loss": 3.3796,
+      "step": 3154
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.446239373023281e-06,
+      "loss": 3.3344,
+      "step": 3155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.444498881241835e-06,
+      "loss": 3.4181,
+      "step": 3156
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.442758000129007e-06,
+      "loss": 3.5272,
+      "step": 3157
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.441016729962064e-06,
+      "loss": 3.3511,
+      "step": 3158
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.439275071018336e-06,
+      "loss": 3.4379,
+      "step": 3159
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.437533023575212e-06,
+      "loss": 3.5475,
+      "step": 3160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.435790587910148e-06,
+      "loss": 3.2941,
+      "step": 3161
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.434047764300658e-06,
+      "loss": 3.476,
+      "step": 3162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.432304553024316e-06,
+      "loss": 3.4986,
+      "step": 3163
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.430560954358764e-06,
+      "loss": 3.3858,
+      "step": 3164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.428816968581701e-06,
+      "loss": 3.406,
+      "step": 3165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.427072595970888e-06,
+      "loss": 3.2812,
+      "step": 3166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.425327836804149e-06,
+      "loss": 3.3584,
+      "step": 3167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.423582691359371e-06,
+      "loss": 3.4985,
+      "step": 3168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.421837159914497e-06,
+      "loss": 3.476,
+      "step": 3169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.420091242747536e-06,
+      "loss": 3.4075,
+      "step": 3170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.418344940136559e-06,
+      "loss": 3.3633,
+      "step": 3171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4165982523596945e-06,
+      "loss": 3.444,
+      "step": 3172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.414851179695137e-06,
+      "loss": 3.4616,
+      "step": 3173
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.413103722421139e-06,
+      "loss": 3.3787,
+      "step": 3174
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.411355880816014e-06,
+      "loss": 3.5306,
+      "step": 3175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.409607655158139e-06,
+      "loss": 3.4461,
+      "step": 3176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.407859045725951e-06,
+      "loss": 3.4601,
+      "step": 3177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.406110052797947e-06,
+      "loss": 3.4001,
+      "step": 3178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.404360676652688e-06,
+      "loss": 3.424,
+      "step": 3179
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.402610917568794e-06,
+      "loss": 3.3162,
+      "step": 3180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.400860775824946e-06,
+      "loss": 3.3596,
+      "step": 3181
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.399110251699887e-06,
+      "loss": 3.4418,
+      "step": 3182
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.397359345472419e-06,
+      "loss": 3.5008,
+      "step": 3183
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.395608057421406e-06,
+      "loss": 3.4041,
+      "step": 3184
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.393856387825773e-06,
+      "loss": 3.4288,
+      "step": 3185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.392104336964507e-06,
+      "loss": 3.4472,
+      "step": 3186
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.390351905116652e-06,
+      "loss": 3.4783,
+      "step": 3187
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.388599092561315e-06,
+      "loss": 3.4516,
+      "step": 3188
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.386845899577666e-06,
+      "loss": 3.3317,
+      "step": 3189
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.385092326444931e-06,
+      "loss": 3.4069,
+      "step": 3190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.383338373442401e-06,
+      "loss": 3.4475,
+      "step": 3191
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3815840408494244e-06,
+      "loss": 3.391,
+      "step": 3192
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.37982932894541e-06,
+      "loss": 3.4064,
+      "step": 3193
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.378074238009827e-06,
+      "loss": 3.4822,
+      "step": 3194
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.376318768322208e-06,
+      "loss": 3.3705,
+      "step": 3195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.374562920162143e-06,
+      "loss": 3.4393,
+      "step": 3196
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.372806693809283e-06,
+      "loss": 3.4222,
+      "step": 3197
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3710500895433404e-06,
+      "loss": 3.4164,
+      "step": 3198
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.369293107644084e-06,
+      "loss": 3.3884,
+      "step": 3199
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.367535748391349e-06,
+      "loss": 3.3506,
+      "step": 3200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.365778012065024e-06,
+      "loss": 3.5318,
+      "step": 3201
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.364019898945063e-06,
+      "loss": 3.3879,
+      "step": 3202
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3622614093114785e-06,
+      "loss": 3.3815,
+      "step": 3203
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.360502543444339e-06,
+      "loss": 3.3756,
+      "step": 3204
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3587433016237786e-06,
+      "loss": 3.3535,
+      "step": 3205
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3569836841299905e-06,
+      "loss": 3.4757,
+      "step": 3206
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.35522369124322e-06,
+      "loss": 3.3884,
+      "step": 3207
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3534633232437854e-06,
+      "loss": 3.2394,
+      "step": 3208
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3517025804120535e-06,
+      "loss": 3.4678,
+      "step": 3209
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.349941463028456e-06,
+      "loss": 3.3677,
+      "step": 3210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.348179971373484e-06,
+      "loss": 3.4594,
+      "step": 3211
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3464181057276864e-06,
+      "loss": 3.432,
+      "step": 3212
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.344655866371673e-06,
+      "loss": 3.4644,
+      "step": 3213
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.342893253586112e-06,
+      "loss": 3.484,
+      "step": 3214
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.341130267651735e-06,
+      "loss": 3.3675,
+      "step": 3215
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.339366908849326e-06,
+      "loss": 3.5067,
+      "step": 3216
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.337603177459734e-06,
+      "loss": 3.465,
+      "step": 3217
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.335839073763866e-06,
+      "loss": 3.511,
+      "step": 3218
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.334074598042686e-06,
+      "loss": 3.2921,
+      "step": 3219
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3323097505772225e-06,
+      "loss": 3.4087,
+      "step": 3220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.330544531648556e-06,
+      "loss": 3.5283,
+      "step": 3221
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.328778941537835e-06,
+      "loss": 3.4624,
+      "step": 3222
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3270129805262594e-06,
+      "loss": 3.425,
+      "step": 3223
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.325246648895089e-06,
+      "loss": 3.4523,
+      "step": 3224
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.323479946925647e-06,
+      "loss": 3.3794,
+      "step": 3225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.321712874899313e-06,
+      "loss": 3.5653,
+      "step": 3226
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.319945433097525e-06,
+      "loss": 3.432,
+      "step": 3227
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.318177621801781e-06,
+      "loss": 3.3951,
+      "step": 3228
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.316409441293637e-06,
+      "loss": 3.3502,
+      "step": 3229
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.314640891854711e-06,
+      "loss": 3.5866,
+      "step": 3230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.312871973766673e-06,
+      "loss": 3.3552,
+      "step": 3231
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.311102687311257e-06,
+      "loss": 3.4463,
+      "step": 3232
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.309333032770256e-06,
+      "loss": 3.4572,
+      "step": 3233
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.307563010425517e-06,
+      "loss": 3.394,
+      "step": 3234
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.3057926205589514e-06,
+      "loss": 3.5005,
+      "step": 3235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.304021863452525e-06,
+      "loss": 3.5608,
+      "step": 3236
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.3022507393882635e-06,
+      "loss": 3.4147,
+      "step": 3237
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.300479248648251e-06,
+      "loss": 3.4354,
+      "step": 3238
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.298707391514632e-06,
+      "loss": 3.4774,
+      "step": 3239
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.296935168269605e-06,
+      "loss": 3.3985,
+      "step": 3240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.295162579195428e-06,
+      "loss": 3.4536,
+      "step": 3241
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.293389624574422e-06,
+      "loss": 3.453,
+      "step": 3242
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.29161630468896e-06,
+      "loss": 3.519,
+      "step": 3243
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.289842619821475e-06,
+      "loss": 3.4361,
+      "step": 3244
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.288068570254463e-06,
+      "loss": 3.4913,
+      "step": 3245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.286294156270469e-06,
+      "loss": 3.4869,
+      "step": 3246
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.284519378152104e-06,
+      "loss": 3.3532,
+      "step": 3247
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.282744236182033e-06,
+      "loss": 3.3942,
+      "step": 3248
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.280968730642982e-06,
+      "loss": 3.4054,
+      "step": 3249
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.27919286181773e-06,
+      "loss": 3.3391,
+      "step": 3250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.277416629989119e-06,
+      "loss": 3.4246,
+      "step": 3251
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2756400354400445e-06,
+      "loss": 3.3856,
+      "step": 3252
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.273863078453463e-06,
+      "loss": 3.5267,
+      "step": 3253
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.272085759312389e-06,
+      "loss": 3.439,
+      "step": 3254
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.27030807829989e-06,
+      "loss": 3.3837,
+      "step": 3255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2685300356990975e-06,
+      "loss": 3.4453,
+      "step": 3256
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.266751631793196e-06,
+      "loss": 3.3796,
+      "step": 3257
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.264972866865429e-06,
+      "loss": 3.4734,
+      "step": 3258
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.263193741199097e-06,
+      "loss": 3.4841,
+      "step": 3259
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.261414255077561e-06,
+      "loss": 3.4787,
+      "step": 3260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.259634408784233e-06,
+      "loss": 3.5317,
+      "step": 3261
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.257854202602588e-06,
+      "loss": 3.4206,
+      "step": 3262
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.256073636816159e-06,
+      "loss": 3.4647,
+      "step": 3263
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.254292711708529e-06,
+      "loss": 3.4491,
+      "step": 3264
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.252511427563347e-06,
+      "loss": 3.4992,
+      "step": 3265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.250729784664316e-06,
+      "loss": 3.4848,
+      "step": 3266
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2489477832951905e-06,
+      "loss": 3.4587,
+      "step": 3267
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.24716542373979e-06,
+      "loss": 3.5197,
+      "step": 3268
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.245382706281989e-06,
+      "loss": 3.3908,
+      "step": 3269
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.243599631205715e-06,
+      "loss": 3.3941,
+      "step": 3270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2418161987949594e-06,
+      "loss": 3.4783,
+      "step": 3271
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.240032409333765e-06,
+      "loss": 3.4853,
+      "step": 3272
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.238248263106231e-06,
+      "loss": 3.3283,
+      "step": 3273
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.236463760396516e-06,
+      "loss": 3.3806,
+      "step": 3274
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.234678901488837e-06,
+      "loss": 3.6099,
+      "step": 3275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.232893686667466e-06,
+      "loss": 3.4903,
+      "step": 3276
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.231108116216727e-06,
+      "loss": 3.3387,
+      "step": 3277
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.229322190421011e-06,
+      "loss": 3.5383,
+      "step": 3278
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.227535909564754e-06,
+      "loss": 3.3073,
+      "step": 3279
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.225749273932457e-06,
+      "loss": 3.4541,
+      "step": 3280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2239622838086745e-06,
+      "loss": 3.2609,
+      "step": 3281
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.222174939478015e-06,
+      "loss": 3.4692,
+      "step": 3282
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.220387241225148e-06,
+      "loss": 3.3593,
+      "step": 3283
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.218599189334799e-06,
+      "loss": 3.5865,
+      "step": 3284
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.216810784091746e-06,
+      "loss": 3.3081,
+      "step": 3285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.215022025780823e-06,
+      "loss": 3.5243,
+      "step": 3286
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.213232914686928e-06,
+      "loss": 3.428,
+      "step": 3287
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.211443451095007e-06,
+      "loss": 3.2883,
+      "step": 3288
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.209653635290063e-06,
+      "loss": 3.4956,
+      "step": 3289
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.207863467557162e-06,
+      "loss": 3.4289,
+      "step": 3290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2060729481814176e-06,
+      "loss": 3.4453,
+      "step": 3291
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.204282077448002e-06,
+      "loss": 3.397,
+      "step": 3292
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.202490855642148e-06,
+      "loss": 3.3965,
+      "step": 3293
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.200699283049138e-06,
+      "loss": 3.4436,
+      "step": 3294
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.198907359954312e-06,
+      "loss": 3.3549,
+      "step": 3295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.197115086643069e-06,
+      "loss": 3.4424,
+      "step": 3296
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.195322463400861e-06,
+      "loss": 3.4246,
+      "step": 3297
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.193529490513195e-06,
+      "loss": 3.4662,
+      "step": 3298
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.191736168265635e-06,
+      "loss": 3.4378,
+      "step": 3299
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.189942496943803e-06,
+      "loss": 3.5275,
+      "step": 3300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.188148476833369e-06,
+      "loss": 3.3979,
+      "step": 3301
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.186354108220071e-06,
+      "loss": 3.4867,
+      "step": 3302
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.18455939138969e-06,
+      "loss": 3.3343,
+      "step": 3303
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.182764326628068e-06,
+      "loss": 3.4289,
+      "step": 3304
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.180968914221104e-06,
+      "loss": 3.4677,
+      "step": 3305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.179173154454748e-06,
+      "loss": 3.4698,
+      "step": 3306
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.177377047615008e-06,
+      "loss": 3.4206,
+      "step": 3307
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.175580593987952e-06,
+      "loss": 3.3849,
+      "step": 3308
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.173783793859691e-06,
+      "loss": 3.4304,
+      "step": 3309
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.171986647516403e-06,
+      "loss": 3.5744,
+      "step": 3310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.170189155244314e-06,
+      "loss": 3.4897,
+      "step": 3311
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1683913173297105e-06,
+      "loss": 3.3919,
+      "step": 3312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.166593134058929e-06,
+      "loss": 3.4341,
+      "step": 3313
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.164794605718366e-06,
+      "loss": 3.4924,
+      "step": 3314
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.162995732594466e-06,
+      "loss": 3.4129,
+      "step": 3315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.161196514973735e-06,
+      "loss": 3.4367,
+      "step": 3316
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.159396953142733e-06,
+      "loss": 3.5418,
+      "step": 3317
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1575970473880705e-06,
+      "loss": 3.4175,
+      "step": 3318
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.155796797996417e-06,
+      "loss": 3.482,
+      "step": 3319
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.153996205254495e-06,
+      "loss": 3.4018,
+      "step": 3320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.152195269449083e-06,
+      "loss": 3.4246,
+      "step": 3321
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.150393990867011e-06,
+      "loss": 3.4464,
+      "step": 3322
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.148592369795165e-06,
+      "loss": 3.4215,
+      "step": 3323
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.146790406520491e-06,
+      "loss": 3.5426,
+      "step": 3324
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1449881013299804e-06,
+      "loss": 3.3577,
+      "step": 3325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.143185454510686e-06,
+      "loss": 3.4747,
+      "step": 3326
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.141382466349711e-06,
+      "loss": 3.4741,
+      "step": 3327
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.139579137134215e-06,
+      "loss": 3.2964,
+      "step": 3328
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.137775467151411e-06,
+      "loss": 3.363,
+      "step": 3329
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.135971456688567e-06,
+      "loss": 3.374,
+      "step": 3330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.134167106033004e-06,
+      "loss": 3.4191,
+      "step": 3331
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.132362415472099e-06,
+      "loss": 3.3937,
+      "step": 3332
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.130557385293282e-06,
+      "loss": 3.4026,
+      "step": 3333
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.128752015784036e-06,
+      "loss": 3.4213,
+      "step": 3334
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.126946307231902e-06,
+      "loss": 3.4567,
+      "step": 3335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.125140259924468e-06,
+      "loss": 3.4842,
+      "step": 3336
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.123333874149385e-06,
+      "loss": 3.3783,
+      "step": 3337
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.121527150194349e-06,
+      "loss": 3.4792,
+      "step": 3338
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.119720088347117e-06,
+      "loss": 3.4363,
+      "step": 3339
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1179126888954954e-06,
+      "loss": 3.3622,
+      "step": 3340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.116104952127348e-06,
+      "loss": 3.4573,
+      "step": 3341
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.114296878330586e-06,
+      "loss": 3.4532,
+      "step": 3342
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.11248846779318e-06,
+      "loss": 3.5121,
+      "step": 3343
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1106797208031554e-06,
+      "loss": 3.3979,
+      "step": 3344
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.108870637648585e-06,
+      "loss": 3.378,
+      "step": 3345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1070612186175995e-06,
+      "loss": 3.4993,
+      "step": 3346
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1052514639983835e-06,
+      "loss": 3.4333,
+      "step": 3347
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1034413740791705e-06,
+      "loss": 3.3826,
+      "step": 3348
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.101630949148255e-06,
+      "loss": 3.4841,
+      "step": 3349
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.099820189493977e-06,
+      "loss": 3.3296,
+      "step": 3350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0980090954047345e-06,
+      "loss": 3.4617,
+      "step": 3351
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.096197667168978e-06,
+      "loss": 3.4235,
+      "step": 3352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.094385905075211e-06,
+      "loss": 3.483,
+      "step": 3353
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0925738094119865e-06,
+      "loss": 3.4126,
+      "step": 3354
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.090761380467919e-06,
+      "loss": 3.4931,
+      "step": 3355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.088948618531668e-06,
+      "loss": 3.4159,
+      "step": 3356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.087135523891949e-06,
+      "loss": 3.5558,
+      "step": 3357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.085322096837532e-06,
+      "loss": 3.4383,
+      "step": 3358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.08350833765724e-06,
+      "loss": 3.333,
+      "step": 3359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.081694246639943e-06,
+      "loss": 3.4428,
+      "step": 3360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.079879824074572e-06,
+      "loss": 3.5359,
+      "step": 3361
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0780650702501066e-06,
+      "loss": 3.3231,
+      "step": 3362
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.076249985455579e-06,
+      "loss": 3.4079,
+      "step": 3363
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0744345699800755e-06,
+      "loss": 3.4562,
+      "step": 3364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.072618824112734e-06,
+      "loss": 3.4866,
+      "step": 3365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.070802748142744e-06,
+      "loss": 3.4995,
+      "step": 3366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.068986342359352e-06,
+      "loss": 3.4369,
+      "step": 3367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.067169607051851e-06,
+      "loss": 3.3848,
+      "step": 3368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.065352542509593e-06,
+      "loss": 3.5268,
+      "step": 3369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.063535149021974e-06,
+      "loss": 3.3851,
+      "step": 3370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.061717426878453e-06,
+      "loss": 3.5486,
+      "step": 3371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.059899376368531e-06,
+      "loss": 3.4497,
+      "step": 3372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.058080997781766e-06,
+      "loss": 3.4452,
+      "step": 3373
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.056262291407773e-06,
+      "loss": 3.3759,
+      "step": 3374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.05444325753621e-06,
+      "loss": 3.3507,
+      "step": 3375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.052623896456794e-06,
+      "loss": 3.5006,
+      "step": 3376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0508042084592896e-06,
+      "loss": 3.3294,
+      "step": 3377
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0489841938335155e-06,
+      "loss": 3.3509,
+      "step": 3378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.047163852869346e-06,
+      "loss": 3.4664,
+      "step": 3379
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.045343185856701e-06,
+      "loss": 3.4859,
+      "step": 3380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0435221930855545e-06,
+      "loss": 3.4554,
+      "step": 3381
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.041700874845934e-06,
+      "loss": 3.3987,
+      "step": 3382
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.039879231427919e-06,
+      "loss": 3.3354,
+      "step": 3383
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.038057263121639e-06,
+      "loss": 3.4844,
+      "step": 3384
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.036234970217276e-06,
+      "loss": 3.2972,
+      "step": 3385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.034412353005062e-06,
+      "loss": 3.3906,
+      "step": 3386
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.032589411775285e-06,
+      "loss": 3.3692,
+      "step": 3387
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.030766146818281e-06,
+      "loss": 3.3074,
+      "step": 3388
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0289425584244365e-06,
+      "loss": 3.5043,
+      "step": 3389
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.027118646884195e-06,
+      "loss": 3.5004,
+      "step": 3390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.025294412488045e-06,
+      "loss": 3.4967,
+      "step": 3391
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.02346985552653e-06,
+      "loss": 3.3734,
+      "step": 3392
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.021644976290245e-06,
+      "loss": 3.4345,
+      "step": 3393
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.019819775069834e-06,
+      "loss": 3.4365,
+      "step": 3394
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.0179942521559975e-06,
+      "loss": 3.4933,
+      "step": 3395
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.016168407839478e-06,
+      "loss": 3.3293,
+      "step": 3396
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.01434224241108e-06,
+      "loss": 3.461,
+      "step": 3397
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.0125157561616505e-06,
+      "loss": 3.399,
+      "step": 3398
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.010688949382093e-06,
+      "loss": 3.351,
+      "step": 3399
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.00886182236336e-06,
+      "loss": 3.4309,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.007034375396453e-06,
+      "loss": 3.3239,
+      "step": 3401
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.005206608772429e-06,
+      "loss": 3.4088,
+      "step": 3402
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.003378522782393e-06,
+      "loss": 3.3457,
+      "step": 3403
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.001550117717499e-06,
+      "loss": 3.3524,
+      "step": 3404
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.999721393868956e-06,
+      "loss": 3.4751,
+      "step": 3405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.997892351528023e-06,
+      "loss": 3.342,
+      "step": 3406
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.996062990986005e-06,
+      "loss": 3.5206,
+      "step": 3407
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.994233312534266e-06,
+      "loss": 3.4798,
+      "step": 3408
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.992403316464211e-06,
+      "loss": 3.4621,
+      "step": 3409
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9905730030673044e-06,
+      "loss": 3.5746,
+      "step": 3410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.988742372635056e-06,
+      "loss": 3.3954,
+      "step": 3411
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.986911425459028e-06,
+      "loss": 3.4437,
+      "step": 3412
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.98508016183083e-06,
+      "loss": 3.4515,
+      "step": 3413
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.983248582042126e-06,
+      "loss": 3.4322,
+      "step": 3414
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.98141668638463e-06,
+      "loss": 3.4273,
+      "step": 3415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.979584475150103e-06,
+      "loss": 3.3999,
+      "step": 3416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9777519486303595e-06,
+      "loss": 3.424,
+      "step": 3417
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.975919107117263e-06,
+      "loss": 3.3822,
+      "step": 3418
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.974085950902726e-06,
+      "loss": 3.3924,
+      "step": 3419
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9722524802787125e-06,
+      "loss": 3.3986,
+      "step": 3420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.970418695537239e-06,
+      "loss": 3.3888,
+      "step": 3421
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9685845969703644e-06,
+      "loss": 3.3967,
+      "step": 3422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.966750184870209e-06,
+      "loss": 3.3511,
+      "step": 3423
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9649154595289326e-06,
+      "loss": 3.5069,
+      "step": 3424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.963080421238749e-06,
+      "loss": 3.4828,
+      "step": 3425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.961245070291922e-06,
+      "loss": 3.5212,
+      "step": 3426
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.959409406980769e-06,
+      "loss": 3.3215,
+      "step": 3427
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.957573431597646e-06,
+      "loss": 3.3954,
+      "step": 3428
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9557371444349694e-06,
+      "loss": 3.5543,
+      "step": 3429
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.953900545785204e-06,
+      "loss": 3.474,
+      "step": 3430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.952063635940858e-06,
+      "loss": 3.3012,
+      "step": 3431
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.950226415194495e-06,
+      "loss": 3.4891,
+      "step": 3432
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.948388883838726e-06,
+      "loss": 3.4296,
+      "step": 3433
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.94655104216621e-06,
+      "loss": 3.3092,
+      "step": 3434
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.944712890469659e-06,
+      "loss": 3.3946,
+      "step": 3435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.942874429041834e-06,
+      "loss": 3.5026,
+      "step": 3436
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.94103565817554e-06,
+      "loss": 3.5051,
+      "step": 3437
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.939196578163637e-06,
+      "loss": 3.4582,
+      "step": 3438
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9373571892990334e-06,
+      "loss": 3.5678,
+      "step": 3439
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9355174918746834e-06,
+      "loss": 3.4511,
+      "step": 3440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9336774861835945e-06,
+      "loss": 3.5224,
+      "step": 3441
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.93183717251882e-06,
+      "loss": 3.5578,
+      "step": 3442
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9299965511734635e-06,
+      "loss": 3.5068,
+      "step": 3443
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.92815562244068e-06,
+      "loss": 3.4028,
+      "step": 3444
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.92631438661367e-06,
+      "loss": 3.5745,
+      "step": 3445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9244728439856814e-06,
+      "loss": 3.4866,
+      "step": 3446
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.922630994850017e-06,
+      "loss": 3.3891,
+      "step": 3447
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.920788839500025e-06,
+      "loss": 3.4633,
+      "step": 3448
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.918946378229103e-06,
+      "loss": 3.3343,
+      "step": 3449
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.917103611330694e-06,
+      "loss": 3.3827,
+      "step": 3450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.915260539098296e-06,
+      "loss": 3.3718,
+      "step": 3451
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.913417161825449e-06,
+      "loss": 3.4237,
+      "step": 3452
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.911573479805747e-06,
+      "loss": 3.3764,
+      "step": 3453
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.90972949333283e-06,
+      "loss": 3.3833,
+      "step": 3454
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.907885202700384e-06,
+      "loss": 3.4807,
+      "step": 3455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.906040608202149e-06,
+      "loss": 3.4834,
+      "step": 3456
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.904195710131912e-06,
+      "loss": 3.4342,
+      "step": 3457
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.902350508783502e-06,
+      "loss": 3.3997,
+      "step": 3458
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.900505004450803e-06,
+      "loss": 3.2648,
+      "step": 3459
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.898659197427748e-06,
+      "loss": 3.4193,
+      "step": 3460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.896813088008315e-06,
+      "loss": 3.4235,
+      "step": 3461
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.89496667648653e-06,
+      "loss": 3.4381,
+      "step": 3462
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.893119963156467e-06,
+      "loss": 3.4199,
+      "step": 3463
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.891272948312251e-06,
+      "loss": 3.5042,
+      "step": 3464
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.889425632248051e-06,
+      "loss": 3.5979,
+      "step": 3465
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.8875780152580896e-06,
+      "loss": 3.3515,
+      "step": 3466
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.885730097636629e-06,
+      "loss": 3.4224,
+      "step": 3467
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.8838818796779875e-06,
+      "loss": 3.4579,
+      "step": 3468
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.882033361676526e-06,
+      "loss": 3.4391,
+      "step": 3469
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.880184543926656e-06,
+      "loss": 3.467,
+      "step": 3470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.8783354267228345e-06,
+      "loss": 3.3655,
+      "step": 3471
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.876486010359569e-06,
+      "loss": 3.5059,
+      "step": 3472
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.8746362951314115e-06,
+      "loss": 3.4054,
+      "step": 3473
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.872786281332965e-06,
+      "loss": 3.4164,
+      "step": 3474
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.870935969258879e-06,
+      "loss": 3.4988,
+      "step": 3475
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.869085359203844e-06,
+      "loss": 3.3671,
+      "step": 3476
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.86723445146261e-06,
+      "loss": 3.4453,
+      "step": 3477
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.865383246329967e-06,
+      "loss": 3.4408,
+      "step": 3478
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.863531744100749e-06,
+      "loss": 3.4357,
+      "step": 3479
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.861679945069846e-06,
+      "loss": 3.3538,
+      "step": 3480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8598278495321904e-06,
+      "loss": 3.583,
+      "step": 3481
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8579754577827606e-06,
+      "loss": 3.4353,
+      "step": 3482
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.856122770116585e-06,
+      "loss": 3.4231,
+      "step": 3483
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.854269786828741e-06,
+      "loss": 3.3194,
+      "step": 3484
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.852416508214346e-06,
+      "loss": 3.4137,
+      "step": 3485
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8505629345685706e-06,
+      "loss": 3.4163,
+      "step": 3486
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.848709066186632e-06,
+      "loss": 3.5052,
+      "step": 3487
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.84685490336379e-06,
+      "loss": 3.4363,
+      "step": 3488
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.845000446395354e-06,
+      "loss": 3.4154,
+      "step": 3489
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8431456955766855e-06,
+      "loss": 3.3656,
+      "step": 3490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8412906512031805e-06,
+      "loss": 3.4756,
+      "step": 3491
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.839435313570293e-06,
+      "loss": 3.4884,
+      "step": 3492
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.837579682973519e-06,
+      "loss": 3.4057,
+      "step": 3493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8357237597084015e-06,
+      "loss": 3.5419,
+      "step": 3494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.833867544070531e-06,
+      "loss": 3.5138,
+      "step": 3495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.832011036355543e-06,
+      "loss": 3.4595,
+      "step": 3496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.83015423685912e-06,
+      "loss": 3.3717,
+      "step": 3497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.828297145876993e-06,
+      "loss": 3.4323,
+      "step": 3498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.82643976370494e-06,
+      "loss": 3.4042,
+      "step": 3499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.824582090638777e-06,
+      "loss": 3.3599,
+      "step": 3500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.822724126974377e-06,
+      "loss": 3.4909,
+      "step": 3501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.820865873007656e-06,
+      "loss": 3.4746,
+      "step": 3502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.819007329034572e-06,
+      "loss": 3.4713,
+      "step": 3503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.817148495351131e-06,
+      "loss": 3.399,
+      "step": 3504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.815289372253393e-06,
+      "loss": 3.311,
+      "step": 3505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.813429960037451e-06,
+      "loss": 3.3403,
+      "step": 3506
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.811570258999453e-06,
+      "loss": 3.4324,
+      "step": 3507
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.80971026943559e-06,
+      "loss": 3.4073,
+      "step": 3508
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.807849991642102e-06,
+      "loss": 3.5075,
+      "step": 3509
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.805989425915269e-06,
+      "loss": 3.5586,
+      "step": 3510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8041285725514215e-06,
+      "loss": 3.4868,
+      "step": 3511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.802267431846935e-06,
+      "loss": 3.4967,
+      "step": 3512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.800406004098231e-06,
+      "loss": 3.542,
+      "step": 3513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7985442896017765e-06,
+      "loss": 3.4979,
+      "step": 3514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.79668228865408e-06,
+      "loss": 3.4759,
+      "step": 3515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.794820001551703e-06,
+      "loss": 3.419,
+      "step": 3516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7929574285912495e-06,
+      "loss": 3.4169,
+      "step": 3517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.791094570069366e-06,
+      "loss": 3.5095,
+      "step": 3518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.789231426282745e-06,
+      "loss": 3.5828,
+      "step": 3519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.787367997528133e-06,
+      "loss": 3.4315,
+      "step": 3520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.785504284102309e-06,
+      "loss": 3.3157,
+      "step": 3521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.783640286302109e-06,
+      "loss": 3.4398,
+      "step": 3522
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.781776004424403e-06,
+      "loss": 3.5147,
+      "step": 3523
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.779911438766117e-06,
+      "loss": 3.3945,
+      "step": 3524
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7780465896242155e-06,
+      "loss": 3.4422,
+      "step": 3525
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.77618145729571e-06,
+      "loss": 3.5565,
+      "step": 3526
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.774316042077658e-06,
+      "loss": 3.452,
+      "step": 3527
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.772450344267159e-06,
+      "loss": 3.3778,
+      "step": 3528
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.770584364161361e-06,
+      "loss": 3.4425,
+      "step": 3529
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.768718102057457e-06,
+      "loss": 3.3876,
+      "step": 3530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.76685155825268e-06,
+      "loss": 3.4245,
+      "step": 3531
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.764984733044316e-06,
+      "loss": 3.45,
+      "step": 3532
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.763117626729687e-06,
+      "loss": 3.4384,
+      "step": 3533
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7612502396061685e-06,
+      "loss": 3.4545,
+      "step": 3534
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.759382571971171e-06,
+      "loss": 3.4402,
+      "step": 3535
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.757514624122159e-06,
+      "loss": 3.4944,
+      "step": 3536
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.755646396356637e-06,
+      "loss": 3.447,
+      "step": 3537
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7537778889721514e-06,
+      "loss": 3.4158,
+      "step": 3538
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7519091022662986e-06,
+      "loss": 3.468,
+      "step": 3539
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.750040036536718e-06,
+      "loss": 3.3757,
+      "step": 3540
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.748170692081093e-06,
+      "loss": 3.3919,
+      "step": 3541
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.746301069197148e-06,
+      "loss": 3.4844,
+      "step": 3542
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.744431168182659e-06,
+      "loss": 3.4826,
+      "step": 3543
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.742560989335438e-06,
+      "loss": 3.5232,
+      "step": 3544
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.740690532953348e-06,
+      "loss": 3.3975,
+      "step": 3545
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.738819799334293e-06,
+      "loss": 3.363,
+      "step": 3546
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.736948788776221e-06,
+      "loss": 3.4121,
+      "step": 3547
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.735077501577126e-06,
+      "loss": 3.4677,
+      "step": 3548
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.733205938035045e-06,
+      "loss": 3.4119,
+      "step": 3549
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.731334098448056e-06,
+      "loss": 3.4597,
+      "step": 3550
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.729461983114287e-06,
+      "loss": 3.4864,
+      "step": 3551
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.727589592331907e-06,
+      "loss": 3.361,
+      "step": 3552
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.725716926399127e-06,
+      "loss": 3.4122,
+      "step": 3553
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.723843985614201e-06,
+      "loss": 3.4138,
+      "step": 3554
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.721970770275435e-06,
+      "loss": 3.4222,
+      "step": 3555
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.72009728068117e-06,
+      "loss": 3.4593,
+      "step": 3556
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.718223517129792e-06,
+      "loss": 3.3559,
+      "step": 3557
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.716349479919737e-06,
+      "loss": 3.3803,
+      "step": 3558
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.714475169349475e-06,
+      "loss": 3.6088,
+      "step": 3559
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.712600585717526e-06,
+      "loss": 3.5179,
+      "step": 3560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.710725729322454e-06,
+      "loss": 3.4501,
+      "step": 3561
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.708850600462862e-06,
+      "loss": 3.4886,
+      "step": 3562
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.7069751994373975e-06,
+      "loss": 3.4998,
+      "step": 3563
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.705099526544757e-06,
+      "loss": 3.4048,
+      "step": 3564
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.703223582083672e-06,
+      "loss": 3.3808,
+      "step": 3565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.701347366352923e-06,
+      "loss": 3.4793,
+      "step": 3566
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.699470879651331e-06,
+      "loss": 3.4189,
+      "step": 3567
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.697594122277762e-06,
+      "loss": 3.3808,
+      "step": 3568
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.695717094531122e-06,
+      "loss": 3.3597,
+      "step": 3569
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.693839796710365e-06,
+      "loss": 3.4599,
+      "step": 3570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.691962229114484e-06,
+      "loss": 3.5254,
+      "step": 3571
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.690084392042514e-06,
+      "loss": 3.5481,
+      "step": 3572
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.688206285793539e-06,
+      "loss": 3.4165,
+      "step": 3573
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.686327910666679e-06,
+      "loss": 3.528,
+      "step": 3574
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.684449266961101e-06,
+      "loss": 3.3694,
+      "step": 3575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.682570354976013e-06,
+      "loss": 3.4766,
+      "step": 3576
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.680691175010668e-06,
+      "loss": 3.4847,
+      "step": 3577
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.678811727364356e-06,
+      "loss": 3.4796,
+      "step": 3578
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.676932012336419e-06,
+      "loss": 3.4366,
+      "step": 3579
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.675052030226231e-06,
+      "loss": 3.3969,
+      "step": 3580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.673171781333217e-06,
+      "loss": 3.3657,
+      "step": 3581
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.671291265956839e-06,
+      "loss": 3.506,
+      "step": 3582
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.669410484396606e-06,
+      "loss": 3.5007,
+      "step": 3583
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.667529436952064e-06,
+      "loss": 3.5942,
+      "step": 3584
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.665648123922809e-06,
+      "loss": 3.3947,
+      "step": 3585
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.66376654560847e-06,
+      "loss": 3.4433,
+      "step": 3586
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.661884702308725e-06,
+      "loss": 3.4017,
+      "step": 3587
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6600025943232935e-06,
+      "loss": 3.4894,
+      "step": 3588
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.658120221951932e-06,
+      "loss": 3.4626,
+      "step": 3589
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.656237585494447e-06,
+      "loss": 3.465,
+      "step": 3590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.654354685250681e-06,
+      "loss": 3.3983,
+      "step": 3591
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.652471521520519e-06,
+      "loss": 3.4428,
+      "step": 3592
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.650588094603893e-06,
+      "loss": 3.5212,
+      "step": 3593
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6487044048007696e-06,
+      "loss": 3.5038,
+      "step": 3594
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.646820452411165e-06,
+      "loss": 3.4207,
+      "step": 3595
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.644936237735129e-06,
+      "loss": 3.4966,
+      "step": 3596
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.643051761072761e-06,
+      "loss": 3.4271,
+      "step": 3597
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.641167022724196e-06,
+      "loss": 3.4137,
+      "step": 3598
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.639282022989615e-06,
+      "loss": 3.5173,
+      "step": 3599
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.637396762169238e-06,
+      "loss": 3.453,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.635511240563328e-06,
+      "loss": 3.501,
+      "step": 3601
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.633625458472187e-06,
+      "loss": 3.4262,
+      "step": 3602
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.631739416196165e-06,
+      "loss": 3.3938,
+      "step": 3603
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.629853114035643e-06,
+      "loss": 3.4918,
+      "step": 3604
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.627966552291053e-06,
+      "loss": 3.5442,
+      "step": 3605
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.626079731262864e-06,
+      "loss": 3.4592,
+      "step": 3606
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.624192651251586e-06,
+      "loss": 3.4483,
+      "step": 3607
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.622305312557773e-06,
+      "loss": 3.3176,
+      "step": 3608
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6204177154820195e-06,
+      "loss": 3.5149,
+      "step": 3609
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.618529860324955e-06,
+      "loss": 3.4634,
+      "step": 3610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.616641747387258e-06,
+      "loss": 3.46,
+      "step": 3611
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.614753376969647e-06,
+      "loss": 3.4948,
+      "step": 3612
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6128647493728755e-06,
+      "loss": 3.4365,
+      "step": 3613
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.610975864897746e-06,
+      "loss": 3.4338,
+      "step": 3614
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6090867238450985e-06,
+      "loss": 3.4541,
+      "step": 3615
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.607197326515808e-06,
+      "loss": 3.4013,
+      "step": 3616
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.605307673210801e-06,
+      "loss": 3.4648,
+      "step": 3617
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.603417764231039e-06,
+      "loss": 3.4447,
+      "step": 3618
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.601527599877522e-06,
+      "loss": 3.5161,
+      "step": 3619
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.599637180451295e-06,
+      "loss": 3.4339,
+      "step": 3620
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.597746506253444e-06,
+      "loss": 3.5403,
+      "step": 3621
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.59585557758509e-06,
+      "loss": 3.4837,
+      "step": 3622
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5939643947474e-06,
+      "loss": 3.2757,
+      "step": 3623
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5920729580415795e-06,
+      "loss": 3.4561,
+      "step": 3624
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.590181267768876e-06,
+      "loss": 3.5066,
+      "step": 3625
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5882893242305724e-06,
+      "loss": 3.4189,
+      "step": 3626
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.586397127727999e-06,
+      "loss": 3.3949,
+      "step": 3627
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.584504678562521e-06,
+      "loss": 3.3999,
+      "step": 3628
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.582611977035547e-06,
+      "loss": 3.3984,
+      "step": 3629
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.580719023448523e-06,
+      "loss": 3.4966,
+      "step": 3630
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.578825818102936e-06,
+      "loss": 3.3954,
+      "step": 3631
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.576932361300315e-06,
+      "loss": 3.3823,
+      "step": 3632
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.57503865334223e-06,
+      "loss": 3.4669,
+      "step": 3633
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.573144694530287e-06,
+      "loss": 3.497,
+      "step": 3634
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.571250485166132e-06,
+      "loss": 3.5353,
+      "step": 3635
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.569356025551454e-06,
+      "loss": 3.4679,
+      "step": 3636
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5674613159879815e-06,
+      "loss": 3.4748,
+      "step": 3637
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.565566356777481e-06,
+      "loss": 3.4199,
+      "step": 3638
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5636711482217595e-06,
+      "loss": 3.3867,
+      "step": 3639
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.561775690622663e-06,
+      "loss": 3.4772,
+      "step": 3640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.55987998428208e-06,
+      "loss": 3.4586,
+      "step": 3641
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.557984029501935e-06,
+      "loss": 3.3846,
+      "step": 3642
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.556087826584192e-06,
+      "loss": 3.4068,
+      "step": 3643
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.554191375830861e-06,
+      "loss": 3.538,
+      "step": 3644
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.552294677543983e-06,
+      "loss": 3.4746,
+      "step": 3645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.550397732025643e-06,
+      "loss": 3.5175,
+      "step": 3646
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.548500539577964e-06,
+      "loss": 3.3597,
+      "step": 3647
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.54660310050311e-06,
+      "loss": 3.3873,
+      "step": 3648
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5447054151032805e-06,
+      "loss": 3.4638,
+      "step": 3649
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.54280748368072e-06,
+      "loss": 3.3845,
+      "step": 3650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.540909306537708e-06,
+      "loss": 3.4838,
+      "step": 3651
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.539010883976562e-06,
+      "loss": 3.4072,
+      "step": 3652
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.537112216299643e-06,
+      "loss": 3.5342,
+      "step": 3653
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.535213303809349e-06,
+      "loss": 3.4261,
+      "step": 3654
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.533314146808113e-06,
+      "loss": 3.417,
+      "step": 3655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5314147455984155e-06,
+      "loss": 3.4911,
+      "step": 3656
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.529515100482768e-06,
+      "loss": 3.4864,
+      "step": 3657
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.527615211763725e-06,
+      "loss": 3.3502,
+      "step": 3658
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.525715079743878e-06,
+      "loss": 3.4519,
+      "step": 3659
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.523814704725861e-06,
+      "loss": 3.5408,
+      "step": 3660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.521914087012338e-06,
+      "loss": 3.4074,
+      "step": 3661
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.520013226906021e-06,
+      "loss": 3.392,
+      "step": 3662
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.518112124709659e-06,
+      "loss": 3.468,
+      "step": 3663
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.516210780726032e-06,
+      "loss": 3.3914,
+      "step": 3664
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.514309195257967e-06,
+      "loss": 3.3807,
+      "step": 3665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.512407368608329e-06,
+      "loss": 3.5649,
+      "step": 3666
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5105053010800135e-06,
+      "loss": 3.5308,
+      "step": 3667
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.508602992975963e-06,
+      "loss": 3.4553,
+      "step": 3668
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.506700444599153e-06,
+      "loss": 3.4397,
+      "step": 3669
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.504797656252601e-06,
+      "loss": 3.3373,
+      "step": 3670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.502894628239363e-06,
+      "loss": 3.4049,
+      "step": 3671
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.500991360862528e-06,
+      "loss": 3.3712,
+      "step": 3672
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.499087854425225e-06,
+      "loss": 3.3458,
+      "step": 3673
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.497184109230628e-06,
+      "loss": 3.3601,
+      "step": 3674
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.495280125581937e-06,
+      "loss": 3.3992,
+      "step": 3675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.493375903782402e-06,
+      "loss": 3.2951,
+      "step": 3676
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4914714441353e-06,
+      "loss": 3.4325,
+      "step": 3677
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.489566746943956e-06,
+      "loss": 3.3939,
+      "step": 3678
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.487661812511725e-06,
+      "loss": 3.5123,
+      "step": 3679
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.485756641142005e-06,
+      "loss": 3.3469,
+      "step": 3680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.483851233138227e-06,
+      "loss": 3.5026,
+      "step": 3681
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.481945588803864e-06,
+      "loss": 3.4789,
+      "step": 3682
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.480039708442425e-06,
+      "loss": 3.5422,
+      "step": 3683
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.478133592357455e-06,
+      "loss": 3.473,
+      "step": 3684
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.476227240852539e-06,
+      "loss": 3.4938,
+      "step": 3685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.474320654231299e-06,
+      "loss": 3.5152,
+      "step": 3686
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.472413832797394e-06,
+      "loss": 3.3693,
+      "step": 3687
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.470506776854519e-06,
+      "loss": 3.3477,
+      "step": 3688
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.468599486706408e-06,
+      "loss": 3.4923,
+      "step": 3689
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.466691962656835e-06,
+      "loss": 3.4387,
+      "step": 3690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.464784205009604e-06,
+      "loss": 3.3461,
+      "step": 3691
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.462876214068563e-06,
+      "loss": 3.4635,
+      "step": 3692
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4609679901375935e-06,
+      "loss": 3.4088,
+      "step": 3693
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4590595335206154e-06,
+      "loss": 3.3847,
+      "step": 3694
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4571508445215866e-06,
+      "loss": 3.3753,
+      "step": 3695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.455241923444499e-06,
+      "loss": 3.3579,
+      "step": 3696
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.453332770593385e-06,
+      "loss": 3.4542,
+      "step": 3697
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.451423386272312e-06,
+      "loss": 3.439,
+      "step": 3698
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.449513770785385e-06,
+      "loss": 3.4101,
+      "step": 3699
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.447603924436744e-06,
+      "loss": 3.4356,
+      "step": 3700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.445693847530567e-06,
+      "loss": 3.4356,
+      "step": 3701
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.443783540371072e-06,
+      "loss": 3.3949,
+      "step": 3702
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.441873003262506e-06,
+      "loss": 3.3947,
+      "step": 3703
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.43996223650916e-06,
+      "loss": 3.4188,
+      "step": 3704
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.438051240415357e-06,
+      "loss": 3.3427,
+      "step": 3705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.436140015285461e-06,
+      "loss": 3.3255,
+      "step": 3706
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.434228561423869e-06,
+      "loss": 3.4285,
+      "step": 3707
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.432316879135012e-06,
+      "loss": 3.365,
+      "step": 3708
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4304049687233624e-06,
+      "loss": 3.5272,
+      "step": 3709
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4284928304934306e-06,
+      "loss": 3.3405,
+      "step": 3710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4265804647497545e-06,
+      "loss": 3.4743,
+      "step": 3711
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4246678717969145e-06,
+      "loss": 3.4208,
+      "step": 3712
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.422755051939528e-06,
+      "loss": 3.495,
+      "step": 3713
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.420842005482248e-06,
+      "loss": 3.3828,
+      "step": 3714
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.418928732729759e-06,
+      "loss": 3.5657,
+      "step": 3715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.417015233986786e-06,
+      "loss": 3.4066,
+      "step": 3716
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.415101509558089e-06,
+      "loss": 3.4133,
+      "step": 3717
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.413187559748465e-06,
+      "loss": 3.5086,
+      "step": 3718
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.411273384862744e-06,
+      "loss": 3.5581,
+      "step": 3719
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.409358985205793e-06,
+      "loss": 3.3287,
+      "step": 3720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.407444361082517e-06,
+      "loss": 3.4312,
+      "step": 3721
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.405529512797857e-06,
+      "loss": 3.3648,
+      "step": 3722
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.403614440656782e-06,
+      "loss": 3.4165,
+      "step": 3723
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.401699144964306e-06,
+      "loss": 3.3364,
+      "step": 3724
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.399783626025478e-06,
+      "loss": 3.4211,
+      "step": 3725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.397867884145372e-06,
+      "loss": 3.4958,
+      "step": 3726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.395951919629112e-06,
+      "loss": 3.3583,
+      "step": 3727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.394035732781847e-06,
+      "loss": 3.4827,
+      "step": 3728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3921193239087665e-06,
+      "loss": 3.3201,
+      "step": 3729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.390202693315093e-06,
+      "loss": 3.4325,
+      "step": 3730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.388285841306088e-06,
+      "loss": 3.3642,
+      "step": 3731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.38636876818704e-06,
+      "loss": 3.302,
+      "step": 3732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.384451474263283e-06,
+      "loss": 3.3796,
+      "step": 3733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.382533959840178e-06,
+      "loss": 3.5754,
+      "step": 3734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3806162252231265e-06,
+      "loss": 3.4572,
+      "step": 3735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.378698270717562e-06,
+      "loss": 3.3647,
+      "step": 3736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.376780096628955e-06,
+      "loss": 3.3427,
+      "step": 3737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.374861703262809e-06,
+      "loss": 3.4387,
+      "step": 3738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3729430909246625e-06,
+      "loss": 3.4132,
+      "step": 3739
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.371024259920091e-06,
+      "loss": 3.4335,
+      "step": 3740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3691052105547056e-06,
+      "loss": 3.5447,
+      "step": 3741
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3671859431341464e-06,
+      "loss": 3.3039,
+      "step": 3742
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.365266457964095e-06,
+      "loss": 3.4364,
+      "step": 3743
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3633467553502625e-06,
+      "loss": 3.4252,
+      "step": 3744
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.361426835598399e-06,
+      "loss": 3.4646,
+      "step": 3745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.359506699014287e-06,
+      "loss": 3.3647,
+      "step": 3746
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.357586345903739e-06,
+      "loss": 3.5235,
+      "step": 3747
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3556657765726116e-06,
+      "loss": 3.3618,
+      "step": 3748
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.353744991326789e-06,
+      "loss": 3.434,
+      "step": 3749
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.351823990472192e-06,
+      "loss": 3.4902,
+      "step": 3750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3499027743147745e-06,
+      "loss": 3.4619,
+      "step": 3751
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.347981343160526e-06,
+      "loss": 3.3587,
+      "step": 3752
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.34605969731547e-06,
+      "loss": 3.4226,
+      "step": 3753
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.344137837085662e-06,
+      "loss": 3.3668,
+      "step": 3754
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.342215762777197e-06,
+      "loss": 3.5118,
+      "step": 3755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.340293474696198e-06,
+      "loss": 3.4122,
+      "step": 3756
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.338370973148826e-06,
+      "loss": 3.4769,
+      "step": 3757
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.336448258441276e-06,
+      "loss": 3.5421,
+      "step": 3758
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.334525330879772e-06,
+      "loss": 3.4678,
+      "step": 3759
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.332602190770576e-06,
+      "loss": 3.4195,
+      "step": 3760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.330678838419988e-06,
+      "loss": 3.4451,
+      "step": 3761
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.328755274134333e-06,
+      "loss": 3.3914,
+      "step": 3762
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.326831498219974e-06,
+      "loss": 3.5655,
+      "step": 3763
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.32490751098331e-06,
+      "loss": 3.3571,
+      "step": 3764
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.322983312730769e-06,
+      "loss": 3.4144,
+      "step": 3765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.321058903768815e-06,
+      "loss": 3.5275,
+      "step": 3766
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.31913428440395e-06,
+      "loss": 3.5014,
+      "step": 3767
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.317209454942699e-06,
+      "loss": 3.335,
+      "step": 3768
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.315284415691629e-06,
+      "loss": 3.4207,
+      "step": 3769
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.31335916695734e-06,
+      "loss": 3.4986,
+      "step": 3770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.311433709046459e-06,
+      "loss": 3.5559,
+      "step": 3771
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.309508042265654e-06,
+      "loss": 3.4692,
+      "step": 3772
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.307582166921622e-06,
+      "loss": 3.3376,
+      "step": 3773
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.305656083321092e-06,
+      "loss": 3.4235,
+      "step": 3774
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.303729791770831e-06,
+      "loss": 3.4504,
+      "step": 3775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.301803292577635e-06,
+      "loss": 3.3219,
+      "step": 3776
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.299876586048334e-06,
+      "loss": 3.3734,
+      "step": 3777
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.297949672489792e-06,
+      "loss": 3.329,
+      "step": 3778
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2960225522089054e-06,
+      "loss": 3.3831,
+      "step": 3779
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.294095225512604e-06,
+      "loss": 3.4054,
+      "step": 3780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.292167692707849e-06,
+      "loss": 3.5047,
+      "step": 3781
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.290239954101638e-06,
+      "loss": 3.4623,
+      "step": 3782
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.288312010000994e-06,
+      "loss": 3.4384,
+      "step": 3783
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.286383860712982e-06,
+      "loss": 3.5079,
+      "step": 3784
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.284455506544695e-06,
+      "loss": 3.4356,
+      "step": 3785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.282526947803255e-06,
+      "loss": 3.4063,
+      "step": 3786
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.280598184795823e-06,
+      "loss": 3.3964,
+      "step": 3787
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.27866921782959e-06,
+      "loss": 3.3707,
+      "step": 3788
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.27674004721178e-06,
+      "loss": 3.425,
+      "step": 3789
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.274810673249647e-06,
+      "loss": 3.4305,
+      "step": 3790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.272881096250482e-06,
+      "loss": 3.3123,
+      "step": 3791
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.270951316521602e-06,
+      "loss": 3.4454,
+      "step": 3792
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.269021334370362e-06,
+      "loss": 3.3806,
+      "step": 3793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2670911501041495e-06,
+      "loss": 3.5146,
+      "step": 3794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.265160764030378e-06,
+      "loss": 3.431,
+      "step": 3795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.263230176456497e-06,
+      "loss": 3.5281,
+      "step": 3796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.261299387689992e-06,
+      "loss": 3.3982,
+      "step": 3797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.259368398038373e-06,
+      "loss": 3.4651,
+      "step": 3798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2574372078091845e-06,
+      "loss": 3.4128,
+      "step": 3799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.255505817310009e-06,
+      "loss": 3.3536,
+      "step": 3800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.253574226848451e-06,
+      "loss": 3.4366,
+      "step": 3801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.251642436732155e-06,
+      "loss": 3.3397,
+      "step": 3802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.249710447268793e-06,
+      "loss": 3.4922,
+      "step": 3803
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.247778258766069e-06,
+      "loss": 3.4569,
+      "step": 3804
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.24584587153172e-06,
+      "loss": 3.453,
+      "step": 3805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2439132858735166e-06,
+      "loss": 3.3736,
+      "step": 3806
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.241980502099255e-06,
+      "loss": 3.3113,
+      "step": 3807
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.240047520516768e-06,
+      "loss": 3.5234,
+      "step": 3808
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.23811434143392e-06,
+      "loss": 3.4458,
+      "step": 3809
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.236180965158601e-06,
+      "loss": 3.4341,
+      "step": 3810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.234247391998741e-06,
+      "loss": 3.4697,
+      "step": 3811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.232313622262297e-06,
+      "loss": 3.5389,
+      "step": 3812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.230379656257255e-06,
+      "loss": 3.5699,
+      "step": 3813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.228445494291635e-06,
+      "loss": 3.3709,
+      "step": 3814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.226511136673488e-06,
+      "loss": 3.5101,
+      "step": 3815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.224576583710898e-06,
+      "loss": 3.2622,
+      "step": 3816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.222641835711974e-06,
+      "loss": 3.4751,
+      "step": 3817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.220706892984866e-06,
+      "loss": 3.4692,
+      "step": 3818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.218771755837743e-06,
+      "loss": 3.4362,
+      "step": 3819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.216836424578816e-06,
+      "loss": 3.4494,
+      "step": 3820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.21490089951632e-06,
+      "loss": 3.4214,
+      "step": 3821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.212965180958524e-06,
+      "loss": 3.4834,
+      "step": 3822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.211029269213724e-06,
+      "loss": 3.3895,
+      "step": 3823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.209093164590253e-06,
+      "loss": 3.371,
+      "step": 3824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.207156867396469e-06,
+      "loss": 3.5184,
+      "step": 3825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.205220377940763e-06,
+      "loss": 3.495,
+      "step": 3826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.203283696531558e-06,
+      "loss": 3.4296,
+      "step": 3827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.2013468234773034e-06,
+      "loss": 3.5064,
+      "step": 3828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.199409759086487e-06,
+      "loss": 3.352,
+      "step": 3829
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.197472503667616e-06,
+      "loss": 3.4485,
+      "step": 3830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.195535057529237e-06,
+      "loss": 3.4835,
+      "step": 3831
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.193597420979925e-06,
+      "loss": 3.4046,
+      "step": 3832
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.191659594328281e-06,
+      "loss": 3.4984,
+      "step": 3833
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.189721577882942e-06,
+      "loss": 3.4799,
+      "step": 3834
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.187783371952572e-06,
+      "loss": 3.4042,
+      "step": 3835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.185844976845866e-06,
+      "loss": 3.4223,
+      "step": 3836
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.183906392871548e-06,
+      "loss": 3.4358,
+      "step": 3837
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1819676203383745e-06,
+      "loss": 3.5227,
+      "step": 3838
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.180028659555132e-06,
+      "loss": 3.2857,
+      "step": 3839
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.178089510830631e-06,
+      "loss": 3.3956,
+      "step": 3840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.17615017447372e-06,
+      "loss": 3.3646,
+      "step": 3841
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1742106507932756e-06,
+      "loss": 3.2658,
+      "step": 3842
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1722709400981984e-06,
+      "loss": 3.3783,
+      "step": 3843
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.170331042697425e-06,
+      "loss": 3.4222,
+      "step": 3844
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.168390958899921e-06,
+      "loss": 3.4327,
+      "step": 3845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.166450689014679e-06,
+      "loss": 3.396,
+      "step": 3846
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.164510233350723e-06,
+      "loss": 3.5084,
+      "step": 3847
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1625695922171055e-06,
+      "loss": 3.3178,
+      "step": 3848
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.160628765922909e-06,
+      "loss": 3.3574,
+      "step": 3849
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.158687754777247e-06,
+      "loss": 3.3379,
+      "step": 3850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.156746559089262e-06,
+      "loss": 3.4642,
+      "step": 3851
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.154805179168122e-06,
+      "loss": 3.4206,
+      "step": 3852
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.15286361532303e-06,
+      "loss": 3.4014,
+      "step": 3853
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.150921867863215e-06,
+      "loss": 3.3809,
+      "step": 3854
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.148979937097937e-06,
+      "loss": 3.3988,
+      "step": 3855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.147037823336482e-06,
+      "loss": 3.3106,
+      "step": 3856
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.14509552688817e-06,
+      "loss": 3.3861,
+      "step": 3857
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.143153048062345e-06,
+      "loss": 3.3799,
+      "step": 3858
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.141210387168383e-06,
+      "loss": 3.4195,
+      "step": 3859
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.139267544515689e-06,
+      "loss": 3.3555,
+      "step": 3860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1373245204136965e-06,
+      "loss": 3.3923,
+      "step": 3861
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.135381315171867e-06,
+      "loss": 3.4213,
+      "step": 3862
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.133437929099693e-06,
+      "loss": 3.3516,
+      "step": 3863
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.131494362506693e-06,
+      "loss": 3.4071,
+      "step": 3864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.129550615702416e-06,
+      "loss": 3.3658,
+      "step": 3865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.12760668899644e-06,
+      "loss": 3.4609,
+      "step": 3866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.125662582698371e-06,
+      "loss": 3.4675,
+      "step": 3867
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.123718297117844e-06,
+      "loss": 3.5352,
+      "step": 3868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1217738325645216e-06,
+      "loss": 3.4106,
+      "step": 3869
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.119829189348095e-06,
+      "loss": 3.461,
+      "step": 3870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.117884367778286e-06,
+      "loss": 3.4089,
+      "step": 3871
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.115939368164842e-06,
+      "loss": 3.4475,
+      "step": 3872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.113994190817539e-06,
+      "loss": 3.4023,
+      "step": 3873
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.112048836046185e-06,
+      "loss": 3.4931,
+      "step": 3874
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.110103304160611e-06,
+      "loss": 3.3921,
+      "step": 3875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.10815759547068e-06,
+      "loss": 3.5016,
+      "step": 3876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.106211710286281e-06,
+      "loss": 3.4544,
+      "step": 3877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1042656489173325e-06,
+      "loss": 3.495,
+      "step": 3878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.10231941167378e-06,
+      "loss": 3.4125,
+      "step": 3879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1003729988656005e-06,
+      "loss": 3.5497,
+      "step": 3880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.098426410802791e-06,
+      "loss": 3.473,
+      "step": 3881
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.096479647795385e-06,
+      "loss": 3.4836,
+      "step": 3882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.09453271015344e-06,
+      "loss": 3.4381,
+      "step": 3883
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.09258559818704e-06,
+      "loss": 3.35,
+      "step": 3884
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.090638312206298e-06,
+      "loss": 3.4273,
+      "step": 3885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.088690852521358e-06,
+      "loss": 3.4545,
+      "step": 3886
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.086743219442388e-06,
+      "loss": 3.4342,
+      "step": 3887
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.084795413279583e-06,
+      "loss": 3.3246,
+      "step": 3888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.0828474343431664e-06,
+      "loss": 3.3635,
+      "step": 3889
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.080899282943391e-06,
+      "loss": 3.2696,
+      "step": 3890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.078950959390536e-06,
+      "loss": 3.4994,
+      "step": 3891
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.077002463994908e-06,
+      "loss": 3.3789,
+      "step": 3892
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.075053797066837e-06,
+      "loss": 3.3783,
+      "step": 3893
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0731049589166895e-06,
+      "loss": 3.4147,
+      "step": 3894
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.071155949854851e-06,
+      "loss": 3.5186,
+      "step": 3895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.069206770191737e-06,
+      "loss": 3.3939,
+      "step": 3896
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.06725742023779e-06,
+      "loss": 3.4407,
+      "step": 3897
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0653079003034805e-06,
+      "loss": 3.3309,
+      "step": 3898
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.063358210699305e-06,
+      "loss": 3.4444,
+      "step": 3899
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.061408351735787e-06,
+      "loss": 3.3863,
+      "step": 3900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.059458323723479e-06,
+      "loss": 3.2684,
+      "step": 3901
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.057508126972956e-06,
+      "loss": 3.4499,
+      "step": 3902
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.055557761794826e-06,
+      "loss": 3.4078,
+      "step": 3903
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.053607228499719e-06,
+      "loss": 3.3486,
+      "step": 3904
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.051656527398293e-06,
+      "loss": 3.3389,
+      "step": 3905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.049705658801233e-06,
+      "loss": 3.4114,
+      "step": 3906
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.047754623019251e-06,
+      "loss": 3.3001,
+      "step": 3907
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.045803420363085e-06,
+      "loss": 3.5619,
+      "step": 3908
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0438520511435e-06,
+      "loss": 3.3861,
+      "step": 3909
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.041900515671289e-06,
+      "loss": 3.3632,
+      "step": 3910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.039948814257266e-06,
+      "loss": 3.419,
+      "step": 3911
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.037996947212278e-06,
+      "loss": 3.3523,
+      "step": 3912
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.036044914847197e-06,
+      "loss": 3.3937,
+      "step": 3913
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.034092717472917e-06,
+      "loss": 3.4767,
+      "step": 3914
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.032140355400362e-06,
+      "loss": 3.3742,
+      "step": 3915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.030187828940485e-06,
+      "loss": 3.4718,
+      "step": 3916
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.028235138404258e-06,
+      "loss": 3.3002,
+      "step": 3917
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.026282284102683e-06,
+      "loss": 3.4767,
+      "step": 3918
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0243292663467915e-06,
+      "loss": 3.4611,
+      "step": 3919
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.022376085447632e-06,
+      "loss": 3.4476,
+      "step": 3920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.020422741716288e-06,
+      "loss": 3.4556,
+      "step": 3921
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.018469235463866e-06,
+      "loss": 3.5679,
+      "step": 3922
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.016515567001495e-06,
+      "loss": 3.3956,
+      "step": 3923
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.014561736640334e-06,
+      "loss": 3.4116,
+      "step": 3924
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0126077446915685e-06,
+      "loss": 3.344,
+      "step": 3925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.010653591466403e-06,
+      "loss": 3.4116,
+      "step": 3926
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.008699277276076e-06,
+      "loss": 3.4607,
+      "step": 3927
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.006744802431849e-06,
+      "loss": 3.4326,
+      "step": 3928
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0047901672450035e-06,
+      "loss": 3.4335,
+      "step": 3929
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.002835372026854e-06,
+      "loss": 3.4062,
+      "step": 3930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.000880417088738e-06,
+      "loss": 3.5098,
+      "step": 3931
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.998925302742017e-06,
+      "loss": 3.4484,
+      "step": 3932
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9969700292980785e-06,
+      "loss": 3.4153,
+      "step": 3933
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9950145970683375e-06,
+      "loss": 3.4599,
+      "step": 3934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.993059006364229e-06,
+      "loss": 3.4544,
+      "step": 3935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.991103257497221e-06,
+      "loss": 3.4621,
+      "step": 3936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9891473507788e-06,
+      "loss": 3.4725,
+      "step": 3937
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.98719128652048e-06,
+      "loss": 3.5611,
+      "step": 3938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.985235065033801e-06,
+      "loss": 3.5192,
+      "step": 3939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.983278686630327e-06,
+      "loss": 3.5233,
+      "step": 3940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.981322151621648e-06,
+      "loss": 3.5957,
+      "step": 3941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9793654603193775e-06,
+      "loss": 3.3958,
+      "step": 3942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.977408613035154e-06,
+      "loss": 3.3433,
+      "step": 3943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.975451610080643e-06,
+      "loss": 3.3319,
+      "step": 3944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9734944517675295e-06,
+      "loss": 3.3203,
+      "step": 3945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.971537138407532e-06,
+      "loss": 3.3374,
+      "step": 3946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9695796703123855e-06,
+      "loss": 3.4968,
+      "step": 3947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.967622047793853e-06,
+      "loss": 3.4006,
+      "step": 3948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.965664271163724e-06,
+      "loss": 3.5837,
+      "step": 3949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9637063407338065e-06,
+      "loss": 3.3491,
+      "step": 3950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.961748256815939e-06,
+      "loss": 3.4089,
+      "step": 3951
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.959790019721985e-06,
+      "loss": 3.4685,
+      "step": 3952
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.957831629763825e-06,
+      "loss": 3.4896,
+      "step": 3953
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.955873087253371e-06,
+      "loss": 3.4406,
+      "step": 3954
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.953914392502559e-06,
+      "loss": 3.4623,
+      "step": 3955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.951955545823342e-06,
+      "loss": 3.3645,
+      "step": 3956
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.949996547527707e-06,
+      "loss": 3.3869,
+      "step": 3957
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.948037397927659e-06,
+      "loss": 3.4249,
+      "step": 3958
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.946078097335227e-06,
+      "loss": 3.3908,
+      "step": 3959
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9441186460624675e-06,
+      "loss": 3.4921,
+      "step": 3960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.94215904442146e-06,
+      "loss": 3.4687,
+      "step": 3961
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.940199292724304e-06,
+      "loss": 3.5223,
+      "step": 3962
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.938239391283127e-06,
+      "loss": 3.4589,
+      "step": 3963
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.936279340410082e-06,
+      "loss": 3.4284,
+      "step": 3964
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9343191404173395e-06,
+      "loss": 3.3293,
+      "step": 3965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9323587916170985e-06,
+      "loss": 3.5258,
+      "step": 3966
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.930398294321583e-06,
+      "loss": 3.3049,
+      "step": 3967
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.928437648843036e-06,
+      "loss": 3.4688,
+      "step": 3968
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.926476855493726e-06,
+      "loss": 3.4469,
+      "step": 3969
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.924515914585947e-06,
+      "loss": 3.433,
+      "step": 3970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.922554826432013e-06,
+      "loss": 3.4855,
+      "step": 3971
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.920593591344264e-06,
+      "loss": 3.5656,
+      "step": 3972
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9186322096350645e-06,
+      "loss": 3.4503,
+      "step": 3973
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.916670681616797e-06,
+      "loss": 3.3533,
+      "step": 3974
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.914709007601873e-06,
+      "loss": 3.4627,
+      "step": 3975
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.912747187902727e-06,
+      "loss": 3.4402,
+      "step": 3976
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.910785222831812e-06,
+      "loss": 3.3196,
+      "step": 3977
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.908823112701607e-06,
+      "loss": 3.2371,
+      "step": 3978
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.906860857824618e-06,
+      "loss": 3.3585,
+      "step": 3979
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.904898458513365e-06,
+      "loss": 3.3063,
+      "step": 3980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.9029359150803985e-06,
+      "loss": 3.3861,
+      "step": 3981
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.90097322783829e-06,
+      "loss": 3.4239,
+      "step": 3982
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.899010397099634e-06,
+      "loss": 3.4108,
+      "step": 3983
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8970474231770445e-06,
+      "loss": 3.3684,
+      "step": 3984
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.895084306383166e-06,
+      "loss": 3.4236,
+      "step": 3985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.893121047030654e-06,
+      "loss": 3.3867,
+      "step": 3986
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.891157645432199e-06,
+      "loss": 3.5776,
+      "step": 3987
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8891941019005095e-06,
+      "loss": 3.3919,
+      "step": 3988
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.887230416748312e-06,
+      "loss": 3.4058,
+      "step": 3989
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.885266590288361e-06,
+      "loss": 3.4561,
+      "step": 3990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.883302622833434e-06,
+      "loss": 3.4701,
+      "step": 3991
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.881338514696327e-06,
+      "loss": 3.5038,
+      "step": 3992
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.879374266189859e-06,
+      "loss": 3.4844,
+      "step": 3993
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.877409877626876e-06,
+      "loss": 3.4026,
+      "step": 3994
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.875445349320241e-06,
+      "loss": 3.5578,
+      "step": 3995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.87348068158284e-06,
+      "loss": 3.2667,
+      "step": 3996
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.871515874727587e-06,
+      "loss": 3.2819,
+      "step": 3997
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.869550929067407e-06,
+      "loss": 3.4811,
+      "step": 3998
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.867585844915259e-06,
+      "loss": 3.4818,
+      "step": 3999
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.865620622584116e-06,
+      "loss": 3.4437,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.863655262386978e-06,
+      "loss": 3.5231,
+      "step": 4001
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.861689764636863e-06,
+      "loss": 3.341,
+      "step": 4002
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.859724129646814e-06,
+      "loss": 3.4024,
+      "step": 4003
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.857758357729892e-06,
+      "loss": 3.4202,
+      "step": 4004
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.855792449199185e-06,
+      "loss": 3.2676,
+      "step": 4005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8538264043678e-06,
+      "loss": 3.4128,
+      "step": 4006
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.851860223548863e-06,
+      "loss": 3.3008,
+      "step": 4007
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.849893907055527e-06,
+      "loss": 3.4343,
+      "step": 4008
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.847927455200963e-06,
+      "loss": 3.402,
+      "step": 4009
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.845960868298366e-06,
+      "loss": 3.4579,
+      "step": 4010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.843994146660949e-06,
+      "loss": 3.4592,
+      "step": 4011
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8420272906019506e-06,
+      "loss": 3.4341,
+      "step": 4012
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.840060300434627e-06,
+      "loss": 3.4699,
+      "step": 4013
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.838093176472259e-06,
+      "loss": 3.4486,
+      "step": 4014
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.836125919028146e-06,
+      "loss": 3.3955,
+      "step": 4015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8341585284156114e-06,
+      "loss": 3.4287,
+      "step": 4016
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.832191004947996e-06,
+      "loss": 3.4517,
+      "step": 4017
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.830223348938668e-06,
+      "loss": 3.4486,
+      "step": 4018
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.828255560701008e-06,
+      "loss": 3.4755,
+      "step": 4019
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.826287640548425e-06,
+      "loss": 3.4828,
+      "step": 4020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.824319588794347e-06,
+      "loss": 3.4679,
+      "step": 4021
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.822351405752221e-06,
+      "loss": 3.5252,
+      "step": 4022
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.820383091735518e-06,
+      "loss": 3.5329,
+      "step": 4023
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8184146470577265e-06,
+      "loss": 3.4133,
+      "step": 4024
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.816446072032357e-06,
+      "loss": 3.3697,
+      "step": 4025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.814477366972945e-06,
+      "loss": 3.4245,
+      "step": 4026
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.812508532193039e-06,
+      "loss": 3.5211,
+      "step": 4027
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.810539568006213e-06,
+      "loss": 3.3747,
+      "step": 4028
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.808570474726062e-06,
+      "loss": 3.3525,
+      "step": 4029
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8066012526661985e-06,
+      "loss": 3.3001,
+      "step": 4030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8046319021402595e-06,
+      "loss": 3.4935,
+      "step": 4031
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.802662423461898e-06,
+      "loss": 3.4773,
+      "step": 4032
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.800692816944793e-06,
+      "loss": 3.3914,
+      "step": 4033
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.798723082902636e-06,
+      "loss": 3.4562,
+      "step": 4034
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.796753221649147e-06,
+      "loss": 3.411,
+      "step": 4035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.794783233498062e-06,
+      "loss": 3.4778,
+      "step": 4036
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7928131187631345e-06,
+      "loss": 3.5465,
+      "step": 4037
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.790842877758147e-06,
+      "loss": 3.4338,
+      "step": 4038
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.788872510796892e-06,
+      "loss": 3.4637,
+      "step": 4039
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.786902018193189e-06,
+      "loss": 3.4229,
+      "step": 4040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.784931400260875e-06,
+      "loss": 3.3803,
+      "step": 4041
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7829606573138065e-06,
+      "loss": 3.3767,
+      "step": 4042
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.78098978966586e-06,
+      "loss": 3.4408,
+      "step": 4043
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.779018797630934e-06,
+      "loss": 3.5112,
+      "step": 4044
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.777047681522943e-06,
+      "loss": 3.329,
+      "step": 4045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.775076441655827e-06,
+      "loss": 3.4136,
+      "step": 4046
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.773105078343539e-06,
+      "loss": 3.3026,
+      "step": 4047
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.771133591900056e-06,
+      "loss": 3.4026,
+      "step": 4048
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.769161982639372e-06,
+      "loss": 3.441,
+      "step": 4049
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.767190250875505e-06,
+      "loss": 3.4858,
+      "step": 4050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.765218396922487e-06,
+      "loss": 3.5187,
+      "step": 4051
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.763246421094373e-06,
+      "loss": 3.4718,
+      "step": 4052
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7612743237052374e-06,
+      "loss": 3.4533,
+      "step": 4053
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.75930210506917e-06,
+      "loss": 3.3884,
+      "step": 4054
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.757329765500286e-06,
+      "loss": 3.525,
+      "step": 4055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.755357305312716e-06,
+      "loss": 3.2697,
+      "step": 4056
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.753384724820608e-06,
+      "loss": 3.3652,
+      "step": 4057
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.751412024338135e-06,
+      "loss": 3.5848,
+      "step": 4058
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.749439204179484e-06,
+      "loss": 3.5578,
+      "step": 4059
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.747466264658863e-06,
+      "loss": 3.2798,
+      "step": 4060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7454932060904974e-06,
+      "loss": 3.3484,
+      "step": 4061
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.743520028788638e-06,
+      "loss": 3.3722,
+      "step": 4062
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.741546733067547e-06,
+      "loss": 3.3329,
+      "step": 4063
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.739573319241505e-06,
+      "loss": 3.4591,
+      "step": 4064
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.737599787624819e-06,
+      "loss": 3.4264,
+      "step": 4065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.735626138531807e-06,
+      "loss": 3.4742,
+      "step": 4066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.733652372276809e-06,
+      "loss": 3.4411,
+      "step": 4067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.731678489174186e-06,
+      "loss": 3.4385,
+      "step": 4068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.729704489538313e-06,
+      "loss": 3.3618,
+      "step": 4069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.727730373683586e-06,
+      "loss": 3.4845,
+      "step": 4070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7257561419244225e-06,
+      "loss": 3.4109,
+      "step": 4071
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.72378179457525e-06,
+      "loss": 3.4095,
+      "step": 4072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.721807331950522e-06,
+      "loss": 3.3583,
+      "step": 4073
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.719832754364707e-06,
+      "loss": 3.4501,
+      "step": 4074
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.717858062132295e-06,
+      "loss": 3.4763,
+      "step": 4075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7158832555677904e-06,
+      "loss": 3.3036,
+      "step": 4076
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.713908334985718e-06,
+      "loss": 3.5464,
+      "step": 4077
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.71193330070062e-06,
+      "loss": 3.4041,
+      "step": 4078
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.709958153027057e-06,
+      "loss": 3.5297,
+      "step": 4079
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7079828922796086e-06,
+      "loss": 3.5222,
+      "step": 4080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.706007518772869e-06,
+      "loss": 3.5139,
+      "step": 4081
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7040320328214535e-06,
+      "loss": 3.5883,
+      "step": 4082
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.702056434739996e-06,
+      "loss": 3.3738,
+      "step": 4083
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7000807248431466e-06,
+      "loss": 3.5153,
+      "step": 4084
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.698104903445572e-06,
+      "loss": 3.3852,
+      "step": 4085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.696128970861957e-06,
+      "loss": 3.4115,
+      "step": 4086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.694152927407008e-06,
+      "loss": 3.4929,
+      "step": 4087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.692176773395446e-06,
+      "loss": 3.3403,
+      "step": 4088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6902005091420095e-06,
+      "loss": 3.43,
+      "step": 4089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.688224134961454e-06,
+      "loss": 3.361,
+      "step": 4090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.686247651168554e-06,
+      "loss": 3.4565,
+      "step": 4091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.684271058078101e-06,
+      "loss": 3.5072,
+      "step": 4092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.682294356004903e-06,
+      "loss": 3.3646,
+      "step": 4093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.680317545263786e-06,
+      "loss": 3.3387,
+      "step": 4094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6783406261695964e-06,
+      "loss": 3.4143,
+      "step": 4095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.67636359903719e-06,
+      "loss": 3.4207,
+      "step": 4096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.674386464181449e-06,
+      "loss": 3.4525,
+      "step": 4097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.672409221917265e-06,
+      "loss": 3.482,
+      "step": 4098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6704318725595546e-06,
+      "loss": 3.4829,
+      "step": 4099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.668454416423243e-06,
+      "loss": 3.3713,
+      "step": 4100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.666476853823278e-06,
+      "loss": 3.3698,
+      "step": 4101
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.664499185074623e-06,
+      "loss": 3.5407,
+      "step": 4102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.662521410492257e-06,
+      "loss": 3.3548,
+      "step": 4103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.66054353039118e-06,
+      "loss": 3.4917,
+      "step": 4104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.658565545086401e-06,
+      "loss": 3.3814,
+      "step": 4105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.656587454892955e-06,
+      "loss": 3.4765,
+      "step": 4106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.654609260125886e-06,
+      "loss": 3.3999,
+      "step": 4107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.65263096110026e-06,
+      "loss": 3.3436,
+      "step": 4108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.650652558131155e-06,
+      "loss": 3.516,
+      "step": 4109
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.648674051533671e-06,
+      "loss": 3.3724,
+      "step": 4110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.646695441622918e-06,
+      "loss": 3.3523,
+      "step": 4111
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.64471672871403e-06,
+      "loss": 3.3481,
+      "step": 4112
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.642737913122152e-06,
+      "loss": 3.4166,
+      "step": 4113
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.640758995162446e-06,
+      "loss": 3.3962,
+      "step": 4114
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.638779975150092e-06,
+      "loss": 3.4375,
+      "step": 4115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.636800853400285e-06,
+      "loss": 3.5121,
+      "step": 4116
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.634821630228235e-06,
+      "loss": 3.391,
+      "step": 4117
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.632842305949171e-06,
+      "loss": 3.4581,
+      "step": 4118
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.630862880878338e-06,
+      "loss": 3.4918,
+      "step": 4119
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.628883355330993e-06,
+      "loss": 3.4915,
+      "step": 4120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.626903729622414e-06,
+      "loss": 3.3762,
+      "step": 4121
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6249240040678935e-06,
+      "loss": 3.3785,
+      "step": 4122
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.622944178982734e-06,
+      "loss": 3.4587,
+      "step": 4123
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.620964254682267e-06,
+      "loss": 3.4319,
+      "step": 4124
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.618984231481825e-06,
+      "loss": 3.4203,
+      "step": 4125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.617004109696768e-06,
+      "loss": 3.4004,
+      "step": 4126
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.615023889642463e-06,
+      "loss": 3.4512,
+      "step": 4127
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.613043571634299e-06,
+      "loss": 3.4795,
+      "step": 4128
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6110631559876775e-06,
+      "loss": 3.4048,
+      "step": 4129
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.609082643018013e-06,
+      "loss": 3.5119,
+      "step": 4130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.607102033040744e-06,
+      "loss": 3.4787,
+      "step": 4131
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.605121326371316e-06,
+      "loss": 3.4981,
+      "step": 4132
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.603140523325192e-06,
+      "loss": 3.4245,
+      "step": 4133
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.601159624217854e-06,
+      "loss": 3.4272,
+      "step": 4134
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.599178629364794e-06,
+      "loss": 3.3503,
+      "step": 4135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.5971975390815236e-06,
+      "loss": 3.3888,
+      "step": 4136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.595216353683568e-06,
+      "loss": 3.4504,
+      "step": 4137
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.593235073486465e-06,
+      "loss": 3.3336,
+      "step": 4138
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.591253698805773e-06,
+      "loss": 3.4276,
+      "step": 4139
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.589272229957061e-06,
+      "loss": 3.4326,
+      "step": 4140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.587290667255913e-06,
+      "loss": 3.4134,
+      "step": 4141
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.585309011017931e-06,
+      "loss": 3.486,
+      "step": 4142
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.583327261558731e-06,
+      "loss": 3.4345,
+      "step": 4143
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.58134541919394e-06,
+      "loss": 3.5114,
+      "step": 4144
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.579363484239205e-06,
+      "loss": 3.3583,
+      "step": 4145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.577381457010186e-06,
+      "loss": 3.5091,
+      "step": 4146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.575399337822555e-06,
+      "loss": 3.3843,
+      "step": 4147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.573417126992004e-06,
+      "loss": 3.3402,
+      "step": 4148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.571434824834233e-06,
+      "loss": 3.4259,
+      "step": 4149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.569452431664962e-06,
+      "loss": 3.4477,
+      "step": 4150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5674699477999235e-06,
+      "loss": 3.382,
+      "step": 4151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.565487373554867e-06,
+      "loss": 3.4277,
+      "step": 4152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.563504709245548e-06,
+      "loss": 3.3768,
+      "step": 4153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5615219551877476e-06,
+      "loss": 3.3264,
+      "step": 4154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.559539111697254e-06,
+      "loss": 3.3344,
+      "step": 4155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5575561790898705e-06,
+      "loss": 3.4747,
+      "step": 4156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.555573157681416e-06,
+      "loss": 3.5261,
+      "step": 4157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.553590047787723e-06,
+      "loss": 3.371,
+      "step": 4158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.55160684972464e-06,
+      "loss": 3.3669,
+      "step": 4159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.549623563808026e-06,
+      "loss": 3.4233,
+      "step": 4160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.547640190353756e-06,
+      "loss": 3.5381,
+      "step": 4161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.545656729677717e-06,
+      "loss": 3.423,
+      "step": 4162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.543673182095816e-06,
+      "loss": 3.3591,
+      "step": 4163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5416895479239665e-06,
+      "loss": 3.4274,
+      "step": 4164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.539705827478099e-06,
+      "loss": 3.5655,
+      "step": 4165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5377220210741566e-06,
+      "loss": 3.4015,
+      "step": 4166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5357381290280986e-06,
+      "loss": 3.3703,
+      "step": 4167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5337541516558945e-06,
+      "loss": 3.3479,
+      "step": 4168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.531770089273532e-06,
+      "loss": 3.4664,
+      "step": 4169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5297859421970055e-06,
+      "loss": 3.461,
+      "step": 4170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.527801710742331e-06,
+      "loss": 3.4463,
+      "step": 4171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.52581739522553e-06,
+      "loss": 3.4885,
+      "step": 4172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5238329959626435e-06,
+      "loss": 3.4329,
+      "step": 4173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.521848513269723e-06,
+      "loss": 3.4832,
+      "step": 4174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.519863947462836e-06,
+      "loss": 3.4292,
+      "step": 4175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.517879298858057e-06,
+      "loss": 3.5256,
+      "step": 4176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5158945677714804e-06,
+      "loss": 3.574,
+      "step": 4177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.513909754519211e-06,
+      "loss": 3.48,
+      "step": 4178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.511924859417367e-06,
+      "loss": 3.4589,
+      "step": 4179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.509939882782077e-06,
+      "loss": 3.3395,
+      "step": 4180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.507954824929488e-06,
+      "loss": 3.4292,
+      "step": 4181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.505969686175756e-06,
+      "loss": 3.4189,
+      "step": 4182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.50398446683705e-06,
+      "loss": 3.4453,
+      "step": 4183
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.501999167229554e-06,
+      "loss": 3.3939,
+      "step": 4184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.500013787669462e-06,
+      "loss": 3.3562,
+      "step": 4185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.498028328472983e-06,
+      "loss": 3.3365,
+      "step": 4186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.496042789956339e-06,
+      "loss": 3.4115,
+      "step": 4187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.494057172435761e-06,
+      "loss": 3.3691,
+      "step": 4188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.492071476227496e-06,
+      "loss": 3.3202,
+      "step": 4189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.490085701647805e-06,
+      "loss": 3.4102,
+      "step": 4190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.488099849012954e-06,
+      "loss": 3.4243,
+      "step": 4191
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.486113918639232e-06,
+      "loss": 3.3638,
+      "step": 4192
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.484127910842932e-06,
+      "loss": 3.478,
+      "step": 4193
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.482141825940363e-06,
+      "loss": 3.4825,
+      "step": 4194
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.480155664247846e-06,
+      "loss": 3.4524,
+      "step": 4195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.478169426081712e-06,
+      "loss": 3.3969,
+      "step": 4196
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4761831117583075e-06,
+      "loss": 3.3523,
+      "step": 4197
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.474196721593988e-06,
+      "loss": 3.5302,
+      "step": 4198
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.472210255905127e-06,
+      "loss": 3.3287,
+      "step": 4199
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4702237150081e-06,
+      "loss": 3.4171,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.468237099219302e-06,
+      "loss": 3.3087,
+      "step": 4201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.466250408855142e-06,
+      "loss": 3.4992,
+      "step": 4202
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4642636442320305e-06,
+      "loss": 3.509,
+      "step": 4203
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4622768056664e-06,
+      "loss": 3.4975,
+      "step": 4204
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4602898934746905e-06,
+      "loss": 3.4311,
+      "step": 4205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.458302907973355e-06,
+      "loss": 3.4135,
+      "step": 4206
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.456315849478857e-06,
+      "loss": 3.2512,
+      "step": 4207
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.454328718307671e-06,
+      "loss": 3.4694,
+      "step": 4208
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4523415147762844e-06,
+      "loss": 3.3864,
+      "step": 4209
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4503542392011975e-06,
+      "loss": 3.4803,
+      "step": 4210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.448366891898919e-06,
+      "loss": 3.3869,
+      "step": 4211
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.446379473185972e-06,
+      "loss": 3.5282,
+      "step": 4212
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.444391983378888e-06,
+      "loss": 3.4687,
+      "step": 4213
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.442404422794212e-06,
+      "loss": 3.4179,
+      "step": 4214
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4404167917484985e-06,
+      "loss": 3.3911,
+      "step": 4215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.438429090558316e-06,
+      "loss": 3.3911,
+      "step": 4216
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.436441319540241e-06,
+      "loss": 3.4439,
+      "step": 4217
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.434453479010863e-06,
+      "loss": 3.474,
+      "step": 4218
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4324655692867825e-06,
+      "loss": 3.3311,
+      "step": 4219
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4304775906846095e-06,
+      "loss": 3.4494,
+      "step": 4220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.428489543520967e-06,
+      "loss": 3.4297,
+      "step": 4221
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.426501428112489e-06,
+      "loss": 3.4297,
+      "step": 4222
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.424513244775818e-06,
+      "loss": 3.4553,
+      "step": 4223
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.42252499382761e-06,
+      "loss": 3.4015,
+      "step": 4224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.420536675584529e-06,
+      "loss": 3.4893,
+      "step": 4225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.418548290363254e-06,
+      "loss": 3.4386,
+      "step": 4226
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.416559838480468e-06,
+      "loss": 3.4312,
+      "step": 4227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.41457132025287e-06,
+      "loss": 3.4622,
+      "step": 4228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.41258273599717e-06,
+      "loss": 3.4439,
+      "step": 4229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.410594086030084e-06,
+      "loss": 3.4434,
+      "step": 4230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.408605370668342e-06,
+      "loss": 3.4487,
+      "step": 4231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.406616590228684e-06,
+      "loss": 3.4495,
+      "step": 4232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.404627745027861e-06,
+      "loss": 3.3438,
+      "step": 4233
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.402638835382628e-06,
+      "loss": 3.3889,
+      "step": 4234
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.400649861609763e-06,
+      "loss": 3.3746,
+      "step": 4235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.398660824026039e-06,
+      "loss": 3.4966,
+      "step": 4236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.396671722948253e-06,
+      "loss": 3.4355,
+      "step": 4237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.394682558693204e-06,
+      "loss": 3.4638,
+      "step": 4238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.392693331577701e-06,
+      "loss": 3.3979,
+      "step": 4239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.390704041918565e-06,
+      "loss": 3.4501,
+      "step": 4240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.388714690032631e-06,
+      "loss": 3.3642,
+      "step": 4241
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.386725276236736e-06,
+      "loss": 3.3889,
+      "step": 4242
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3847358008477315e-06,
+      "loss": 3.3229,
+      "step": 4243
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.38274626418248e-06,
+      "loss": 3.3208,
+      "step": 4244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.380756666557848e-06,
+      "loss": 3.3996,
+      "step": 4245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.378767008290718e-06,
+      "loss": 3.4505,
+      "step": 4246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3767772896979806e-06,
+      "loss": 3.3523,
+      "step": 4247
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.374787511096533e-06,
+      "loss": 3.5341,
+      "step": 4248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.372797672803285e-06,
+      "loss": 3.3572,
+      "step": 4249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.370807775135155e-06,
+      "loss": 3.4168,
+      "step": 4250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.368817818409071e-06,
+      "loss": 3.4963,
+      "step": 4251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.366827802941968e-06,
+      "loss": 3.383,
+      "step": 4252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.364837729050797e-06,
+      "loss": 3.545,
+      "step": 4253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.362847597052509e-06,
+      "loss": 3.4319,
+      "step": 4254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.360857407264071e-06,
+      "loss": 3.4826,
+      "step": 4255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.358867160002459e-06,
+      "loss": 3.5126,
+      "step": 4256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.356876855584654e-06,
+      "loss": 3.5417,
+      "step": 4257
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.354886494327648e-06,
+      "loss": 3.4188,
+      "step": 4258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.352896076548447e-06,
+      "loss": 3.3537,
+      "step": 4259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3509056025640575e-06,
+      "loss": 3.5419,
+      "step": 4260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.348915072691501e-06,
+      "loss": 3.4079,
+      "step": 4261
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3469244872478046e-06,
+      "loss": 3.38,
+      "step": 4262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.344933846550007e-06,
+      "loss": 3.5055,
+      "step": 4263
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3429431509151515e-06,
+      "loss": 3.5705,
+      "step": 4264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.340952400660299e-06,
+      "loss": 3.5062,
+      "step": 4265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.338961596102506e-06,
+      "loss": 3.4193,
+      "step": 4266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.336970737558849e-06,
+      "loss": 3.3753,
+      "step": 4267
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.334979825346409e-06,
+      "loss": 3.5173,
+      "step": 4268
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.332988859782273e-06,
+      "loss": 3.3553,
+      "step": 4269
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.330997841183538e-06,
+      "loss": 3.5535,
+      "step": 4270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.329006769867316e-06,
+      "loss": 3.487,
+      "step": 4271
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.327015646150716e-06,
+      "loss": 3.434,
+      "step": 4272
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.325024470350862e-06,
+      "loss": 3.4372,
+      "step": 4273
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.323033242784889e-06,
+      "loss": 3.383,
+      "step": 4274
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.321041963769933e-06,
+      "loss": 3.4237,
+      "step": 4275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.319050633623141e-06,
+      "loss": 3.4029,
+      "step": 4276
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3170592526616745e-06,
+      "loss": 3.3427,
+      "step": 4277
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.315067821202691e-06,
+      "loss": 3.3664,
+      "step": 4278
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.313076339563366e-06,
+      "loss": 3.3459,
+      "step": 4279
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.31108480806088e-06,
+      "loss": 3.4959,
+      "step": 4280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.309093227012418e-06,
+      "loss": 3.4774,
+      "step": 4281
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.307101596735178e-06,
+      "loss": 3.472,
+      "step": 4282
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.305109917546364e-06,
+      "loss": 3.3894,
+      "step": 4283
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.303118189763187e-06,
+      "loss": 3.4082,
+      "step": 4284
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.301126413702865e-06,
+      "loss": 3.4315,
+      "step": 4285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.299134589682629e-06,
+      "loss": 3.464,
+      "step": 4286
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2971427180197086e-06,
+      "loss": 3.4657,
+      "step": 4287
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.295150799031349e-06,
+      "loss": 3.4646,
+      "step": 4288
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.293158833034799e-06,
+      "loss": 3.4708,
+      "step": 4289
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.291166820347316e-06,
+      "loss": 3.3527,
+      "step": 4290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2891747612861645e-06,
+      "loss": 3.4773,
+      "step": 4291
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.287182656168618e-06,
+      "loss": 3.4066,
+      "step": 4292
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2851905053119525e-06,
+      "loss": 3.387,
+      "step": 4293
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.283198309033457e-06,
+      "loss": 3.4333,
+      "step": 4294
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.281206067650424e-06,
+      "loss": 3.4635,
+      "step": 4295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.279213781480157e-06,
+      "loss": 3.3984,
+      "step": 4296
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2772214508399625e-06,
+      "loss": 3.4944,
+      "step": 4297
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.275229076047156e-06,
+      "loss": 3.4506,
+      "step": 4298
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2732366574190595e-06,
+      "loss": 3.3135,
+      "step": 4299
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.271244195273002e-06,
+      "loss": 3.3988,
+      "step": 4300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.269251689926322e-06,
+      "loss": 3.469,
+      "step": 4301
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2672591416963605e-06,
+      "loss": 3.3967,
+      "step": 4302
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2652665509004665e-06,
+      "loss": 3.4414,
+      "step": 4303
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.263273917856e-06,
+      "loss": 3.4484,
+      "step": 4304
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.261281242880322e-06,
+      "loss": 3.346,
+      "step": 4305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2592885262908015e-06,
+      "loss": 3.5164,
+      "step": 4306
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.257295768404818e-06,
+      "loss": 3.3598,
+      "step": 4307
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.255302969539753e-06,
+      "loss": 3.4428,
+      "step": 4308
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.253310130012997e-06,
+      "loss": 3.3545,
+      "step": 4309
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.251317250141949e-06,
+      "loss": 3.449,
+      "step": 4310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.249324330244006e-06,
+      "loss": 3.4108,
+      "step": 4311
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.247331370636582e-06,
+      "loss": 3.3881,
+      "step": 4312
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.245338371637091e-06,
+      "loss": 3.4979,
+      "step": 4313
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.243345333562954e-06,
+      "loss": 3.468,
+      "step": 4314
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.241352256731599e-06,
+      "loss": 3.4236,
+      "step": 4315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.239359141460461e-06,
+      "loss": 3.3916,
+      "step": 4316
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.237365988066979e-06,
+      "loss": 3.4054,
+      "step": 4317
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2353727968686e-06,
+      "loss": 3.3936,
+      "step": 4318
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.233379568182778e-06,
+      "loss": 3.4898,
+      "step": 4319
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2313863023269675e-06,
+      "loss": 3.4972,
+      "step": 4320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.229392999618636e-06,
+      "loss": 3.4668,
+      "step": 4321
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.227399660375253e-06,
+      "loss": 3.4163,
+      "step": 4322
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.225406284914293e-06,
+      "loss": 3.4398,
+      "step": 4323
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.22341287355324e-06,
+      "loss": 3.4429,
+      "step": 4324
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.22141942660958e-06,
+      "loss": 3.4159,
+      "step": 4325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.219425944400805e-06,
+      "loss": 3.5319,
+      "step": 4326
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.217432427244414e-06,
+      "loss": 3.3083,
+      "step": 4327
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.215438875457914e-06,
+      "loss": 3.4201,
+      "step": 4328
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.213445289358813e-06,
+      "loss": 3.4807,
+      "step": 4329
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.211451669264624e-06,
+      "loss": 3.3392,
+      "step": 4330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.209458015492872e-06,
+      "loss": 3.4806,
+      "step": 4331
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.207464328361078e-06,
+      "loss": 3.3759,
+      "step": 4332
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.205470608186779e-06,
+      "loss": 3.408,
+      "step": 4333
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.203476855287507e-06,
+      "loss": 3.4911,
+      "step": 4334
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.201483069980805e-06,
+      "loss": 3.4519,
+      "step": 4335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.199489252584221e-06,
+      "loss": 3.4959,
+      "step": 4336
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.197495403415308e-06,
+      "loss": 3.407,
+      "step": 4337
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.195501522791619e-06,
+      "loss": 3.3699,
+      "step": 4338
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.19350761103072e-06,
+      "loss": 3.4382,
+      "step": 4339
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.191513668450178e-06,
+      "loss": 3.4808,
+      "step": 4340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.189519695367562e-06,
+      "loss": 3.3636,
+      "step": 4341
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.187525692100452e-06,
+      "loss": 3.4799,
+      "step": 4342
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.185531658966428e-06,
+      "loss": 3.5002,
+      "step": 4343
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.183537596283075e-06,
+      "loss": 3.4592,
+      "step": 4344
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.181543504367988e-06,
+      "loss": 3.4336,
+      "step": 4345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.17954938353876e-06,
+      "loss": 3.4141,
+      "step": 4346
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.177555234112993e-06,
+      "loss": 3.5024,
+      "step": 4347
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.175561056408291e-06,
+      "loss": 3.3509,
+      "step": 4348
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.173566850742263e-06,
+      "loss": 3.5221,
+      "step": 4349
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.171572617432523e-06,
+      "loss": 3.4031,
+      "step": 4350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.169578356796691e-06,
+      "loss": 3.463,
+      "step": 4351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.167584069152388e-06,
+      "loss": 3.358,
+      "step": 4352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.16558975481724e-06,
+      "loss": 3.3638,
+      "step": 4353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1635954141088815e-06,
+      "loss": 3.4883,
+      "step": 4354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1616010473449465e-06,
+      "loss": 3.4425,
+      "step": 4355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1596066548430725e-06,
+      "loss": 3.3836,
+      "step": 4356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.157612236920905e-06,
+      "loss": 3.3471,
+      "step": 4357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.155617793896091e-06,
+      "loss": 3.3799,
+      "step": 4358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.153623326086283e-06,
+      "loss": 3.3782,
+      "step": 4359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.151628833809138e-06,
+      "loss": 3.468,
+      "step": 4360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.149634317382313e-06,
+      "loss": 3.3838,
+      "step": 4361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.147639777123472e-06,
+      "loss": 3.4175,
+      "step": 4362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.145645213350284e-06,
+      "loss": 3.4314,
+      "step": 4363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.143650626380417e-06,
+      "loss": 3.4304,
+      "step": 4364
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.141656016531546e-06,
+      "loss": 3.36,
+      "step": 4365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.139661384121352e-06,
+      "loss": 3.4135,
+      "step": 4366
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1376667294675155e-06,
+      "loss": 3.3594,
+      "step": 4367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1356720528877215e-06,
+      "loss": 3.3154,
+      "step": 4368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.133677354699658e-06,
+      "loss": 3.4119,
+      "step": 4369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.131682635221019e-06,
+      "loss": 3.3305,
+      "step": 4370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1296878947694995e-06,
+      "loss": 3.3582,
+      "step": 4371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.127693133662801e-06,
+      "loss": 3.4859,
+      "step": 4372
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.125698352218621e-06,
+      "loss": 3.3434,
+      "step": 4373
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.123703550754669e-06,
+      "loss": 3.5017,
+      "step": 4374
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.121708729588653e-06,
+      "loss": 3.4266,
+      "step": 4375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1197138890382835e-06,
+      "loss": 3.5174,
+      "step": 4376
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.117719029421277e-06,
+      "loss": 3.4104,
+      "step": 4377
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.115724151055351e-06,
+      "loss": 3.5222,
+      "step": 4378
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.113729254258228e-06,
+      "loss": 3.5004,
+      "step": 4379
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.111734339347629e-06,
+      "loss": 3.3437,
+      "step": 4380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.109739406641283e-06,
+      "loss": 3.3665,
+      "step": 4381
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.107744456456919e-06,
+      "loss": 3.3072,
+      "step": 4382
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1057494891122695e-06,
+      "loss": 3.5107,
+      "step": 4383
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.103754504925071e-06,
+      "loss": 3.391,
+      "step": 4384
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.10175950421306e-06,
+      "loss": 3.4755,
+      "step": 4385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.099764487293977e-06,
+      "loss": 3.3844,
+      "step": 4386
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.097769454485566e-06,
+      "loss": 3.5357,
+      "step": 4387
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.095774406105572e-06,
+      "loss": 3.4681,
+      "step": 4388
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.0937793424717415e-06,
+      "loss": 3.3688,
+      "step": 4389
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.09178426390183e-06,
+      "loss": 3.4843,
+      "step": 4390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.089789170713585e-06,
+      "loss": 3.4503,
+      "step": 4391
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.087794063224764e-06,
+      "loss": 3.4463,
+      "step": 4392
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.085798941753127e-06,
+      "loss": 3.4543,
+      "step": 4393
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.083803806616428e-06,
+      "loss": 3.5738,
+      "step": 4394
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.0818086581324345e-06,
+      "loss": 3.486,
+      "step": 4395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.079813496618908e-06,
+      "loss": 3.4195,
+      "step": 4396
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0778183223936154e-06,
+      "loss": 3.3791,
+      "step": 4397
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.075823135774322e-06,
+      "loss": 3.3862,
+      "step": 4398
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.073827937078805e-06,
+      "loss": 3.4927,
+      "step": 4399
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.071832726624828e-06,
+      "loss": 3.3603,
+      "step": 4400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.069837504730169e-06,
+      "loss": 3.5192,
+      "step": 4401
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.067842271712606e-06,
+      "loss": 3.4527,
+      "step": 4402
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.065847027889913e-06,
+      "loss": 3.4044,
+      "step": 4403
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.06385177357987e-06,
+      "loss": 3.4837,
+      "step": 4404
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.061856509100259e-06,
+      "loss": 3.4521,
+      "step": 4405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0598612347688605e-06,
+      "loss": 3.4172,
+      "step": 4406
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.05786595090346e-06,
+      "loss": 3.3319,
+      "step": 4407
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0558706578218455e-06,
+      "loss": 3.4125,
+      "step": 4408
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0538753558418e-06,
+      "loss": 3.4322,
+      "step": 4409
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0518800452811135e-06,
+      "loss": 3.3594,
+      "step": 4410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.049884726457577e-06,
+      "loss": 3.4987,
+      "step": 4411
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.04788939968898e-06,
+      "loss": 3.4081,
+      "step": 4412
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.045894065293117e-06,
+      "loss": 3.4916,
+      "step": 4413
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.043898723587779e-06,
+      "loss": 3.3038,
+      "step": 4414
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.041903374890763e-06,
+      "loss": 3.3411,
+      "step": 4415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.039908019519863e-06,
+      "loss": 3.4748,
+      "step": 4416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.037912657792881e-06,
+      "loss": 3.3686,
+      "step": 4417
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.035917290027607e-06,
+      "loss": 3.4379,
+      "step": 4418
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.033921916541845e-06,
+      "loss": 3.4846,
+      "step": 4419
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.031926537653396e-06,
+      "loss": 3.5042,
+      "step": 4420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.029931153680059e-06,
+      "loss": 3.4534,
+      "step": 4421
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.027935764939634e-06,
+      "loss": 3.3994,
+      "step": 4422
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.025940371749928e-06,
+      "loss": 3.4868,
+      "step": 4423
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.023944974428739e-06,
+      "loss": 3.448,
+      "step": 4424
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.021949573293871e-06,
+      "loss": 3.4425,
+      "step": 4425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.019954168663133e-06,
+      "loss": 3.4716,
+      "step": 4426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.017958760854325e-06,
+      "loss": 3.4422,
+      "step": 4427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.015963350185253e-06,
+      "loss": 3.4288,
+      "step": 4428
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.013967936973727e-06,
+      "loss": 3.4392,
+      "step": 4429
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.011972521537547e-06,
+      "loss": 3.4453,
+      "step": 4430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.009977104194525e-06,
+      "loss": 3.4247,
+      "step": 4431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.007981685262466e-06,
+      "loss": 3.4468,
+      "step": 4432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.005986265059176e-06,
+      "loss": 3.3996,
+      "step": 4433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.003990843902463e-06,
+      "loss": 3.4375,
+      "step": 4434
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0019954221101366e-06,
+      "loss": 3.416,
+      "step": 4435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5e-06,
+      "loss": 3.4278,
+      "step": 4436
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.998004577889865e-06,
+      "loss": 3.4613,
+      "step": 4437
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.996009156097538e-06,
+      "loss": 3.3692,
+      "step": 4438
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.994013734940826e-06,
+      "loss": 3.4037,
+      "step": 4439
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.992018314737535e-06,
+      "loss": 3.4265,
+      "step": 4440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.990022895805477e-06,
+      "loss": 3.4838,
+      "step": 4441
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.988027478462454e-06,
+      "loss": 3.4952,
+      "step": 4442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.986032063026276e-06,
+      "loss": 3.3381,
+      "step": 4443
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9840366498147495e-06,
+      "loss": 3.3634,
+      "step": 4444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.982041239145679e-06,
+      "loss": 3.3705,
+      "step": 4445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.980045831336871e-06,
+      "loss": 3.4432,
+      "step": 4446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.97805042670613e-06,
+      "loss": 3.3873,
+      "step": 4447
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.976055025571265e-06,
+      "loss": 3.3134,
+      "step": 4448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.974059628250074e-06,
+      "loss": 3.4367,
+      "step": 4449
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.972064235060366e-06,
+      "loss": 3.3859,
+      "step": 4450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.970068846319941e-06,
+      "loss": 3.4128,
+      "step": 4451
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.968073462346605e-06,
+      "loss": 3.4834,
+      "step": 4452
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.966078083458154e-06,
+      "loss": 3.3183,
+      "step": 4453
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.964082709972394e-06,
+      "loss": 3.3224,
+      "step": 4454
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.962087342207121e-06,
+      "loss": 3.3636,
+      "step": 4455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9600919804801375e-06,
+      "loss": 3.421,
+      "step": 4456
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.958096625109238e-06,
+      "loss": 3.426,
+      "step": 4457
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.956101276412222e-06,
+      "loss": 3.4553,
+      "step": 4458
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.954105934706886e-06,
+      "loss": 3.4096,
+      "step": 4459
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.952110600311021e-06,
+      "loss": 3.4989,
+      "step": 4460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.950115273542425e-06,
+      "loss": 3.3844,
+      "step": 4461
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.948119954718888e-06,
+      "loss": 3.393,
+      "step": 4462
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.946124644158202e-06,
+      "loss": 3.387,
+      "step": 4463
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.944129342178156e-06,
+      "loss": 3.3187,
+      "step": 4464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.942134049096541e-06,
+      "loss": 3.4094,
+      "step": 4465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.940138765231141e-06,
+      "loss": 3.4875,
+      "step": 4466
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.938143490899743e-06,
+      "loss": 3.3712,
+      "step": 4467
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.936148226420133e-06,
+      "loss": 3.3679,
+      "step": 4468
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.934152972110089e-06,
+      "loss": 3.3728,
+      "step": 4469
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9321577282873965e-06,
+      "loss": 3.4841,
+      "step": 4470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.930162495269833e-06,
+      "loss": 3.4468,
+      "step": 4471
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9281672733751745e-06,
+      "loss": 3.4265,
+      "step": 4472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9261720629211975e-06,
+      "loss": 3.5342,
+      "step": 4473
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.924176864225678e-06,
+      "loss": 3.4957,
+      "step": 4474
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.922181677606385e-06,
+      "loss": 3.4733,
+      "step": 4475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9201865033810934e-06,
+      "loss": 3.4149,
+      "step": 4476
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.918191341867566e-06,
+      "loss": 3.4165,
+      "step": 4477
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.916196193383573e-06,
+      "loss": 3.4752,
+      "step": 4478
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.914201058246876e-06,
+      "loss": 3.4088,
+      "step": 4479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.912205936775236e-06,
+      "loss": 3.4464,
+      "step": 4480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.910210829286416e-06,
+      "loss": 3.4717,
+      "step": 4481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.908215736098171e-06,
+      "loss": 3.406,
+      "step": 4482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.906220657528259e-06,
+      "loss": 3.4694,
+      "step": 4483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.90422559389443e-06,
+      "loss": 3.3764,
+      "step": 4484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.902230545514436e-06,
+      "loss": 3.4151,
+      "step": 4485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.900235512706025e-06,
+      "loss": 3.3961,
+      "step": 4486
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.898240495786942e-06,
+      "loss": 3.4905,
+      "step": 4487
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8962454950749304e-06,
+      "loss": 3.4028,
+      "step": 4488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.894250510887732e-06,
+      "loss": 3.5122,
+      "step": 4489
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.892255543543083e-06,
+      "loss": 3.4399,
+      "step": 4490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8902605933587185e-06,
+      "loss": 3.4323,
+      "step": 4491
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8882656606523736e-06,
+      "loss": 3.5002,
+      "step": 4492
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.886270745741775e-06,
+      "loss": 3.4386,
+      "step": 4493
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.88427584894465e-06,
+      "loss": 3.4025,
+      "step": 4494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.882280970578725e-06,
+      "loss": 3.38,
+      "step": 4495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.880286110961718e-06,
+      "loss": 3.3445,
+      "step": 4496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.878291270411349e-06,
+      "loss": 3.2507,
+      "step": 4497
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.876296449245332e-06,
+      "loss": 3.405,
+      "step": 4498
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.874301647781378e-06,
+      "loss": 3.4541,
+      "step": 4499
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8723068663372005e-06,
+      "loss": 3.4257,
+      "step": 4500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8703121052305005e-06,
+      "loss": 3.4351,
+      "step": 4501
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.868317364778981e-06,
+      "loss": 3.4287,
+      "step": 4502
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.866322645300343e-06,
+      "loss": 3.4065,
+      "step": 4503
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.864327947112281e-06,
+      "loss": 3.4447,
+      "step": 4504
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.862333270532486e-06,
+      "loss": 3.4803,
+      "step": 4505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.86033861587865e-06,
+      "loss": 3.3425,
+      "step": 4506
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8583439834684555e-06,
+      "loss": 3.5079,
+      "step": 4507
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.856349373619586e-06,
+      "loss": 3.5021,
+      "step": 4508
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.854354786649718e-06,
+      "loss": 3.3643,
+      "step": 4509
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.85236022287653e-06,
+      "loss": 3.3784,
+      "step": 4510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.850365682617689e-06,
+      "loss": 3.4066,
+      "step": 4511
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8483711661908635e-06,
+      "loss": 3.3195,
+      "step": 4512
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8463766739137184e-06,
+      "loss": 3.4236,
+      "step": 4513
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.84438220610391e-06,
+      "loss": 3.4287,
+      "step": 4514
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.842387763079097e-06,
+      "loss": 3.3383,
+      "step": 4515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.84039334515693e-06,
+      "loss": 3.4078,
+      "step": 4516
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.838398952655057e-06,
+      "loss": 3.4576,
+      "step": 4517
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.83640458589112e-06,
+      "loss": 3.4336,
+      "step": 4518
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.834410245182762e-06,
+      "loss": 3.3819,
+      "step": 4519
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8324159308476145e-06,
+      "loss": 3.3743,
+      "step": 4520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.830421643203311e-06,
+      "loss": 3.4156,
+      "step": 4521
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.82842738256748e-06,
+      "loss": 3.4618,
+      "step": 4522
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.826433149257738e-06,
+      "loss": 3.4327,
+      "step": 4523
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.82443894359171e-06,
+      "loss": 3.51,
+      "step": 4524
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.822444765887008e-06,
+      "loss": 3.3547,
+      "step": 4525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.820450616461239e-06,
+      "loss": 3.3528,
+      "step": 4526
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.818456495632013e-06,
+      "loss": 3.2869,
+      "step": 4527
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.816462403716925e-06,
+      "loss": 3.5932,
+      "step": 4528
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8144683410335735e-06,
+      "loss": 3.3765,
+      "step": 4529
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.81247430789955e-06,
+      "loss": 3.474,
+      "step": 4530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.810480304632439e-06,
+      "loss": 3.4749,
+      "step": 4531
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.808486331549824e-06,
+      "loss": 3.3722,
+      "step": 4532
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.806492388969281e-06,
+      "loss": 3.2986,
+      "step": 4533
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.804498477208382e-06,
+      "loss": 3.3752,
+      "step": 4534
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.802504596584694e-06,
+      "loss": 3.3851,
+      "step": 4535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.800510747415781e-06,
+      "loss": 3.4967,
+      "step": 4536
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7985169300191956e-06,
+      "loss": 3.41,
+      "step": 4537
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.796523144712494e-06,
+      "loss": 3.4385,
+      "step": 4538
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.794529391813223e-06,
+      "loss": 3.3717,
+      "step": 4539
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7925356716389235e-06,
+      "loss": 3.4689,
+      "step": 4540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7905419845071304e-06,
+      "loss": 3.3243,
+      "step": 4541
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.788548330735377e-06,
+      "loss": 3.4942,
+      "step": 4542
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.786554710641191e-06,
+      "loss": 3.4321,
+      "step": 4543
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.784561124542088e-06,
+      "loss": 3.4512,
+      "step": 4544
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.782567572755587e-06,
+      "loss": 3.3745,
+      "step": 4545
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.780574055599198e-06,
+      "loss": 3.2913,
+      "step": 4546
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.778580573390421e-06,
+      "loss": 3.4356,
+      "step": 4547
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.776587126446761e-06,
+      "loss": 3.4657,
+      "step": 4548
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.774593715085707e-06,
+      "loss": 3.4787,
+      "step": 4549
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.772600339624748e-06,
+      "loss": 3.5169,
+      "step": 4550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.770607000381364e-06,
+      "loss": 3.5563,
+      "step": 4551
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.768613697673033e-06,
+      "loss": 3.4917,
+      "step": 4552
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.766620431817224e-06,
+      "loss": 3.4435,
+      "step": 4553
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.764627203131401e-06,
+      "loss": 3.4375,
+      "step": 4554
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.762634011933023e-06,
+      "loss": 3.4775,
+      "step": 4555
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.760640858539541e-06,
+      "loss": 3.5066,
+      "step": 4556
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.758647743268403e-06,
+      "loss": 3.3579,
+      "step": 4557
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.756654666437048e-06,
+      "loss": 3.3604,
+      "step": 4558
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.75466162836291e-06,
+      "loss": 3.3312,
+      "step": 4559
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.752668629363419e-06,
+      "loss": 3.3992,
+      "step": 4560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7506756697559945e-06,
+      "loss": 3.4416,
+      "step": 4561
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.748682749858053e-06,
+      "loss": 3.4366,
+      "step": 4562
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.746689869987004e-06,
+      "loss": 3.5035,
+      "step": 4563
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.744697030460248e-06,
+      "loss": 3.4317,
+      "step": 4564
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.742704231595184e-06,
+      "loss": 3.3399,
+      "step": 4565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.740711473709199e-06,
+      "loss": 3.4216,
+      "step": 4566
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.738718757119682e-06,
+      "loss": 3.3709,
+      "step": 4567
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.736726082144003e-06,
+      "loss": 3.3966,
+      "step": 4568
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.734733449099534e-06,
+      "loss": 3.4206,
+      "step": 4569
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.732740858303642e-06,
+      "loss": 3.4018,
+      "step": 4570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.73074831007368e-06,
+      "loss": 3.4729,
+      "step": 4571
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.728755804726998e-06,
+      "loss": 3.4431,
+      "step": 4572
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7267633425809405e-06,
+      "loss": 3.5929,
+      "step": 4573
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.724770923952844e-06,
+      "loss": 3.4382,
+      "step": 4574
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.722778549160038e-06,
+      "loss": 3.4161,
+      "step": 4575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.720786218519843e-06,
+      "loss": 3.5259,
+      "step": 4576
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.718793932349576e-06,
+      "loss": 3.4897,
+      "step": 4577
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7168016909665445e-06,
+      "loss": 3.3952,
+      "step": 4578
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.714809494688049e-06,
+      "loss": 3.4634,
+      "step": 4579
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.712817343831384e-06,
+      "loss": 3.4611,
+      "step": 4580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.710825238713837e-06,
+      "loss": 3.4232,
+      "step": 4581
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7088331796526845e-06,
+      "loss": 3.3579,
+      "step": 4582
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.706841166965202e-06,
+      "loss": 3.38,
+      "step": 4583
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7048492009686525e-06,
+      "loss": 3.4093,
+      "step": 4584
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.702857281980292e-06,
+      "loss": 3.4081,
+      "step": 4585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.700865410317373e-06,
+      "loss": 3.5347,
+      "step": 4586
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6988735862971355e-06,
+      "loss": 3.3549,
+      "step": 4587
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.696881810236815e-06,
+      "loss": 3.4742,
+      "step": 4588
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.694890082453638e-06,
+      "loss": 3.4108,
+      "step": 4589
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.692898403264825e-06,
+      "loss": 3.3645,
+      "step": 4590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.690906772987584e-06,
+      "loss": 3.4465,
+      "step": 4591
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.688915191939123e-06,
+      "loss": 3.4292,
+      "step": 4592
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.686923660436635e-06,
+      "loss": 3.4404,
+      "step": 4593
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6849321787973116e-06,
+      "loss": 3.3871,
+      "step": 4594
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.682940747338328e-06,
+      "loss": 3.4685,
+      "step": 4595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.680949366376858e-06,
+      "loss": 3.3921,
+      "step": 4596
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.678958036230067e-06,
+      "loss": 3.4725,
+      "step": 4597
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.676966757215112e-06,
+      "loss": 3.5213,
+      "step": 4598
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.674975529649137e-06,
+      "loss": 3.4282,
+      "step": 4599
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.672984353849285e-06,
+      "loss": 3.4032,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.670993230132685e-06,
+      "loss": 3.4101,
+      "step": 4601
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.669002158816462e-06,
+      "loss": 3.4924,
+      "step": 4602
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.667011140217729e-06,
+      "loss": 3.3903,
+      "step": 4603
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6650201746535926e-06,
+      "loss": 3.4092,
+      "step": 4604
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.663029262441152e-06,
+      "loss": 3.443,
+      "step": 4605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6610384038974945e-06,
+      "loss": 3.4165,
+      "step": 4606
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.659047599339703e-06,
+      "loss": 3.4388,
+      "step": 4607
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.657056849084849e-06,
+      "loss": 3.5019,
+      "step": 4608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.655066153449995e-06,
+      "loss": 3.4191,
+      "step": 4609
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.653075512752196e-06,
+      "loss": 3.3858,
+      "step": 4610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6510849273085016e-06,
+      "loss": 3.43,
+      "step": 4611
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.649094397435944e-06,
+      "loss": 3.4603,
+      "step": 4612
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.647103923451555e-06,
+      "loss": 3.3195,
+      "step": 4613
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.645113505672353e-06,
+      "loss": 3.4894,
+      "step": 4614
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.643123144415348e-06,
+      "loss": 3.3876,
+      "step": 4615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.641132839997544e-06,
+      "loss": 3.2833,
+      "step": 4616
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6391425927359316e-06,
+      "loss": 3.5317,
+      "step": 4617
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.637152402947494e-06,
+      "loss": 3.4137,
+      "step": 4618
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.635162270949206e-06,
+      "loss": 3.4497,
+      "step": 4619
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.633172197058034e-06,
+      "loss": 3.3837,
+      "step": 4620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6311821815909294e-06,
+      "loss": 3.3988,
+      "step": 4621
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6291922248648455e-06,
+      "loss": 3.3702,
+      "step": 4622
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.627202327196715e-06,
+      "loss": 3.4502,
+      "step": 4623
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.625212488903467e-06,
+      "loss": 3.3634,
+      "step": 4624
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.62322271030202e-06,
+      "loss": 3.3405,
+      "step": 4625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.621232991709282e-06,
+      "loss": 3.5264,
+      "step": 4626
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.619243333442153e-06,
+      "loss": 3.3762,
+      "step": 4627
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.617253735817522e-06,
+      "loss": 3.4158,
+      "step": 4628
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.615264199152269e-06,
+      "loss": 3.4861,
+      "step": 4629
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.613274723763265e-06,
+      "loss": 3.4035,
+      "step": 4630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.61128530996737e-06,
+      "loss": 3.3885,
+      "step": 4631
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6092959580814355e-06,
+      "loss": 3.4324,
+      "step": 4632
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.607306668422301e-06,
+      "loss": 3.415,
+      "step": 4633
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.605317441306798e-06,
+      "loss": 3.4576,
+      "step": 4634
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.603328277051748e-06,
+      "loss": 3.396,
+      "step": 4635
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6013391759739615e-06,
+      "loss": 3.3622,
+      "step": 4636
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.59935013839024e-06,
+      "loss": 3.4613,
+      "step": 4637
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.597361164617373e-06,
+      "loss": 3.3478,
+      "step": 4638
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.595372254972143e-06,
+      "loss": 3.4757,
+      "step": 4639
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.593383409771318e-06,
+      "loss": 3.3721,
+      "step": 4640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5913946293316605e-06,
+      "loss": 3.3617,
+      "step": 4641
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.589405913969919e-06,
+      "loss": 3.4603,
+      "step": 4642
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.587417264002833e-06,
+      "loss": 3.3112,
+      "step": 4643
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.585428679747133e-06,
+      "loss": 3.4842,
+      "step": 4644
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.583440161519535e-06,
+      "loss": 3.3768,
+      "step": 4645
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.581451709636748e-06,
+      "loss": 3.4598,
+      "step": 4646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.579463324415472e-06,
+      "loss": 3.385,
+      "step": 4647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.57747500617239e-06,
+      "loss": 3.3903,
+      "step": 4648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5754867552241825e-06,
+      "loss": 3.4772,
+      "step": 4649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.573498571887511e-06,
+      "loss": 3.342,
+      "step": 4650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.571510456479033e-06,
+      "loss": 3.3872,
+      "step": 4651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.569522409315392e-06,
+      "loss": 3.5712,
+      "step": 4652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.56753443071322e-06,
+      "loss": 3.3844,
+      "step": 4653
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.565546520989139e-06,
+      "loss": 3.2613,
+      "step": 4654
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.56355868045976e-06,
+      "loss": 3.4753,
+      "step": 4655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.561570909441686e-06,
+      "loss": 3.4934,
+      "step": 4656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.559583208251502e-06,
+      "loss": 3.37,
+      "step": 4657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.557595577205789e-06,
+      "loss": 3.4836,
+      "step": 4658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5556080166211135e-06,
+      "loss": 3.3576,
+      "step": 4659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.553620526814029e-06,
+      "loss": 3.4918,
+      "step": 4660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5516331081010815e-06,
+      "loss": 3.399,
+      "step": 4661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.549645760798804e-06,
+      "loss": 3.4612,
+      "step": 4662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.547658485223716e-06,
+      "loss": 3.4201,
+      "step": 4663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.545671281692331e-06,
+      "loss": 3.4539,
+      "step": 4664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.543684150521146e-06,
+      "loss": 3.4579,
+      "step": 4665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.541697092026647e-06,
+      "loss": 3.472,
+      "step": 4666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.539710106525311e-06,
+      "loss": 3.3227,
+      "step": 4667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.537723194333603e-06,
+      "loss": 3.4593,
+      "step": 4668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.535736355767972e-06,
+      "loss": 3.5104,
+      "step": 4669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.533749591144861e-06,
+      "loss": 3.2692,
+      "step": 4670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.531762900780698e-06,
+      "loss": 3.425,
+      "step": 4671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5297762849919e-06,
+      "loss": 3.383,
+      "step": 4672
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.527789744094875e-06,
+      "loss": 3.3192,
+      "step": 4673
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.525803278406011e-06,
+      "loss": 3.5202,
+      "step": 4674
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5238168882416924e-06,
+      "loss": 3.4349,
+      "step": 4675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.521830573918289e-06,
+      "loss": 3.3876,
+      "step": 4676
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.519844335752156e-06,
+      "loss": 3.2799,
+      "step": 4677
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5178581740596376e-06,
+      "loss": 3.4729,
+      "step": 4678
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.51587208915707e-06,
+      "loss": 3.3161,
+      "step": 4679
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.513886081360769e-06,
+      "loss": 3.3232,
+      "step": 4680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5119001509870465e-06,
+      "loss": 3.4271,
+      "step": 4681
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.509914298352197e-06,
+      "loss": 3.4572,
+      "step": 4682
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.507928523772506e-06,
+      "loss": 3.364,
+      "step": 4683
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.505942827564242e-06,
+      "loss": 3.4628,
+      "step": 4684
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.503957210043663e-06,
+      "loss": 3.5155,
+      "step": 4685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.501971671527019e-06,
+      "loss": 3.3878,
+      "step": 4686
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.49998621233054e-06,
+      "loss": 3.3176,
+      "step": 4687
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.498000832770448e-06,
+      "loss": 3.4771,
+      "step": 4688
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.496015533162952e-06,
+      "loss": 3.3644,
+      "step": 4689
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.494030313824246e-06,
+      "loss": 3.3765,
+      "step": 4690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.492045175070513e-06,
+      "loss": 3.5136,
+      "step": 4691
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.490060117217925e-06,
+      "loss": 3.4715,
+      "step": 4692
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.488075140582635e-06,
+      "loss": 3.3214,
+      "step": 4693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4860902454807905e-06,
+      "loss": 3.2889,
+      "step": 4694
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4841054322285195e-06,
+      "loss": 3.3596,
+      "step": 4695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4821207011419435e-06,
+      "loss": 3.4596,
+      "step": 4696
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.480136052537166e-06,
+      "loss": 3.5671,
+      "step": 4697
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.478151486730276e-06,
+      "loss": 3.3936,
+      "step": 4698
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4761670040373565e-06,
+      "loss": 3.5588,
+      "step": 4699
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.474182604774471e-06,
+      "loss": 3.4164,
+      "step": 4700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.472198289257671e-06,
+      "loss": 3.1753,
+      "step": 4701
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.470214057802995e-06,
+      "loss": 3.4374,
+      "step": 4702
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.468229910726471e-06,
+      "loss": 3.3686,
+      "step": 4703
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.466245848344106e-06,
+      "loss": 3.4132,
+      "step": 4704
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.464261870971902e-06,
+      "loss": 3.3942,
+      "step": 4705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.462277978925845e-06,
+      "loss": 3.3423,
+      "step": 4706
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.460294172521903e-06,
+      "loss": 3.3161,
+      "step": 4707
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.458310452076034e-06,
+      "loss": 3.4722,
+      "step": 4708
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.456326817904186e-06,
+      "loss": 3.5719,
+      "step": 4709
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.454343270322284e-06,
+      "loss": 3.4294,
+      "step": 4710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4523598096462465e-06,
+      "loss": 3.527,
+      "step": 4711
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.450376436191975e-06,
+      "loss": 3.5217,
+      "step": 4712
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.448393150275362e-06,
+      "loss": 3.5297,
+      "step": 4713
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.446409952212278e-06,
+      "loss": 3.355,
+      "step": 4714
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.444426842318586e-06,
+      "loss": 3.3911,
+      "step": 4715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.442443820910133e-06,
+      "loss": 3.4039,
+      "step": 4716
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.440460888302749e-06,
+      "loss": 3.5483,
+      "step": 4717
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.438478044812254e-06,
+      "loss": 3.458,
+      "step": 4718
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.436495290754452e-06,
+      "loss": 3.4892,
+      "step": 4719
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.434512626445134e-06,
+      "loss": 3.439,
+      "step": 4720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4325300522000765e-06,
+      "loss": 3.3878,
+      "step": 4721
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.430547568335039e-06,
+      "loss": 3.4415,
+      "step": 4722
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4285651751657676e-06,
+      "loss": 3.4322,
+      "step": 4723
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.426582873007999e-06,
+      "loss": 3.4291,
+      "step": 4724
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.424600662177446e-06,
+      "loss": 3.4488,
+      "step": 4725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.422618542989815e-06,
+      "loss": 3.4792,
+      "step": 4726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.420636515760797e-06,
+      "loss": 3.5041,
+      "step": 4727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.418654580806062e-06,
+      "loss": 3.4066,
+      "step": 4728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.416672738441271e-06,
+      "loss": 3.4809,
+      "step": 4729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4146909889820705e-06,
+      "loss": 3.413,
+      "step": 4730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4127093327440885e-06,
+      "loss": 3.3436,
+      "step": 4731
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.410727770042941e-06,
+      "loss": 3.4743,
+      "step": 4732
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.408746301194229e-06,
+      "loss": 3.2713,
+      "step": 4733
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.406764926513536e-06,
+      "loss": 3.4302,
+      "step": 4734
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.404783646316434e-06,
+      "loss": 3.3684,
+      "step": 4735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.402802460918478e-06,
+      "loss": 3.443,
+      "step": 4736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4008213706352075e-06,
+      "loss": 3.4169,
+      "step": 4737
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.398840375782148e-06,
+      "loss": 3.3988,
+      "step": 4738
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.396859476674809e-06,
+      "loss": 3.415,
+      "step": 4739
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3948786736286866e-06,
+      "loss": 3.4482,
+      "step": 4740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3928979669592575e-06,
+      "loss": 3.3195,
+      "step": 4741
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.390917356981988e-06,
+      "loss": 3.4134,
+      "step": 4742
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.388936844012326e-06,
+      "loss": 3.416,
+      "step": 4743
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.386956428365701e-06,
+      "loss": 3.3882,
+      "step": 4744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.384976110357538e-06,
+      "loss": 3.3432,
+      "step": 4745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.382995890303233e-06,
+      "loss": 3.4309,
+      "step": 4746
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.381015768518175e-06,
+      "loss": 3.468,
+      "step": 4747
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.379035745317734e-06,
+      "loss": 3.4576,
+      "step": 4748
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.377055821017266e-06,
+      "loss": 3.5366,
+      "step": 4749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.375075995932108e-06,
+      "loss": 3.4079,
+      "step": 4750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.373096270377588e-06,
+      "loss": 3.4471,
+      "step": 4751
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.371116644669008e-06,
+      "loss": 3.3818,
+      "step": 4752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3691371191216635e-06,
+      "loss": 3.4095,
+      "step": 4753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.36715769405083e-06,
+      "loss": 3.4631,
+      "step": 4754
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.365178369771766e-06,
+      "loss": 3.4131,
+      "step": 4755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.363199146599717e-06,
+      "loss": 3.476,
+      "step": 4756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.36122002484991e-06,
+      "loss": 3.4664,
+      "step": 4757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.359241004837555e-06,
+      "loss": 3.4107,
+      "step": 4758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3572620868778485e-06,
+      "loss": 3.3378,
+      "step": 4759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3552832712859715e-06,
+      "loss": 3.4677,
+      "step": 4760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.353304558377083e-06,
+      "loss": 3.4064,
+      "step": 4761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.351325948466331e-06,
+      "loss": 3.4686,
+      "step": 4762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.349347441868848e-06,
+      "loss": 3.4272,
+      "step": 4763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.347369038899744e-06,
+      "loss": 3.4128,
+      "step": 4764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.345390739874117e-06,
+      "loss": 3.4638,
+      "step": 4765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.343412545107047e-06,
+      "loss": 3.5085,
+      "step": 4766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.341434454913602e-06,
+      "loss": 3.2877,
+      "step": 4767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.339456469608823e-06,
+      "loss": 3.4037,
+      "step": 4768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3374785895077434e-06,
+      "loss": 3.364,
+      "step": 4769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.335500814925378e-06,
+      "loss": 3.438,
+      "step": 4770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.333523146176723e-06,
+      "loss": 3.4116,
+      "step": 4771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.331545583576758e-06,
+      "loss": 3.3952,
+      "step": 4772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.329568127440446e-06,
+      "loss": 3.4166,
+      "step": 4773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.327590778082734e-06,
+      "loss": 3.3677,
+      "step": 4774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3256135358185525e-06,
+      "loss": 3.4278,
+      "step": 4775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.323636400962811e-06,
+      "loss": 3.4443,
+      "step": 4776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.321659373830405e-06,
+      "loss": 3.4868,
+      "step": 4777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.319682454736215e-06,
+      "loss": 3.3559,
+      "step": 4778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.317705643995099e-06,
+      "loss": 3.4868,
+      "step": 4779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3157289419219e-06,
+      "loss": 3.4411,
+      "step": 4780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.313752348831448e-06,
+      "loss": 3.4066,
+      "step": 4781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.311775865038548e-06,
+      "loss": 3.3326,
+      "step": 4782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.309799490857991e-06,
+      "loss": 3.4664,
+      "step": 4783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.307823226604555e-06,
+      "loss": 3.5296,
+      "step": 4784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3058470725929925e-06,
+      "loss": 3.3982,
+      "step": 4785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.303871029138044e-06,
+      "loss": 3.3881,
+      "step": 4786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.301895096554432e-06,
+      "loss": 3.4758,
+      "step": 4787
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.299919275156857e-06,
+      "loss": 3.4016,
+      "step": 4788
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.297943565260006e-06,
+      "loss": 3.2423,
+      "step": 4789
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.295967967178549e-06,
+      "loss": 3.4676,
+      "step": 4790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.293992481227134e-06,
+      "loss": 3.4393,
+      "step": 4791
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.292017107720394e-06,
+      "loss": 3.3515,
+      "step": 4792
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2900418469729435e-06,
+      "loss": 3.3584,
+      "step": 4793
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.28806669929938e-06,
+      "loss": 3.464,
+      "step": 4794
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.286091665014283e-06,
+      "loss": 3.349,
+      "step": 4795
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.28411674443221e-06,
+      "loss": 3.4398,
+      "step": 4796
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.282141937867706e-06,
+      "loss": 3.3603,
+      "step": 4797
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.280167245635294e-06,
+      "loss": 3.4306,
+      "step": 4798
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.27819266804948e-06,
+      "loss": 3.3934,
+      "step": 4799
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.276218205424752e-06,
+      "loss": 3.3679,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.274243858075579e-06,
+      "loss": 3.3684,
+      "step": 4801
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2722696263164145e-06,
+      "loss": 3.4197,
+      "step": 4802
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.270295510461688e-06,
+      "loss": 3.3568,
+      "step": 4803
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2683215108258145e-06,
+      "loss": 3.2888,
+      "step": 4804
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.266347627723192e-06,
+      "loss": 3.3425,
+      "step": 4805
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.264373861468195e-06,
+      "loss": 3.4511,
+      "step": 4806
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.262400212375183e-06,
+      "loss": 3.4859,
+      "step": 4807
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.260426680758496e-06,
+      "loss": 3.4017,
+      "step": 4808
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.258453266932455e-06,
+      "loss": 3.4416,
+      "step": 4809
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.256479971211363e-06,
+      "loss": 3.3716,
+      "step": 4810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.254506793909503e-06,
+      "loss": 3.4376,
+      "step": 4811
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.25253373534114e-06,
+      "loss": 3.3668,
+      "step": 4812
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.250560795820519e-06,
+      "loss": 3.422,
+      "step": 4813
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2485879756618685e-06,
+      "loss": 3.3223,
+      "step": 4814
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2466152751793945e-06,
+      "loss": 3.4614,
+      "step": 4815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.244642694687288e-06,
+      "loss": 3.265,
+      "step": 4816
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2426702344997175e-06,
+      "loss": 3.3895,
+      "step": 4817
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.24069789493083e-06,
+      "loss": 3.3279,
+      "step": 4818
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.238725676294764e-06,
+      "loss": 3.4285,
+      "step": 4819
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.236753578905627e-06,
+      "loss": 3.4633,
+      "step": 4820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.234781603077514e-06,
+      "loss": 3.513,
+      "step": 4821
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.232809749124497e-06,
+      "loss": 3.3679,
+      "step": 4822
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.230838017360629e-06,
+      "loss": 3.4919,
+      "step": 4823
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.228866408099945e-06,
+      "loss": 3.4061,
+      "step": 4824
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.226894921656463e-06,
+      "loss": 3.3533,
+      "step": 4825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.224923558344175e-06,
+      "loss": 3.3063,
+      "step": 4826
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2229523184770575e-06,
+      "loss": 3.3805,
+      "step": 4827
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.220981202369067e-06,
+      "loss": 3.5582,
+      "step": 4828
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.219010210334141e-06,
+      "loss": 3.4169,
+      "step": 4829
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.217039342686195e-06,
+      "loss": 3.3626,
+      "step": 4830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.215068599739126e-06,
+      "loss": 3.3831,
+      "step": 4831
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.213097981806813e-06,
+      "loss": 3.2539,
+      "step": 4832
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2111274892031095e-06,
+      "loss": 3.3902,
+      "step": 4833
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2091571222418546e-06,
+      "loss": 3.3785,
+      "step": 4834
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.207186881236866e-06,
+      "loss": 3.426,
+      "step": 4835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.205216766501941e-06,
+      "loss": 3.3981,
+      "step": 4836
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2032467783508545e-06,
+      "loss": 3.4258,
+      "step": 4837
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.201276917097366e-06,
+      "loss": 3.3396,
+      "step": 4838
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1993071830552105e-06,
+      "loss": 3.2931,
+      "step": 4839
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.197337576538103e-06,
+      "loss": 3.3363,
+      "step": 4840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.195368097859743e-06,
+      "loss": 3.3972,
+      "step": 4841
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1933987473338015e-06,
+      "loss": 3.354,
+      "step": 4842
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.191429525273939e-06,
+      "loss": 3.4606,
+      "step": 4843
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.189460431993788e-06,
+      "loss": 3.4345,
+      "step": 4844
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.187491467806962e-06,
+      "loss": 3.4476,
+      "step": 4845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.185522633027057e-06,
+      "loss": 3.3945,
+      "step": 4846
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.183553927967643e-06,
+      "loss": 3.3428,
+      "step": 4847
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.181585352942274e-06,
+      "loss": 3.4809,
+      "step": 4848
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.179616908264484e-06,
+      "loss": 3.3309,
+      "step": 4849
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1776485942477795e-06,
+      "loss": 3.4965,
+      "step": 4850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1756804112056535e-06,
+      "loss": 3.4346,
+      "step": 4851
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.173712359451576e-06,
+      "loss": 3.3619,
+      "step": 4852
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1717444392989935e-06,
+      "loss": 3.3458,
+      "step": 4853
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.169776651061334e-06,
+      "loss": 3.4352,
+      "step": 4854
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.167808995052005e-06,
+      "loss": 3.3306,
+      "step": 4855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.16584147158439e-06,
+      "loss": 3.4469,
+      "step": 4856
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.163874080971855e-06,
+      "loss": 3.4605,
+      "step": 4857
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.161906823527742e-06,
+      "loss": 3.4697,
+      "step": 4858
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.159939699565375e-06,
+      "loss": 3.4353,
+      "step": 4859
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.157972709398051e-06,
+      "loss": 3.4614,
+      "step": 4860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.156005853339052e-06,
+      "loss": 3.3383,
+      "step": 4861
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.154039131701636e-06,
+      "loss": 3.4027,
+      "step": 4862
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.152072544799038e-06,
+      "loss": 3.4647,
+      "step": 4863
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.150106092944475e-06,
+      "loss": 3.4605,
+      "step": 4864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1481397764511395e-06,
+      "loss": 3.3343,
+      "step": 4865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.146173595632203e-06,
+      "loss": 3.4296,
+      "step": 4866
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.144207550800816e-06,
+      "loss": 3.3849,
+      "step": 4867
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.142241642270109e-06,
+      "loss": 3.5282,
+      "step": 4868
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.140275870353187e-06,
+      "loss": 3.529,
+      "step": 4869
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.138310235363138e-06,
+      "loss": 3.503,
+      "step": 4870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.136344737613023e-06,
+      "loss": 3.5368,
+      "step": 4871
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1343793774158845e-06,
+      "loss": 3.4798,
+      "step": 4872
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1324141550847425e-06,
+      "loss": 3.4004,
+      "step": 4873
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.130449070932594e-06,
+      "loss": 3.4517,
+      "step": 4874
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.128484125272415e-06,
+      "loss": 3.2671,
+      "step": 4875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1265193184171605e-06,
+      "loss": 3.3459,
+      "step": 4876
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.124554650679761e-06,
+      "loss": 3.4172,
+      "step": 4877
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1225901223731245e-06,
+      "loss": 3.4026,
+      "step": 4878
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.120625733810142e-06,
+      "loss": 3.4088,
+      "step": 4879
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.118661485303674e-06,
+      "loss": 3.5038,
+      "step": 4880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1166973771665665e-06,
+      "loss": 3.4423,
+      "step": 4881
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.11473340971164e-06,
+      "loss": 3.4068,
+      "step": 4882
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.112769583251689e-06,
+      "loss": 3.4487,
+      "step": 4883
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.110805898099492e-06,
+      "loss": 3.492,
+      "step": 4884
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.108842354567801e-06,
+      "loss": 3.5523,
+      "step": 4885
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.106878952969348e-06,
+      "loss": 3.386,
+      "step": 4886
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.104915693616838e-06,
+      "loss": 3.4314,
+      "step": 4887
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.102952576822957e-06,
+      "loss": 3.4637,
+      "step": 4888
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.100989602900369e-06,
+      "loss": 3.4113,
+      "step": 4889
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0990267721617115e-06,
+      "loss": 3.3697,
+      "step": 4890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.097064084919603e-06,
+      "loss": 3.4505,
+      "step": 4891
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.095101541486636e-06,
+      "loss": 3.3987,
+      "step": 4892
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.093139142175382e-06,
+      "loss": 3.4869,
+      "step": 4893
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.091176887298392e-06,
+      "loss": 3.5227,
+      "step": 4894
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.089214777168188e-06,
+      "loss": 3.5104,
+      "step": 4895
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0872528120972734e-06,
+      "loss": 3.4377,
+      "step": 4896
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.085290992398127e-06,
+      "loss": 3.3186,
+      "step": 4897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.083329318383204e-06,
+      "loss": 3.39,
+      "step": 4898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.081367790364937e-06,
+      "loss": 3.4015,
+      "step": 4899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.079406408655737e-06,
+      "loss": 3.4048,
+      "step": 4900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0774451735679885e-06,
+      "loss": 3.3363,
+      "step": 4901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.075484085414054e-06,
+      "loss": 3.3831,
+      "step": 4902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.073523144506276e-06,
+      "loss": 3.3634,
+      "step": 4903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.071562351156966e-06,
+      "loss": 3.4591,
+      "step": 4904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.069601705678418e-06,
+      "loss": 3.5141,
+      "step": 4905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.067641208382902e-06,
+      "loss": 3.3919,
+      "step": 4906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.065680859582662e-06,
+      "loss": 3.4169,
+      "step": 4907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0637206595899206e-06,
+      "loss": 3.3693,
+      "step": 4908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0617606087168745e-06,
+      "loss": 3.4957,
+      "step": 4909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.059800707275698e-06,
+      "loss": 3.4557,
+      "step": 4910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.057840955578543e-06,
+      "loss": 3.4458,
+      "step": 4911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.055881353937533e-06,
+      "loss": 3.5025,
+      "step": 4912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0539219026647755e-06,
+      "loss": 3.3429,
+      "step": 4913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.051962602072343e-06,
+      "loss": 3.4069,
+      "step": 4914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.050003452472295e-06,
+      "loss": 3.4489,
+      "step": 4915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.048044454176658e-06,
+      "loss": 3.3779,
+      "step": 4916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.046085607497443e-06,
+      "loss": 3.4329,
+      "step": 4917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.044126912746629e-06,
+      "loss": 3.5814,
+      "step": 4918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.042168370236176e-06,
+      "loss": 3.4581,
+      "step": 4919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.040209980278016e-06,
+      "loss": 3.4125,
+      "step": 4920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.038251743184061e-06,
+      "loss": 3.4229,
+      "step": 4921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.036293659266195e-06,
+      "loss": 3.427,
+      "step": 4922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.034335728836278e-06,
+      "loss": 3.4418,
+      "step": 4923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.032377952206148e-06,
+      "loss": 3.4168,
+      "step": 4924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.030420329687616e-06,
+      "loss": 3.5609,
+      "step": 4925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.028462861592469e-06,
+      "loss": 3.1923,
+      "step": 4926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.026505548232471e-06,
+      "loss": 3.492,
+      "step": 4927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02454838991936e-06,
+      "loss": 3.4742,
+      "step": 4928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.022591386964847e-06,
+      "loss": 3.4221,
+      "step": 4929
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.020634539680624e-06,
+      "loss": 3.4621,
+      "step": 4930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.018677848378354e-06,
+      "loss": 3.2982,
+      "step": 4931
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.016721313369674e-06,
+      "loss": 3.4222,
+      "step": 4932
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.014764934966201e-06,
+      "loss": 3.3198,
+      "step": 4933
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.012808713479523e-06,
+      "loss": 3.5416,
+      "step": 4934
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.010852649221202e-06,
+      "loss": 3.4701,
+      "step": 4935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.008896742502782e-06,
+      "loss": 3.4012,
+      "step": 4936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.006940993635773e-06,
+      "loss": 3.4909,
+      "step": 4937
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.004985402931666e-06,
+      "loss": 3.2776,
+      "step": 4938
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.003029970701923e-06,
+      "loss": 3.4793,
+      "step": 4939
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0010746972579865e-06,
+      "loss": 3.3794,
+      "step": 4940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.999119582911263e-06,
+      "loss": 3.4365,
+      "step": 4941
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.997164627973146e-06,
+      "loss": 3.4693,
+      "step": 4942
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.995209832754997e-06,
+      "loss": 3.5101,
+      "step": 4943
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.993255197568154e-06,
+      "loss": 3.3672,
+      "step": 4944
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991300722723924e-06,
+      "loss": 3.4616,
+      "step": 4945
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.989346408533597e-06,
+      "loss": 3.4751,
+      "step": 4946
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.987392255308432e-06,
+      "loss": 3.4916,
+      "step": 4947
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.985438263359667e-06,
+      "loss": 3.5076,
+      "step": 4948
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9834844329985065e-06,
+      "loss": 3.4552,
+      "step": 4949
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9815307645361356e-06,
+      "loss": 3.3376,
+      "step": 4950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.979577258283713e-06,
+      "loss": 3.5612,
+      "step": 4951
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.97762391455237e-06,
+      "loss": 3.5866,
+      "step": 4952
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.975670733653211e-06,
+      "loss": 3.4292,
+      "step": 4953
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9737177158973185e-06,
+      "loss": 3.4604,
+      "step": 4954
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.971764861595744e-06,
+      "loss": 3.3375,
+      "step": 4955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.969812171059516e-06,
+      "loss": 3.4834,
+      "step": 4956
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.967859644599639e-06,
+      "loss": 3.4804,
+      "step": 4957
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.965907282527085e-06,
+      "loss": 3.4822,
+      "step": 4958
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.963955085152805e-06,
+      "loss": 3.4844,
+      "step": 4959
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9620030527877245e-06,
+      "loss": 3.4624,
+      "step": 4960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.960051185742737e-06,
+      "loss": 3.3603,
+      "step": 4961
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9580994843287145e-06,
+      "loss": 3.4558,
+      "step": 4962
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.956147948856503e-06,
+      "loss": 3.5039,
+      "step": 4963
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.954196579636918e-06,
+      "loss": 3.3976,
+      "step": 4964
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.952245376980751e-06,
+      "loss": 3.5012,
+      "step": 4965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.950294341198768e-06,
+      "loss": 3.3619,
+      "step": 4966
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.948343472601708e-06,
+      "loss": 3.3758,
+      "step": 4967
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.946392771500282e-06,
+      "loss": 3.4017,
+      "step": 4968
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.944442238205174e-06,
+      "loss": 3.3953,
+      "step": 4969
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.942491873027044e-06,
+      "loss": 3.3841,
+      "step": 4970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.940541676276522e-06,
+      "loss": 3.3935,
+      "step": 4971
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.938591648264214e-06,
+      "loss": 3.4261,
+      "step": 4972
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9366417893006954e-06,
+      "loss": 3.4694,
+      "step": 4973
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.93469209969652e-06,
+      "loss": 3.4239,
+      "step": 4974
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.932742579762211e-06,
+      "loss": 3.511,
+      "step": 4975
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.930793229808265e-06,
+      "loss": 3.471,
+      "step": 4976
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.92884405014515e-06,
+      "loss": 3.4237,
+      "step": 4977
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.926895041083311e-06,
+      "loss": 3.3343,
+      "step": 4978
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9249462029331635e-06,
+      "loss": 3.3366,
+      "step": 4979
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.922997536005094e-06,
+      "loss": 3.3779,
+      "step": 4980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.921049040609466e-06,
+      "loss": 3.4614,
+      "step": 4981
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.91910071705661e-06,
+      "loss": 3.5065,
+      "step": 4982
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.917152565656835e-06,
+      "loss": 3.4264,
+      "step": 4983
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.91520458672042e-06,
+      "loss": 3.4004,
+      "step": 4984
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9132567805576134e-06,
+      "loss": 3.3918,
+      "step": 4985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.911309147478643e-06,
+      "loss": 3.3919,
+      "step": 4986
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.909361687793704e-06,
+      "loss": 3.4045,
+      "step": 4987
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.907414401812963e-06,
+      "loss": 3.3835,
+      "step": 4988
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.905467289846563e-06,
+      "loss": 3.4532,
+      "step": 4989
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.903520352204615e-06,
+      "loss": 3.4294,
+      "step": 4990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9015735891972095e-06,
+      "loss": 3.3426,
+      "step": 4991
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.899627001134402e-06,
+      "loss": 3.4331,
+      "step": 4992
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.897680588326221e-06,
+      "loss": 3.4145,
+      "step": 4993
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.895734351082668e-06,
+      "loss": 3.4035,
+      "step": 4994
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.893788289713721e-06,
+      "loss": 3.4093,
+      "step": 4995
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.891842404529321e-06,
+      "loss": 3.5437,
+      "step": 4996
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.88989669583939e-06,
+      "loss": 3.4626,
+      "step": 4997
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.887951163953817e-06,
+      "loss": 3.4167,
+      "step": 4998
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.886005809182462e-06,
+      "loss": 3.4507,
+      "step": 4999
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.884060631835159e-06,
+      "loss": 3.4702,
+      "step": 5000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.882115632221715e-06,
+      "loss": 3.3433,
+      "step": 5001
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.880170810651906e-06,
+      "loss": 3.4972,
+      "step": 5002
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.878226167435479e-06,
+      "loss": 3.4878,
+      "step": 5003
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.876281702882156e-06,
+      "loss": 3.4807,
+      "step": 5004
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.87433741730163e-06,
+      "loss": 3.3576,
+      "step": 5005
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.872393311003561e-06,
+      "loss": 3.4155,
+      "step": 5006
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.870449384297585e-06,
+      "loss": 3.4153,
+      "step": 5007
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.86850563749331e-06,
+      "loss": 3.3677,
+      "step": 5008
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.866562070900309e-06,
+      "loss": 3.4243,
+      "step": 5009
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.864618684828135e-06,
+      "loss": 3.4106,
+      "step": 5010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.862675479586306e-06,
+      "loss": 3.4616,
+      "step": 5011
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.860732455484314e-06,
+      "loss": 3.4304,
+      "step": 5012
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.85878961283162e-06,
+      "loss": 3.4303,
+      "step": 5013
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8568469519376585e-06,
+      "loss": 3.3532,
+      "step": 5014
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.854904473111831e-06,
+      "loss": 3.4323,
+      "step": 5015
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.852962176663518e-06,
+      "loss": 3.417,
+      "step": 5016
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.851020062902064e-06,
+      "loss": 3.3723,
+      "step": 5017
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.849078132136785e-06,
+      "loss": 3.4117,
+      "step": 5018
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.847136384676971e-06,
+      "loss": 3.4324,
+      "step": 5019
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.845194820831878e-06,
+      "loss": 3.4594,
+      "step": 5020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.843253440910739e-06,
+      "loss": 3.4396,
+      "step": 5021
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.841312245222754e-06,
+      "loss": 3.4319,
+      "step": 5022
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.839371234077092e-06,
+      "loss": 3.4492,
+      "step": 5023
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.837430407782896e-06,
+      "loss": 3.3213,
+      "step": 5024
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.835489766649279e-06,
+      "loss": 3.2486,
+      "step": 5025
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.833549310985322e-06,
+      "loss": 3.4507,
+      "step": 5026
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8316090411000794e-06,
+      "loss": 3.4281,
+      "step": 5027
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.829668957302576e-06,
+      "loss": 3.2558,
+      "step": 5028
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.827729059901803e-06,
+      "loss": 3.4629,
+      "step": 5029
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.825789349206726e-06,
+      "loss": 3.3647,
+      "step": 5030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8238498255262806e-06,
+      "loss": 3.4136,
+      "step": 5031
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.821910489169371e-06,
+      "loss": 3.2827,
+      "step": 5032
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.819971340444871e-06,
+      "loss": 3.471,
+      "step": 5033
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.818032379661626e-06,
+      "loss": 3.379,
+      "step": 5034
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.816093607128454e-06,
+      "loss": 3.4508,
+      "step": 5035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.814155023154136e-06,
+      "loss": 3.4508,
+      "step": 5036
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.81221662804743e-06,
+      "loss": 3.3598,
+      "step": 5037
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.810278422117061e-06,
+      "loss": 3.5011,
+      "step": 5038
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8083404056717187e-06,
+      "loss": 3.4394,
+      "step": 5039
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8064025790200766e-06,
+      "loss": 3.4044,
+      "step": 5040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.804464942470763e-06,
+      "loss": 3.5323,
+      "step": 5041
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8025274963323844e-06,
+      "loss": 3.4525,
+      "step": 5042
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8005902409135147e-06,
+      "loss": 3.3684,
+      "step": 5043
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7986531765226965e-06,
+      "loss": 3.3556,
+      "step": 5044
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7967163034684435e-06,
+      "loss": 3.4011,
+      "step": 5045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7947796220592386e-06,
+      "loss": 3.5325,
+      "step": 5046
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.792843132603533e-06,
+      "loss": 3.3736,
+      "step": 5047
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7909068354097487e-06,
+      "loss": 3.3599,
+      "step": 5048
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.788970730786278e-06,
+      "loss": 3.3988,
+      "step": 5049
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7870348190414785e-06,
+      "loss": 3.2613,
+      "step": 5050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7850991004836813e-06,
+      "loss": 3.3977,
+      "step": 5051
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7831635754211856e-06,
+      "loss": 3.3991,
+      "step": 5052
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.781228244162258e-06,
+      "loss": 3.3883,
+      "step": 5053
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.779293107015136e-06,
+      "loss": 3.4244,
+      "step": 5054
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7773581642880275e-06,
+      "loss": 3.395,
+      "step": 5055
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7754234162891048e-06,
+      "loss": 3.2655,
+      "step": 5056
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7734888633265133e-06,
+      "loss": 3.4465,
+      "step": 5057
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.771554505708367e-06,
+      "loss": 3.3255,
+      "step": 5058
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.769620343742748e-06,
+      "loss": 3.4813,
+      "step": 5059
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7676863777377055e-06,
+      "loss": 3.4273,
+      "step": 5060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7657526080012596e-06,
+      "loss": 3.4655,
+      "step": 5061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7638190348414007e-06,
+      "loss": 3.5138,
+      "step": 5062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.761885658566083e-06,
+      "loss": 3.3764,
+      "step": 5063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.759952479483232e-06,
+      "loss": 3.4728,
+      "step": 5064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7580194979007456e-06,
+      "loss": 3.3141,
+      "step": 5065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7560867141264834e-06,
+      "loss": 3.3765,
+      "step": 5066
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7541541284682797e-06,
+      "loss": 3.3892,
+      "step": 5067
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7522217412339313e-06,
+      "loss": 3.3857,
+      "step": 5068
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7502895527312077e-06,
+      "loss": 3.4129,
+      "step": 5069
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.748357563267846e-06,
+      "loss": 3.4587,
+      "step": 5070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7464257731515495e-06,
+      "loss": 3.5066,
+      "step": 5071
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7444941826899924e-06,
+      "loss": 3.3724,
+      "step": 5072
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7425627921908163e-06,
+      "loss": 3.44,
+      "step": 5073
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7406316019616297e-06,
+      "loss": 3.3203,
+      "step": 5074
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7387006123100094e-06,
+      "loss": 3.4873,
+      "step": 5075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7367698235435036e-06,
+      "loss": 3.3348,
+      "step": 5076
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.734839235969624e-06,
+      "loss": 3.4211,
+      "step": 5077
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7329088498958522e-06,
+      "loss": 3.4978,
+      "step": 5078
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.730978665629639e-06,
+      "loss": 3.3575,
+      "step": 5079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.729048683478399e-06,
+      "loss": 3.5577,
+      "step": 5080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7271189037495203e-06,
+      "loss": 3.3208,
+      "step": 5081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7251893267503547e-06,
+      "loss": 3.461,
+      "step": 5082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7232599527882226e-06,
+      "loss": 3.398,
+      "step": 5083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7213307821704115e-06,
+      "loss": 3.5119,
+      "step": 5084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.719401815204179e-06,
+      "loss": 3.5193,
+      "step": 5085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7174730521967476e-06,
+      "loss": 3.3818,
+      "step": 5086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.715544493455308e-06,
+      "loss": 3.3805,
+      "step": 5087
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7136161392870185e-06,
+      "loss": 3.4567,
+      "step": 5088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7116879899990054e-06,
+      "loss": 3.319,
+      "step": 5089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7097600458983634e-06,
+      "loss": 3.4362,
+      "step": 5090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.707832307292151e-06,
+      "loss": 3.3693,
+      "step": 5091
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.705904774487396e-06,
+      "loss": 3.5046,
+      "step": 5092
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7039774477910945e-06,
+      "loss": 3.399,
+      "step": 5093
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7020503275102095e-06,
+      "loss": 3.3589,
+      "step": 5094
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7001234139516674e-06,
+      "loss": 3.5455,
+      "step": 5095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6981967074223662e-06,
+      "loss": 3.2906,
+      "step": 5096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.696270208229171e-06,
+      "loss": 3.4469,
+      "step": 5097
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.694343916678909e-06,
+      "loss": 3.3385,
+      "step": 5098
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.69241783307838e-06,
+      "loss": 3.4488,
+      "step": 5099
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.690491957734348e-06,
+      "loss": 3.3679,
+      "step": 5100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.688566290953542e-06,
+      "loss": 3.3523,
+      "step": 5101
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.686640833042662e-06,
+      "loss": 3.3987,
+      "step": 5102
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.684715584308372e-06,
+      "loss": 3.4163,
+      "step": 5103
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6827905450573022e-06,
+      "loss": 3.4607,
+      "step": 5104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.680865715596052e-06,
+      "loss": 3.404,
+      "step": 5105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6789410962311854e-06,
+      "loss": 3.5357,
+      "step": 5106
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.677016687269233e-06,
+      "loss": 3.4263,
+      "step": 5107
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.675092489016693e-06,
+      "loss": 3.3322,
+      "step": 5108
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.673168501780029e-06,
+      "loss": 3.4881,
+      "step": 5109
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6712447258656702e-06,
+      "loss": 3.4738,
+      "step": 5110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6693211615800145e-06,
+      "loss": 3.5311,
+      "step": 5111
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.667397809229426e-06,
+      "loss": 3.3578,
+      "step": 5112
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6654746691202293e-06,
+      "loss": 3.4946,
+      "step": 5113
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6635517415587264e-06,
+      "loss": 3.4718,
+      "step": 5114
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.661629026851174e-06,
+      "loss": 3.5423,
+      "step": 5115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6597065253038016e-06,
+      "loss": 3.4424,
+      "step": 5116
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6577842372228034e-06,
+      "loss": 3.2947,
+      "step": 5117
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.655862162914338e-06,
+      "loss": 3.3921,
+      "step": 5118
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.653940302684531e-06,
+      "loss": 3.3268,
+      "step": 5119
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6520186568394743e-06,
+      "loss": 3.4836,
+      "step": 5120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.650097225685227e-06,
+      "loss": 3.3296,
+      "step": 5121
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6481760095278097e-06,
+      "loss": 3.5118,
+      "step": 5122
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6462550086732116e-06,
+      "loss": 3.4879,
+      "step": 5123
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6443342234273905e-06,
+      "loss": 3.4354,
+      "step": 5124
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.642413654096262e-06,
+      "loss": 3.4921,
+      "step": 5125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.640493300985716e-06,
+      "loss": 3.4022,
+      "step": 5126
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.638573164401603e-06,
+      "loss": 3.3015,
+      "step": 5127
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6366532446497383e-06,
+      "loss": 3.4247,
+      "step": 5128
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6347335420359066e-06,
+      "loss": 3.4608,
+      "step": 5129
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6328140568658552e-06,
+      "loss": 3.5359,
+      "step": 5130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.630894789445296e-06,
+      "loss": 3.3133,
+      "step": 5131
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.62897574007991e-06,
+      "loss": 3.398,
+      "step": 5132
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.62705690907534e-06,
+      "loss": 3.455,
+      "step": 5133
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6251382967371938e-06,
+      "loss": 3.4422,
+      "step": 5134
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6232199033710476e-06,
+      "loss": 3.4882,
+      "step": 5135
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6213017292824404e-06,
+      "loss": 3.4178,
+      "step": 5136
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.619383774776876e-06,
+      "loss": 3.4566,
+      "step": 5137
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6174660401598226e-06,
+      "loss": 3.3264,
+      "step": 5138
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6155485257367184e-06,
+      "loss": 3.3554,
+      "step": 5139
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.61363123181296e-06,
+      "loss": 3.3852,
+      "step": 5140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6117141586939144e-06,
+      "loss": 3.3815,
+      "step": 5141
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.609797306684907e-06,
+      "loss": 3.4061,
+      "step": 5142
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.607880676091233e-06,
+      "loss": 3.3145,
+      "step": 5143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6059642672181537e-06,
+      "loss": 3.4607,
+      "step": 5144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6040480803708887e-06,
+      "loss": 3.3177,
+      "step": 5145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6021321158546286e-06,
+      "loss": 3.4406,
+      "step": 5146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6002163739745244e-06,
+      "loss": 3.3522,
+      "step": 5147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5983008550356946e-06,
+      "loss": 3.363,
+      "step": 5148
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5963855593432197e-06,
+      "loss": 3.3998,
+      "step": 5149
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.594470487202145e-06,
+      "loss": 3.4008,
+      "step": 5150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5925556389174836e-06,
+      "loss": 3.3824,
+      "step": 5151
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.590641014794207e-06,
+      "loss": 3.5338,
+      "step": 5152
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5887266151372567e-06,
+      "loss": 3.3962,
+      "step": 5153
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.586812440251537e-06,
+      "loss": 3.414,
+      "step": 5154
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5848984904419115e-06,
+      "loss": 3.4237,
+      "step": 5155
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.582984766013215e-06,
+      "loss": 3.4225,
+      "step": 5156
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.581071267270243e-06,
+      "loss": 3.4488,
+      "step": 5157
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5791579945177534e-06,
+      "loss": 3.4285,
+      "step": 5158
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.577244948060472e-06,
+      "loss": 3.3678,
+      "step": 5159
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.575332128203087e-06,
+      "loss": 3.3974,
+      "step": 5160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.573419535250248e-06,
+      "loss": 3.4931,
+      "step": 5161
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.571507169506571e-06,
+      "loss": 3.4362,
+      "step": 5162
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.569595031276637e-06,
+      "loss": 3.4921,
+      "step": 5163
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5676831208649887e-06,
+      "loss": 3.399,
+      "step": 5164
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5657714385761333e-06,
+      "loss": 3.4139,
+      "step": 5165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5638599847145396e-06,
+      "loss": 3.4378,
+      "step": 5166
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5619487595846426e-06,
+      "loss": 3.3254,
+      "step": 5167
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5600377634908417e-06,
+      "loss": 3.3526,
+      "step": 5168
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5581269967374955e-06,
+      "loss": 3.347,
+      "step": 5169
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5562164596289296e-06,
+      "loss": 3.4975,
+      "step": 5170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5543061524694344e-06,
+      "loss": 3.4537,
+      "step": 5171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5523960755632573e-06,
+      "loss": 3.4325,
+      "step": 5172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.550486229214616e-06,
+      "loss": 3.2788,
+      "step": 5173
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5485766137276894e-06,
+      "loss": 3.3771,
+      "step": 5174
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5466672294066164e-06,
+      "loss": 3.418,
+      "step": 5175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5447580765555023e-06,
+      "loss": 3.4268,
+      "step": 5176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.542849155478415e-06,
+      "loss": 3.4087,
+      "step": 5177
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5409404664793867e-06,
+      "loss": 3.404,
+      "step": 5178
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5390320098624086e-06,
+      "loss": 3.4197,
+      "step": 5179
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5371237859314387e-06,
+      "loss": 3.4287,
+      "step": 5180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5352157949903987e-06,
+      "loss": 3.3422,
+      "step": 5181
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5333080373431673e-06,
+      "loss": 3.3313,
+      "step": 5182
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5314005132935925e-06,
+      "loss": 3.3535,
+      "step": 5183
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5294932231454838e-06,
+      "loss": 3.4597,
+      "step": 5184
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5275861672026084e-06,
+      "loss": 3.5182,
+      "step": 5185
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5256793457687026e-06,
+      "loss": 3.3268,
+      "step": 5186
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5237727591474612e-06,
+      "loss": 3.3947,
+      "step": 5187
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5218664076425455e-06,
+      "loss": 3.4421,
+      "step": 5188
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.519960291557577e-06,
+      "loss": 3.353,
+      "step": 5189
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.518054411196137e-06,
+      "loss": 3.3479,
+      "step": 5190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.516148766861773e-06,
+      "loss": 3.4995,
+      "step": 5191
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.514243358857997e-06,
+      "loss": 3.4398,
+      "step": 5192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5123381874882755e-06,
+      "loss": 3.3543,
+      "step": 5193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.510433253056045e-06,
+      "loss": 3.3987,
+      "step": 5194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5085285558647007e-06,
+      "loss": 3.3752,
+      "step": 5195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5066240962176e-06,
+      "loss": 3.5084,
+      "step": 5196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5047198744180632e-06,
+      "loss": 3.4728,
+      "step": 5197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5028158907693744e-06,
+      "loss": 3.5062,
+      "step": 5198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.500912145574775e-06,
+      "loss": 3.3574,
+      "step": 5199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4990086391374734e-06,
+      "loss": 3.3523,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.497105371760639e-06,
+      "loss": 3.5638,
+      "step": 5201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.495202343747399e-06,
+      "loss": 3.3388,
+      "step": 5202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4932995554008476e-06,
+      "loss": 3.376,
+      "step": 5203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4913970070240388e-06,
+      "loss": 3.4058,
+      "step": 5204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.489494698919989e-06,
+      "loss": 3.3796,
+      "step": 5205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4875926313916743e-06,
+      "loss": 3.4745,
+      "step": 5206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4856908047420335e-06,
+      "loss": 3.4564,
+      "step": 5207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4837892192739702e-06,
+      "loss": 3.4214,
+      "step": 5208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.481887875290344e-06,
+      "loss": 3.3649,
+      "step": 5209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4799867730939795e-06,
+      "loss": 3.4844,
+      "step": 5210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.478085912987662e-06,
+      "loss": 3.3057,
+      "step": 5211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4761852952741405e-06,
+      "loss": 3.3519,
+      "step": 5212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4742849202561223e-06,
+      "loss": 3.5265,
+      "step": 5213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4723847882362755e-06,
+      "loss": 3.4429,
+      "step": 5214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4704848995172326e-06,
+      "loss": 3.4284,
+      "step": 5215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.468585254401586e-06,
+      "loss": 3.3774,
+      "step": 5216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4666858531918877e-06,
+      "loss": 3.3641,
+      "step": 5217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.464786696190653e-06,
+      "loss": 3.3619,
+      "step": 5218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4628877837003584e-06,
+      "loss": 3.3195,
+      "step": 5219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.460989116023439e-06,
+      "loss": 3.3144,
+      "step": 5220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4590906934622937e-06,
+      "loss": 3.3504,
+      "step": 5221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.457192516319281e-06,
+      "loss": 3.4489,
+      "step": 5222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.45529458489672e-06,
+      "loss": 3.4091,
+      "step": 5223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4533968994968913e-06,
+      "loss": 3.3746,
+      "step": 5224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4514994604220376e-06,
+      "loss": 3.4382,
+      "step": 5225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.449602267974359e-06,
+      "loss": 3.4353,
+      "step": 5226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4477053224560185e-06,
+      "loss": 3.4999,
+      "step": 5227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.445808624169141e-06,
+      "loss": 3.5525,
+      "step": 5228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4439121734158087e-06,
+      "loss": 3.4015,
+      "step": 5229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.442015970498067e-06,
+      "loss": 3.4021,
+      "step": 5230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4401200157179216e-06,
+      "loss": 3.4396,
+      "step": 5231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.438224309377339e-06,
+      "loss": 3.2058,
+      "step": 5232
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4363288517782426e-06,
+      "loss": 3.3816,
+      "step": 5233
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4344336432225205e-06,
+      "loss": 3.3466,
+      "step": 5234
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4325386840120206e-06,
+      "loss": 3.4187,
+      "step": 5235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4306439744485453e-06,
+      "loss": 3.4293,
+      "step": 5236
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.428749514833869e-06,
+      "loss": 3.3905,
+      "step": 5237
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4268553054697142e-06,
+      "loss": 3.5266,
+      "step": 5238
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.424961346657769e-06,
+      "loss": 3.4383,
+      "step": 5239
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.423067638699684e-06,
+      "loss": 3.4984,
+      "step": 5240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4211741818970644e-06,
+      "loss": 3.4487,
+      "step": 5241
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.419280976551479e-06,
+      "loss": 3.5336,
+      "step": 5242
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.417388022964455e-06,
+      "loss": 3.3555,
+      "step": 5243
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.41549532143748e-06,
+      "loss": 3.4401,
+      "step": 5244
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4136028722720014e-06,
+      "loss": 3.4562,
+      "step": 5245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4117106757694284e-06,
+      "loss": 3.5014,
+      "step": 5246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.409818732231126e-06,
+      "loss": 3.3766,
+      "step": 5247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.407927041958421e-06,
+      "loss": 3.399,
+      "step": 5248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4060356052526013e-06,
+      "loss": 3.4237,
+      "step": 5249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.404144422414911e-06,
+      "loss": 3.3149,
+      "step": 5250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.402253493746558e-06,
+      "loss": 3.4175,
+      "step": 5251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.400362819548706e-06,
+      "loss": 3.2989,
+      "step": 5252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3984724001224797e-06,
+      "loss": 3.3863,
+      "step": 5253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3965822357689626e-06,
+      "loss": 3.4803,
+      "step": 5254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3946923267892007e-06,
+      "loss": 3.4784,
+      "step": 5255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3928026734841935e-06,
+      "loss": 3.3859,
+      "step": 5256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3909132761549044e-06,
+      "loss": 3.4168,
+      "step": 5257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3890241351022547e-06,
+      "loss": 3.4192,
+      "step": 5258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.387135250627126e-06,
+      "loss": 3.4002,
+      "step": 5259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.385246623030356e-06,
+      "loss": 3.488,
+      "step": 5260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.383358252612742e-06,
+      "loss": 3.4277,
+      "step": 5261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.381470139675046e-06,
+      "loss": 3.4932,
+      "step": 5262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3795822845179826e-06,
+      "loss": 3.5063,
+      "step": 5263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3776946874422268e-06,
+      "loss": 3.4159,
+      "step": 5264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3758073487484132e-06,
+      "loss": 3.4556,
+      "step": 5265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3739202687371366e-06,
+      "loss": 3.4959,
+      "step": 5266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.372033447708948e-06,
+      "loss": 3.4381,
+      "step": 5267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3701468859643583e-06,
+      "loss": 3.4145,
+      "step": 5268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3682605838038375e-06,
+      "loss": 3.5149,
+      "step": 5269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3663745415278137e-06,
+      "loss": 3.3892,
+      "step": 5270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3644887594366736e-06,
+      "loss": 3.4298,
+      "step": 5271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3626032378307627e-06,
+      "loss": 3.4213,
+      "step": 5272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.360717977010387e-06,
+      "loss": 3.5025,
+      "step": 5273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3588329772758054e-06,
+      "loss": 3.3742,
+      "step": 5274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3569482389272405e-06,
+      "loss": 3.2981,
+      "step": 5275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.355063762264873e-06,
+      "loss": 3.5143,
+      "step": 5276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3531795475888374e-06,
+      "loss": 3.4055,
+      "step": 5277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3512955951992317e-06,
+      "loss": 3.4282,
+      "step": 5278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3494119053961094e-06,
+      "loss": 3.3544,
+      "step": 5279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.347528478479483e-06,
+      "loss": 3.4019,
+      "step": 5280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.345645314749322e-06,
+      "loss": 3.3924,
+      "step": 5281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3437624145055557e-06,
+      "loss": 3.4091,
+      "step": 5282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.34187977804807e-06,
+      "loss": 3.4158,
+      "step": 5283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3399974056767095e-06,
+      "loss": 3.4811,
+      "step": 5284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3381152976912756e-06,
+      "loss": 3.4476,
+      "step": 5285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3362334543915308e-06,
+      "loss": 3.3944,
+      "step": 5286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.334351876077193e-06,
+      "loss": 3.404,
+      "step": 5287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.332470563047936e-06,
+      "loss": 3.5765,
+      "step": 5288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.330589515603395e-06,
+      "loss": 3.3727,
+      "step": 5289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.328708734043163e-06,
+      "loss": 3.4351,
+      "step": 5290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3268282186667853e-06,
+      "loss": 3.4404,
+      "step": 5291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.324947969773771e-06,
+      "loss": 3.4777,
+      "step": 5292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.323067987663583e-06,
+      "loss": 3.4999,
+      "step": 5293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3211882726356447e-06,
+      "loss": 3.4954,
+      "step": 5294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.319308824989334e-06,
+      "loss": 3.3838,
+      "step": 5295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.317429645023987e-06,
+      "loss": 3.3561,
+      "step": 5296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3155507330389004e-06,
+      "loss": 3.5234,
+      "step": 5297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3136720893333223e-06,
+      "loss": 3.4313,
+      "step": 5298
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3117937142064627e-06,
+      "loss": 3.4396,
+      "step": 5299
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.309915607957487e-06,
+      "loss": 3.4628,
+      "step": 5300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3080377708855182e-06,
+      "loss": 3.308,
+      "step": 5301
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3061602032896365e-06,
+      "loss": 3.4436,
+      "step": 5302
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3042829054688797e-06,
+      "loss": 3.3055,
+      "step": 5303
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.30240587772224e-06,
+      "loss": 3.453,
+      "step": 5304
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3005291203486707e-06,
+      "loss": 3.4688,
+      "step": 5305
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2986526336470793e-06,
+      "loss": 3.3586,
+      "step": 5306
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.29677641791633e-06,
+      "loss": 3.5025,
+      "step": 5307
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.294900473455245e-06,
+      "loss": 3.4497,
+      "step": 5308
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.293024800562604e-06,
+      "loss": 3.3183,
+      "step": 5309
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.291149399537139e-06,
+      "loss": 3.4445,
+      "step": 5310
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.289274270677547e-06,
+      "loss": 3.5487,
+      "step": 5311
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.287399414282474e-06,
+      "loss": 3.3466,
+      "step": 5312
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.285524830650526e-06,
+      "loss": 3.4509,
+      "step": 5313
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.283650520080265e-06,
+      "loss": 3.3226,
+      "step": 5314
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2817764828702083e-06,
+      "loss": 3.3783,
+      "step": 5315
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2799027193188316e-06,
+      "loss": 3.4646,
+      "step": 5316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.278029229724566e-06,
+      "loss": 3.4463,
+      "step": 5317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2761560143857998e-06,
+      "loss": 3.4014,
+      "step": 5318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.274283073600876e-06,
+      "loss": 3.4138,
+      "step": 5319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.272410407668095e-06,
+      "loss": 3.4748,
+      "step": 5320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.270538016885715e-06,
+      "loss": 3.4924,
+      "step": 5321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.268665901551945e-06,
+      "loss": 3.3911,
+      "step": 5322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2667940619649575e-06,
+      "loss": 3.3321,
+      "step": 5323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2649224984228756e-06,
+      "loss": 3.416,
+      "step": 5324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.26305121122378e-06,
+      "loss": 3.5201,
+      "step": 5325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2611802006657086e-06,
+      "loss": 3.4575,
+      "step": 5326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.259309467046654e-06,
+      "loss": 3.4159,
+      "step": 5327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2574390106645636e-06,
+      "loss": 3.3183,
+      "step": 5328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2555688318173427e-06,
+      "loss": 3.379,
+      "step": 5329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2536989308028534e-06,
+      "loss": 3.428,
+      "step": 5330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.251829307918909e-06,
+      "loss": 3.4035,
+      "step": 5331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.249959963463283e-06,
+      "loss": 3.342,
+      "step": 5332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2480908977337035e-06,
+      "loss": 3.4796,
+      "step": 5333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.246222111027849e-06,
+      "loss": 3.4409,
+      "step": 5334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2443536036433653e-06,
+      "loss": 3.3656,
+      "step": 5335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.242485375877841e-06,
+      "loss": 3.378,
+      "step": 5336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2406174280288284e-06,
+      "loss": 3.3201,
+      "step": 5337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2387497603938327e-06,
+      "loss": 3.3462,
+      "step": 5338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2368823732703126e-06,
+      "loss": 3.3357,
+      "step": 5339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2350152669556845e-06,
+      "loss": 3.4854,
+      "step": 5340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.23314844174732e-06,
+      "loss": 3.2268,
+      "step": 5341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.231281897942544e-06,
+      "loss": 3.3775,
+      "step": 5342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2294156358386397e-06,
+      "loss": 3.475,
+      "step": 5343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.227549655732843e-06,
+      "loss": 3.3851,
+      "step": 5344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2256839579223444e-06,
+      "loss": 3.4917,
+      "step": 5345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.223818542704291e-06,
+      "loss": 3.4982,
+      "step": 5346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2219534103757866e-06,
+      "loss": 3.3871,
+      "step": 5347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2200885612338846e-06,
+      "loss": 3.394,
+      "step": 5348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2182239955755977e-06,
+      "loss": 3.4588,
+      "step": 5349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.216359713697893e-06,
+      "loss": 3.336,
+      "step": 5350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.214495715897692e-06,
+      "loss": 3.478,
+      "step": 5351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2126320024718696e-06,
+      "loss": 3.5158,
+      "step": 5352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2107685737172557e-06,
+      "loss": 3.3638,
+      "step": 5353
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2089054299306376e-06,
+      "loss": 3.4754,
+      "step": 5354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.207042571408754e-06,
+      "loss": 3.4319,
+      "step": 5355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.205179998448298e-06,
+      "loss": 3.5127,
+      "step": 5356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.203317711345922e-06,
+      "loss": 3.3896,
+      "step": 5357
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.201455710398227e-06,
+      "loss": 3.4456,
+      "step": 5358
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1995939959017697e-06,
+      "loss": 3.4503,
+      "step": 5359
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1977325681530647e-06,
+      "loss": 3.3919,
+      "step": 5360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1958714274485785e-06,
+      "loss": 3.3905,
+      "step": 5361
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.194010574084733e-06,
+      "loss": 3.3545,
+      "step": 5362
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1921500083578994e-06,
+      "loss": 3.4068,
+      "step": 5363
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.19028973056441e-06,
+      "loss": 3.4539,
+      "step": 5364
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.188429741000549e-06,
+      "loss": 3.3438,
+      "step": 5365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.186570039962551e-06,
+      "loss": 3.4834,
+      "step": 5366
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.184710627746609e-06,
+      "loss": 3.3886,
+      "step": 5367
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1828515046488695e-06,
+      "loss": 3.391,
+      "step": 5368
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1809926709654304e-06,
+      "loss": 3.4266,
+      "step": 5369
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.179134126992346e-06,
+      "loss": 3.4515,
+      "step": 5370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1772758730256237e-06,
+      "loss": 3.3955,
+      "step": 5371
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1754179093612246e-06,
+      "loss": 3.4146,
+      "step": 5372
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1735602362950624e-06,
+      "loss": 3.4587,
+      "step": 5373
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.171702854123008e-06,
+      "loss": 3.4135,
+      "step": 5374
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1698457631408813e-06,
+      "loss": 3.3148,
+      "step": 5375
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1679889636444594e-06,
+      "loss": 3.4826,
+      "step": 5376
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.166132455929471e-06,
+      "loss": 3.3809,
+      "step": 5377
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1642762402916005e-06,
+      "loss": 3.3282,
+      "step": 5378
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1624203170264832e-06,
+      "loss": 3.4491,
+      "step": 5379
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.160564686429709e-06,
+      "loss": 3.4964,
+      "step": 5380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1587093487968225e-06,
+      "loss": 3.4387,
+      "step": 5381
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.156854304423318e-06,
+      "loss": 3.4744,
+      "step": 5382
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1549995536046463e-06,
+      "loss": 3.4694,
+      "step": 5383
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.153145096636211e-06,
+      "loss": 3.3477,
+      "step": 5384
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1512909338133683e-06,
+      "loss": 3.4669,
+      "step": 5385
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1494370654314294e-06,
+      "loss": 3.4053,
+      "step": 5386
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.147583491785654e-06,
+      "loss": 3.4378,
+      "step": 5387
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1457302131712595e-06,
+      "loss": 3.4291,
+      "step": 5388
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.143877229883415e-06,
+      "loss": 3.4633,
+      "step": 5389
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1420245422172403e-06,
+      "loss": 3.3361,
+      "step": 5390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1401721504678112e-06,
+      "loss": 3.4324,
+      "step": 5391
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1383200549301563e-06,
+      "loss": 3.2968,
+      "step": 5392
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.136468255899253e-06,
+      "loss": 3.3966,
+      "step": 5393
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.134616753670036e-06,
+      "loss": 3.2911,
+      "step": 5394
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1327655485373914e-06,
+      "loss": 3.4215,
+      "step": 5395
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1309146407961565e-06,
+      "loss": 3.5128,
+      "step": 5396
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1290640307411233e-06,
+      "loss": 3.3647,
+      "step": 5397
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.127213718667036e-06,
+      "loss": 3.3564,
+      "step": 5398
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1253637048685893e-06,
+      "loss": 3.3962,
+      "step": 5399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.123513989640432e-06,
+      "loss": 3.3504,
+      "step": 5400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1216645732771676e-06,
+      "loss": 3.4368,
+      "step": 5401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.119815456073346e-06,
+      "loss": 3.4515,
+      "step": 5402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1179666383234763e-06,
+      "loss": 3.5455,
+      "step": 5403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1161181203220146e-06,
+      "loss": 3.303,
+      "step": 5404
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1142699023633736e-06,
+      "loss": 3.3574,
+      "step": 5405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1124219847419134e-06,
+      "loss": 3.3795,
+      "step": 5406
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1105743677519495e-06,
+      "loss": 3.455,
+      "step": 5407
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.108727051687749e-06,
+      "loss": 3.3465,
+      "step": 5408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1068800368435337e-06,
+      "loss": 3.4997,
+      "step": 5409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.105033323513471e-06,
+      "loss": 3.4684,
+      "step": 5410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1031869119916853e-06,
+      "loss": 3.4297,
+      "step": 5411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1013408025722515e-06,
+      "loss": 3.5018,
+      "step": 5412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.099494995549197e-06,
+      "loss": 3.3889,
+      "step": 5413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0976494912165e-06,
+      "loss": 3.5618,
+      "step": 5414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0958042898680905e-06,
+      "loss": 3.3013,
+      "step": 5415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0939593917978522e-06,
+      "loss": 3.4276,
+      "step": 5416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.092114797299617e-06,
+      "loss": 3.429,
+      "step": 5417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0902705066671717e-06,
+      "loss": 3.4419,
+      "step": 5418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0884265201942543e-06,
+      "loss": 3.3592,
+      "step": 5419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0865828381745515e-06,
+      "loss": 3.3469,
+      "step": 5420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.084739460901705e-06,
+      "loss": 3.4616,
+      "step": 5421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0828963886693066e-06,
+      "loss": 3.3609,
+      "step": 5422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0810536217708985e-06,
+      "loss": 3.4075,
+      "step": 5423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.079211160499975e-06,
+      "loss": 3.424,
+      "step": 5424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.077369005149984e-06,
+      "loss": 3.3768,
+      "step": 5425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0755271560143206e-06,
+      "loss": 3.4261,
+      "step": 5426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0736856133863335e-06,
+      "loss": 3.5139,
+      "step": 5427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0718443775593233e-06,
+      "loss": 3.5569,
+      "step": 5428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0700034488265386e-06,
+      "loss": 3.3506,
+      "step": 5429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0681628274811825e-06,
+      "loss": 3.4157,
+      "step": 5430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0663225138164076e-06,
+      "loss": 3.3365,
+      "step": 5431
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0644825081253187e-06,
+      "loss": 3.3581,
+      "step": 5432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0626428107009674e-06,
+      "loss": 3.4049,
+      "step": 5433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.060803421836363e-06,
+      "loss": 3.3784,
+      "step": 5434
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0589643418244604e-06,
+      "loss": 3.4199,
+      "step": 5435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0571255709581676e-06,
+      "loss": 3.4333,
+      "step": 5436
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0552871095303412e-06,
+      "loss": 3.4197,
+      "step": 5437
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0534489578337907e-06,
+      "loss": 3.4194,
+      "step": 5438
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0516111161612753e-06,
+      "loss": 3.367,
+      "step": 5439
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.049773584805507e-06,
+      "loss": 3.4141,
+      "step": 5440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0479363640591435e-06,
+      "loss": 3.4162,
+      "step": 5441
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.046099454214798e-06,
+      "loss": 3.3662,
+      "step": 5442
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.044262855565032e-06,
+      "loss": 3.4555,
+      "step": 5443
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0424265684023556e-06,
+      "loss": 3.396,
+      "step": 5444
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.040590593019234e-06,
+      "loss": 3.3865,
+      "step": 5445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.038754929708079e-06,
+      "loss": 3.3636,
+      "step": 5446
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.036919578761252e-06,
+      "loss": 3.3251,
+      "step": 5447
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0350845404710687e-06,
+      "loss": 3.4795,
+      "step": 5448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.033249815129793e-06,
+      "loss": 3.459,
+      "step": 5449
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.031415403029636e-06,
+      "loss": 3.4252,
+      "step": 5450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.029581304462763e-06,
+      "loss": 3.4151,
+      "step": 5451
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0277475197212896e-06,
+      "loss": 3.4513,
+      "step": 5452
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0259140490972772e-06,
+      "loss": 3.4533,
+      "step": 5453
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0240808928827397e-06,
+      "loss": 3.32,
+      "step": 5454
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.022248051369644e-06,
+      "loss": 3.4618,
+      "step": 5455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0204155248498996e-06,
+      "loss": 3.3371,
+      "step": 5456
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0185833136153713e-06,
+      "loss": 3.4233,
+      "step": 5457
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0167514179578743e-06,
+      "loss": 3.3349,
+      "step": 5458
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0149198381691706e-06,
+      "loss": 3.2949,
+      "step": 5459
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0130885745409744e-06,
+      "loss": 3.3773,
+      "step": 5460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.011257627364945e-06,
+      "loss": 3.3946,
+      "step": 5461
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.009426996932696e-06,
+      "loss": 3.3792,
+      "step": 5462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.00759668353579e-06,
+      "loss": 3.5678,
+      "step": 5463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0057666874657365e-06,
+      "loss": 3.3684,
+      "step": 5464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0039370090139956e-06,
+      "loss": 3.3867,
+      "step": 5465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0021076484719793e-06,
+      "loss": 3.325,
+      "step": 5466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0002786061310453e-06,
+      "loss": 3.3687,
+      "step": 5467
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9984498822825027e-06,
+      "loss": 3.3337,
+      "step": 5468
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.996621477217609e-06,
+      "loss": 3.3838,
+      "step": 5469
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9947933912275727e-06,
+      "loss": 3.4837,
+      "step": 5470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9929656246035476e-06,
+      "loss": 3.3975,
+      "step": 5471
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.991138177636641e-06,
+      "loss": 3.4057,
+      "step": 5472
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9893110506179084e-06,
+      "loss": 3.2796,
+      "step": 5473
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9874842438383508e-06,
+      "loss": 3.3802,
+      "step": 5474
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9856577575889213e-06,
+      "loss": 3.4676,
+      "step": 5475
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9838315921605237e-06,
+      "loss": 3.4684,
+      "step": 5476
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.982005747844006e-06,
+      "loss": 3.3934,
+      "step": 5477
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.980180224930167e-06,
+      "loss": 3.3048,
+      "step": 5478
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.978355023709758e-06,
+      "loss": 3.4348,
+      "step": 5479
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.976530144473473e-06,
+      "loss": 3.3711,
+      "step": 5480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.974705587511958e-06,
+      "loss": 3.3056,
+      "step": 5481
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9728813531158063e-06,
+      "loss": 3.4104,
+      "step": 5482
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9710574415755635e-06,
+      "loss": 3.3783,
+      "step": 5483
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9692338531817205e-06,
+      "loss": 3.4297,
+      "step": 5484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.967410588224715e-06,
+      "loss": 3.4039,
+      "step": 5485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.965587646994938e-06,
+      "loss": 3.3355,
+      "step": 5486
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9637650297827258e-06,
+      "loss": 3.4212,
+      "step": 5487
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9619427368783617e-06,
+      "loss": 3.3477,
+      "step": 5488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.960120768572081e-06,
+      "loss": 3.3336,
+      "step": 5489
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.958299125154067e-06,
+      "loss": 3.4669,
+      "step": 5490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.956477806914447e-06,
+      "loss": 3.3709,
+      "step": 5491
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9546568141433007e-06,
+      "loss": 3.4849,
+      "step": 5492
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.952836147130655e-06,
+      "loss": 3.444,
+      "step": 5493
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.951015806166485e-06,
+      "loss": 3.4744,
+      "step": 5494
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.949195791540712e-06,
+      "loss": 3.5007,
+      "step": 5495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.947376103543208e-06,
+      "loss": 3.3579,
+      "step": 5496
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9455567424637923e-06,
+      "loss": 3.3853,
+      "step": 5497
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9437377085922288e-06,
+      "loss": 3.3398,
+      "step": 5498
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.941919002218234e-06,
+      "loss": 3.4345,
+      "step": 5499
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9401006236314722e-06,
+      "loss": 3.3267,
+      "step": 5500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.93828257312155e-06,
+      "loss": 3.38,
+      "step": 5501
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.936464850978027e-06,
+      "loss": 3.4113,
+      "step": 5502
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9346474574904104e-06,
+      "loss": 3.4238,
+      "step": 5503
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9328303929481507e-06,
+      "loss": 3.4667,
+      "step": 5504
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.93101365764065e-06,
+      "loss": 3.5196,
+      "step": 5505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9291972518572576e-06,
+      "loss": 3.5505,
+      "step": 5506
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9273811758872663e-06,
+      "loss": 3.5036,
+      "step": 5507
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9255654300199253e-06,
+      "loss": 3.3687,
+      "step": 5508
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9237500145444207e-06,
+      "loss": 3.4354,
+      "step": 5509
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9219349297498934e-06,
+      "loss": 3.3558,
+      "step": 5510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.920120175925428e-06,
+      "loss": 3.4696,
+      "step": 5511
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.918305753360057e-06,
+      "loss": 3.4168,
+      "step": 5512
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9164916623427624e-06,
+      "loss": 3.4985,
+      "step": 5513
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9146779031624683e-06,
+      "loss": 3.4378,
+      "step": 5514
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9128644761080517e-06,
+      "loss": 3.5398,
+      "step": 5515
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9110513814683335e-06,
+      "loss": 3.4125,
+      "step": 5516
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.909238619532082e-06,
+      "loss": 3.4586,
+      "step": 5517
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.907426190588013e-06,
+      "loss": 3.4108,
+      "step": 5518
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9056140949247917e-06,
+      "loss": 3.4434,
+      "step": 5519
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.903802332831024e-06,
+      "loss": 3.5206,
+      "step": 5520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9019909045952668e-06,
+      "loss": 3.515,
+      "step": 5521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9001798105060243e-06,
+      "loss": 3.4436,
+      "step": 5522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.898369050851746e-06,
+      "loss": 3.5542,
+      "step": 5523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8965586259208295e-06,
+      "loss": 3.399,
+      "step": 5524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8947485360016194e-06,
+      "loss": 3.3896,
+      "step": 5525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8929387813824026e-06,
+      "loss": 3.4113,
+      "step": 5526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8911293623514176e-06,
+      "loss": 3.427,
+      "step": 5527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8893202791968466e-06,
+      "loss": 3.3867,
+      "step": 5528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.887511532206821e-06,
+      "loss": 3.3756,
+      "step": 5529
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.885703121669416e-06,
+      "loss": 3.4745,
+      "step": 5530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.883895047872654e-06,
+      "loss": 3.4418,
+      "step": 5531
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8820873111045045e-06,
+      "loss": 3.361,
+      "step": 5532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8802799116528835e-06,
+      "loss": 3.4486,
+      "step": 5533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.878472849805651e-06,
+      "loss": 3.4792,
+      "step": 5534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8766661258506156e-06,
+      "loss": 3.3666,
+      "step": 5535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8748597400755336e-06,
+      "loss": 3.4073,
+      "step": 5536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8730536927681007e-06,
+      "loss": 3.3891,
+      "step": 5537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8712479842159647e-06,
+      "loss": 3.3764,
+      "step": 5538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8694426147067188e-06,
+      "loss": 3.3647,
+      "step": 5539
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8676375845279013e-06,
+      "loss": 3.3852,
+      "step": 5540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8658328939669965e-06,
+      "loss": 3.3809,
+      "step": 5541
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8640285433114336e-06,
+      "loss": 3.3729,
+      "step": 5542
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.862224532848591e-06,
+      "loss": 3.4369,
+      "step": 5543
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.860420862865787e-06,
+      "loss": 3.4755,
+      "step": 5544
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.85861753365029e-06,
+      "loss": 3.3183,
+      "step": 5545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.856814545489315e-06,
+      "loss": 3.3568,
+      "step": 5546
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8550118986700204e-06,
+      "loss": 3.3953,
+      "step": 5547
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.85320959347951e-06,
+      "loss": 3.3958,
+      "step": 5548
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8514076302048366e-06,
+      "loss": 3.3258,
+      "step": 5549
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8496060091329925e-06,
+      "loss": 3.333,
+      "step": 5550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8478047305509205e-06,
+      "loss": 3.3961,
+      "step": 5551
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.846003794745507e-06,
+      "loss": 3.4731,
+      "step": 5552
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.844203202003585e-06,
+      "loss": 3.474,
+      "step": 5553
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.842402952611931e-06,
+      "loss": 3.451,
+      "step": 5554
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.84060304685727e-06,
+      "loss": 3.4675,
+      "step": 5555
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.838803485026265e-06,
+      "loss": 3.2898,
+      "step": 5556
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8370042674055344e-06,
+      "loss": 3.3597,
+      "step": 5557
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.835205394281635e-06,
+      "loss": 3.3819,
+      "step": 5558
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8334068659410706e-06,
+      "loss": 3.422,
+      "step": 5559
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.831608682670291e-06,
+      "loss": 3.3783,
+      "step": 5560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.829810844755687e-06,
+      "loss": 3.4457,
+      "step": 5561
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.828013352483599e-06,
+      "loss": 3.3727,
+      "step": 5562
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8262162061403104e-06,
+      "loss": 3.3755,
+      "step": 5563
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.82441940601205e-06,
+      "loss": 3.4334,
+      "step": 5564
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8226229523849914e-06,
+      "loss": 3.4095,
+      "step": 5565
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8208268455452525e-06,
+      "loss": 3.4868,
+      "step": 5566
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.819031085778899e-06,
+      "loss": 3.3912,
+      "step": 5567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8172356733719343e-06,
+      "loss": 3.3796,
+      "step": 5568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8154406086103115e-06,
+      "loss": 3.4929,
+      "step": 5569
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8136458917799303e-06,
+      "loss": 3.3672,
+      "step": 5570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8118515231666297e-06,
+      "loss": 3.4388,
+      "step": 5571
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.810057503056198e-06,
+      "loss": 3.3287,
+      "step": 5572
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.808263831734367e-06,
+      "loss": 3.4193,
+      "step": 5573
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8064705094868074e-06,
+      "loss": 3.328,
+      "step": 5574
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8046775365991414e-06,
+      "loss": 3.463,
+      "step": 5575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8028849133569326e-06,
+      "loss": 3.2949,
+      "step": 5576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8010926400456895e-06,
+      "loss": 3.4401,
+      "step": 5577
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7993007169508636e-06,
+      "loss": 3.4534,
+      "step": 5578
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.797509144357855e-06,
+      "loss": 3.4362,
+      "step": 5579
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.795717922551998e-06,
+      "loss": 3.443,
+      "step": 5580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.793927051818583e-06,
+      "loss": 3.3334,
+      "step": 5581
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.792136532442838e-06,
+      "loss": 3.4249,
+      "step": 5582
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.790346364709935e-06,
+      "loss": 3.3966,
+      "step": 5583
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7885565489049948e-06,
+      "loss": 3.3901,
+      "step": 5584
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7867670853130724e-06,
+      "loss": 3.4408,
+      "step": 5585
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.784977974219177e-06,
+      "loss": 3.3368,
+      "step": 5586
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.783189215908255e-06,
+      "loss": 3.3254,
+      "step": 5587
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.781400810665201e-06,
+      "loss": 3.3022,
+      "step": 5588
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7796127587748513e-06,
+      "loss": 3.4807,
+      "step": 5589
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7778250605219868e-06,
+      "loss": 3.3935,
+      "step": 5590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7760377161913284e-06,
+      "loss": 3.3298,
+      "step": 5591
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7742507260675446e-06,
+      "loss": 3.4769,
+      "step": 5592
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.772464090435247e-06,
+      "loss": 3.4204,
+      "step": 5593
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7706778095789905e-06,
+      "loss": 3.4161,
+      "step": 5594
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7688918837832723e-06,
+      "loss": 3.4036,
+      "step": 5595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.767106313332535e-06,
+      "loss": 3.3465,
+      "step": 5596
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7653210985111646e-06,
+      "loss": 3.4653,
+      "step": 5597
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.763536239603486e-06,
+      "loss": 3.4706,
+      "step": 5598
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.761751736893772e-06,
+      "loss": 3.4353,
+      "step": 5599
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7599675906662382e-06,
+      "loss": 3.4707,
+      "step": 5600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7581838012050422e-06,
+      "loss": 3.3455,
+      "step": 5601
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7564003687942853e-06,
+      "loss": 3.4884,
+      "step": 5602
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7546172937180144e-06,
+      "loss": 3.3489,
+      "step": 5603
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7528345762602125e-06,
+      "loss": 3.4633,
+      "step": 5604
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7510522167048103e-06,
+      "loss": 3.4872,
+      "step": 5605
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7492702153356852e-06,
+      "loss": 3.4106,
+      "step": 5606
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.747488572436652e-06,
+      "loss": 3.3281,
+      "step": 5607
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7457072882914717e-06,
+      "loss": 3.4317,
+      "step": 5608
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.743926363183843e-06,
+      "loss": 3.3805,
+      "step": 5609
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7421457973974126e-06,
+      "loss": 3.3822,
+      "step": 5610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.740365591215768e-06,
+      "loss": 3.4715,
+      "step": 5611
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7385857449224408e-06,
+      "loss": 3.3147,
+      "step": 5612
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7368062588009025e-06,
+      "loss": 3.4901,
+      "step": 5613
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7350271331345733e-06,
+      "loss": 3.426,
+      "step": 5614
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7332483682068057e-06,
+      "loss": 3.4181,
+      "step": 5615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7314699643009037e-06,
+      "loss": 3.5312,
+      "step": 5616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7296919217001105e-06,
+      "loss": 3.3719,
+      "step": 5617
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7279142406876126e-06,
+      "loss": 3.4931,
+      "step": 5618
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.726136921546537e-06,
+      "loss": 3.4452,
+      "step": 5619
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.724359964559958e-06,
+      "loss": 3.2697,
+      "step": 5620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.722583370010884e-06,
+      "loss": 3.4133,
+      "step": 5621
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.720807138182272e-06,
+      "loss": 3.4162,
+      "step": 5622
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7190312693570197e-06,
+      "loss": 3.2844,
+      "step": 5623
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7172557638179674e-06,
+      "loss": 3.4739,
+      "step": 5624
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.715480621847897e-06,
+      "loss": 3.337,
+      "step": 5625
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.713705843729532e-06,
+      "loss": 3.3658,
+      "step": 5626
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.711931429745541e-06,
+      "loss": 3.4742,
+      "step": 5627
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.710157380178527e-06,
+      "loss": 3.4617,
+      "step": 5628
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.708383695311043e-06,
+      "loss": 3.4782,
+      "step": 5629
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.706610375425578e-06,
+      "loss": 3.3949,
+      "step": 5630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7048374208045713e-06,
+      "loss": 3.4636,
+      "step": 5631
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7030648317303965e-06,
+      "loss": 3.2542,
+      "step": 5632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7012926084853687e-06,
+      "loss": 3.4649,
+      "step": 5633
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6995207513517484e-06,
+      "loss": 3.3677,
+      "step": 5634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6977492606117365e-06,
+      "loss": 3.4275,
+      "step": 5635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.695978136547476e-06,
+      "loss": 3.3802,
+      "step": 5636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.694207379441049e-06,
+      "loss": 3.5127,
+      "step": 5637
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.692436989574485e-06,
+      "loss": 3.425,
+      "step": 5638
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.690666967229747e-06,
+      "loss": 3.4538,
+      "step": 5639
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6888973126887447e-06,
+      "loss": 3.4167,
+      "step": 5640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6871280262333287e-06,
+      "loss": 3.2898,
+      "step": 5641
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6853591081452908e-06,
+      "loss": 3.4602,
+      "step": 5642
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6835905587063627e-06,
+      "loss": 3.3458,
+      "step": 5643
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.681822378198221e-06,
+      "loss": 3.3815,
+      "step": 5644
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.680054566902477e-06,
+      "loss": 3.39,
+      "step": 5645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6782871251006894e-06,
+      "loss": 3.3736,
+      "step": 5646
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6765200530743546e-06,
+      "loss": 3.381,
+      "step": 5647
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6747533511049127e-06,
+      "loss": 3.3012,
+      "step": 5648
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6729870194737427e-06,
+      "loss": 3.4641,
+      "step": 5649
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6712210584621656e-06,
+      "loss": 3.4111,
+      "step": 5650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6694554683514452e-06,
+      "loss": 3.4943,
+      "step": 5651
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6676902494227795e-06,
+      "loss": 3.4768,
+      "step": 5652
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6659254019573157e-06,
+      "loss": 3.4223,
+      "step": 5653
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6641609262361343e-06,
+      "loss": 3.4005,
+      "step": 5654
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6623968225402683e-06,
+      "loss": 3.3126,
+      "step": 5655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.660633091150676e-06,
+      "loss": 3.485,
+      "step": 5656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.658869732348267e-06,
+      "loss": 3.4083,
+      "step": 5657
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6571067464138883e-06,
+      "loss": 3.4062,
+      "step": 5658
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.655344133628328e-06,
+      "loss": 3.4472,
+      "step": 5659
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.653581894272315e-06,
+      "loss": 3.4911,
+      "step": 5660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.651820028626516e-06,
+      "loss": 3.3881,
+      "step": 5661
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6500585369715457e-06,
+      "loss": 3.4802,
+      "step": 5662
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6482974195879486e-06,
+      "loss": 3.5656,
+      "step": 5663
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6465366767562162e-06,
+      "loss": 3.5041,
+      "step": 5664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.64477630875678e-06,
+      "loss": 3.3604,
+      "step": 5665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6430163158700116e-06,
+      "loss": 3.4347,
+      "step": 5666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.641256698376222e-06,
+      "loss": 3.4422,
+      "step": 5667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.639497456555663e-06,
+      "loss": 3.3941,
+      "step": 5668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.637738590688524e-06,
+      "loss": 3.3593,
+      "step": 5669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.635980101054938e-06,
+      "loss": 3.465,
+      "step": 5670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.634221987934977e-06,
+      "loss": 3.4276,
+      "step": 5671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6324642516086523e-06,
+      "loss": 3.3241,
+      "step": 5672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6307068923559166e-06,
+      "loss": 3.4651,
+      "step": 5673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.628949910456663e-06,
+      "loss": 3.4471,
+      "step": 5674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6271933061907197e-06,
+      "loss": 3.358,
+      "step": 5675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6254370798378594e-06,
+      "loss": 3.5902,
+      "step": 5676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6236812316777942e-06,
+      "loss": 3.5348,
+      "step": 5677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.621925761990175e-06,
+      "loss": 3.3673,
+      "step": 5678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6201706710545925e-06,
+      "loss": 3.4036,
+      "step": 5679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6184159591505777e-06,
+      "loss": 3.4562,
+      "step": 5680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6166616265575995e-06,
+      "loss": 3.3985,
+      "step": 5681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6149076735550687e-06,
+      "loss": 3.362,
+      "step": 5682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.613154100422334e-06,
+      "loss": 3.4516,
+      "step": 5683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.611400907438685e-06,
+      "loss": 3.4011,
+      "step": 5684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6096480948833492e-06,
+      "loss": 3.496,
+      "step": 5685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.607895663035496e-06,
+      "loss": 3.4771,
+      "step": 5686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6061436121742285e-06,
+      "loss": 3.4645,
+      "step": 5687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6043919425785957e-06,
+      "loss": 3.4211,
+      "step": 5688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6026406545275823e-06,
+      "loss": 3.3648,
+      "step": 5689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.600889748300114e-06,
+      "loss": 3.3043,
+      "step": 5690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5991392241750536e-06,
+      "loss": 3.3604,
+      "step": 5691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5973890824312074e-06,
+      "loss": 3.349,
+      "step": 5692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5956393233473134e-06,
+      "loss": 3.4261,
+      "step": 5693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5938899472020545e-06,
+      "loss": 3.4332,
+      "step": 5694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.592140954274052e-06,
+      "loss": 3.4292,
+      "step": 5695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5903923448418635e-06,
+      "loss": 3.4466,
+      "step": 5696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.588644119183987e-06,
+      "loss": 3.3647,
+      "step": 5697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.586896277578864e-06,
+      "loss": 3.4634,
+      "step": 5698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.585148820304866e-06,
+      "loss": 3.5439,
+      "step": 5699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.583401747640307e-06,
+      "loss": 3.4537,
+      "step": 5700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.581655059863443e-06,
+      "loss": 3.3887,
+      "step": 5701
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.579908757252465e-06,
+      "loss": 3.4403,
+      "step": 5702
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5781628400855046e-06,
+      "loss": 3.4177,
+      "step": 5703
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5764173086406306e-06,
+      "loss": 3.3934,
+      "step": 5704
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5746721631958506e-06,
+      "loss": 3.4226,
+      "step": 5705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.572927404029112e-06,
+      "loss": 3.3591,
+      "step": 5706
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5711830314182996e-06,
+      "loss": 3.3893,
+      "step": 5707
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.569439045641236e-06,
+      "loss": 3.5643,
+      "step": 5708
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5676954469756856e-06,
+      "loss": 3.462,
+      "step": 5709
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.565952235699345e-06,
+      "loss": 3.5019,
+      "step": 5710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5642094120898535e-06,
+      "loss": 3.4388,
+      "step": 5711
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.562466976424789e-06,
+      "loss": 3.4284,
+      "step": 5712
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5607249289816656e-06,
+      "loss": 3.3013,
+      "step": 5713
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.558983270037937e-06,
+      "loss": 3.4198,
+      "step": 5714
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.557241999870993e-06,
+      "loss": 3.3912,
+      "step": 5715
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.555501118758167e-06,
+      "loss": 3.467,
+      "step": 5716
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.55376062697672e-06,
+      "loss": 3.4868,
+      "step": 5717
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5520205248038616e-06,
+      "loss": 3.4467,
+      "step": 5718
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.550280812516733e-06,
+      "loss": 3.3607,
+      "step": 5719
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.548541490392418e-06,
+      "loss": 3.4016,
+      "step": 5720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5468025587079324e-06,
+      "loss": 3.3109,
+      "step": 5721
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5450640177402376e-06,
+      "loss": 3.5145,
+      "step": 5722
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5433258677662224e-06,
+      "loss": 3.4703,
+      "step": 5723
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5415881090627227e-06,
+      "loss": 3.3127,
+      "step": 5724
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5398507419065073e-06,
+      "loss": 3.2527,
+      "step": 5725
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5381137665742846e-06,
+      "loss": 3.4047,
+      "step": 5726
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5363771833426993e-06,
+      "loss": 3.4682,
+      "step": 5727
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5346409924883343e-06,
+      "loss": 3.4153,
+      "step": 5728
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5329051942877103e-06,
+      "loss": 3.358,
+      "step": 5729
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.531169789017285e-06,
+      "loss": 3.4019,
+      "step": 5730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.529434776953453e-06,
+      "loss": 3.4097,
+      "step": 5731
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.527700158372548e-06,
+      "loss": 3.4537,
+      "step": 5732
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5259659335508412e-06,
+      "loss": 3.375,
+      "step": 5733
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5242321027645354e-06,
+      "loss": 3.3678,
+      "step": 5734
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.522498666289777e-06,
+      "loss": 3.3803,
+      "step": 5735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.520765624402649e-06,
+      "loss": 3.5011,
+      "step": 5736
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5190329773791684e-06,
+      "loss": 3.459,
+      "step": 5737
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.517300725495292e-06,
+      "loss": 3.3554,
+      "step": 5738
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.515568869026913e-06,
+      "loss": 3.3968,
+      "step": 5739
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.513837408249863e-06,
+      "loss": 3.4449,
+      "step": 5740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.512106343439905e-06,
+      "loss": 3.4327,
+      "step": 5741
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.510375674872745e-06,
+      "loss": 3.2489,
+      "step": 5742
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5086454028240226e-06,
+      "loss": 3.4526,
+      "step": 5743
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.506915527569318e-06,
+      "loss": 3.4631,
+      "step": 5744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.505186049384143e-06,
+      "loss": 3.3656,
+      "step": 5745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5034569685439523e-06,
+      "loss": 3.3565,
+      "step": 5746
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.50172828532413e-06,
+      "loss": 3.3347,
+      "step": 5747
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5000000000000015e-06,
+      "loss": 3.4511,
+      "step": 5748
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4982721128468283e-06,
+      "loss": 3.3128,
+      "step": 5749
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.496544624139809e-06,
+      "loss": 3.5406,
+      "step": 5750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4948175341540775e-06,
+      "loss": 3.5021,
+      "step": 5751
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4930908431647062e-06,
+      "loss": 3.3746,
+      "step": 5752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.491364551446697e-06,
+      "loss": 3.4808,
+      "step": 5753
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4896386592749994e-06,
+      "loss": 3.457,
+      "step": 5754
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4879131669244916e-06,
+      "loss": 3.3508,
+      "step": 5755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.486188074669989e-06,
+      "loss": 3.4492,
+      "step": 5756
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4844633827862474e-06,
+      "loss": 3.3589,
+      "step": 5757
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4827390915479515e-06,
+      "loss": 3.5348,
+      "step": 5758
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4810152012297285e-06,
+      "loss": 3.4222,
+      "step": 5759
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.479291712106139e-06,
+      "loss": 3.5802,
+      "step": 5760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.47756862445168e-06,
+      "loss": 3.522,
+      "step": 5761
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.475845938540786e-06,
+      "loss": 3.4598,
+      "step": 5762
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.474123654647828e-06,
+      "loss": 3.372,
+      "step": 5763
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.472401773047107e-06,
+      "loss": 3.4495,
+      "step": 5764
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.470680294012867e-06,
+      "loss": 3.4558,
+      "step": 5765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4689592178192843e-06,
+      "loss": 3.4417,
+      "step": 5766
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4672385447404727e-06,
+      "loss": 3.3293,
+      "step": 5767
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.465518275050481e-06,
+      "loss": 3.4763,
+      "step": 5768
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.463798409023293e-06,
+      "loss": 3.3376,
+      "step": 5769
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4620789469328312e-06,
+      "loss": 3.4799,
+      "step": 5770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.460359889052948e-06,
+      "loss": 3.3927,
+      "step": 5771
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4586412356574367e-06,
+      "loss": 3.4321,
+      "step": 5772
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.456922987020025e-06,
+      "loss": 3.2793,
+      "step": 5773
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.455205143414375e-06,
+      "loss": 3.4206,
+      "step": 5774
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4534877051140844e-06,
+      "loss": 3.4298,
+      "step": 5775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.45177067239269e-06,
+      "loss": 3.3831,
+      "step": 5776
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4500540455236537e-06,
+      "loss": 3.4356,
+      "step": 5777
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.448337824780386e-06,
+      "loss": 3.5553,
+      "step": 5778
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.446622010436226e-06,
+      "loss": 3.4011,
+      "step": 5779
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4449066027644473e-06,
+      "loss": 3.454,
+      "step": 5780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.443191602038262e-06,
+      "loss": 3.4223,
+      "step": 5781
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4414770085308125e-06,
+      "loss": 3.3512,
+      "step": 5782
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.439762822515181e-06,
+      "loss": 3.3976,
+      "step": 5783
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.438049044264382e-06,
+      "loss": 3.4059,
+      "step": 5784
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.436335674051368e-06,
+      "loss": 3.4953,
+      "step": 5785
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4346227121490233e-06,
+      "loss": 3.3987,
+      "step": 5786
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4329101588301713e-06,
+      "loss": 3.4581,
+      "step": 5787
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4311980143675635e-06,
+      "loss": 3.4218,
+      "step": 5788
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.429486279033892e-06,
+      "loss": 3.4257,
+      "step": 5789
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4277749531017824e-06,
+      "loss": 3.4759,
+      "step": 5790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4260640368437955e-06,
+      "loss": 3.409,
+      "step": 5791
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4243535305324244e-06,
+      "loss": 3.4027,
+      "step": 5792
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4226434344401023e-06,
+      "loss": 3.346,
+      "step": 5793
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4209337488391886e-06,
+      "loss": 3.3798,
+      "step": 5794
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4192244740019847e-06,
+      "loss": 3.4273,
+      "step": 5795
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4175156102007237e-06,
+      "loss": 3.459,
+      "step": 5796
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.415807157707574e-06,
+      "loss": 3.4054,
+      "step": 5797
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.414099116794637e-06,
+      "loss": 3.2663,
+      "step": 5798
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4123914877339504e-06,
+      "loss": 3.3106,
+      "step": 5799
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4106842707974874e-06,
+      "loss": 3.4139,
+      "step": 5800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4089774662571502e-06,
+      "loss": 3.4415,
+      "step": 5801
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4072710743847784e-06,
+      "loss": 3.4621,
+      "step": 5802
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4055650954521496e-06,
+      "loss": 3.3837,
+      "step": 5803
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4038595297309712e-06,
+      "loss": 3.3017,
+      "step": 5804
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4021543774928873e-06,
+      "loss": 3.391,
+      "step": 5805
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.400449639009471e-06,
+      "loss": 3.3763,
+      "step": 5806
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.398745314552236e-06,
+      "loss": 3.4782,
+      "step": 5807
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.397041404392626e-06,
+      "loss": 3.4935,
+      "step": 5808
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.395337908802021e-06,
+      "loss": 3.4608,
+      "step": 5809
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.393634828051733e-06,
+      "loss": 3.4639,
+      "step": 5810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3919321624130115e-06,
+      "loss": 3.4847,
+      "step": 5811
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3902299121570332e-06,
+      "loss": 3.3343,
+      "step": 5812
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.388528077554915e-06,
+      "loss": 3.3633,
+      "step": 5813
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.386826658877706e-06,
+      "loss": 3.3752,
+      "step": 5814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3851256563963866e-06,
+      "loss": 3.4141,
+      "step": 5815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.383425070381874e-06,
+      "loss": 3.3607,
+      "step": 5816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3817249011050194e-06,
+      "loss": 3.46,
+      "step": 5817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3800251488366016e-06,
+      "loss": 3.4159,
+      "step": 5818
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.378325813847341e-06,
+      "loss": 3.3833,
+      "step": 5819
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3766268964078857e-06,
+      "loss": 3.3768,
+      "step": 5820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.374928396788821e-06,
+      "loss": 3.4919,
+      "step": 5821
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3732303152606638e-06,
+      "loss": 3.3682,
+      "step": 5822
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.371532652093865e-06,
+      "loss": 3.4883,
+      "step": 5823
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3698354075588105e-06,
+      "loss": 3.3184,
+      "step": 5824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3681385819258133e-06,
+      "loss": 3.3975,
+      "step": 5825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.366442175465125e-06,
+      "loss": 3.3269,
+      "step": 5826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.364746188446933e-06,
+      "loss": 3.3707,
+      "step": 5827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.363050621141354e-06,
+      "loss": 3.4402,
+      "step": 5828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3613554738184347e-06,
+      "loss": 3.3769,
+      "step": 5829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3596607467481602e-06,
+      "loss": 3.3336,
+      "step": 5830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3579664402004477e-06,
+      "loss": 3.3758,
+      "step": 5831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.356272554445146e-06,
+      "loss": 3.3931,
+      "step": 5832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.354579089752037e-06,
+      "loss": 3.4198,
+      "step": 5833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3528860463908375e-06,
+      "loss": 3.4833,
+      "step": 5834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.351193424631196e-06,
+      "loss": 3.4113,
+      "step": 5835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.349501224742692e-06,
+      "loss": 3.4728,
+      "step": 5836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3478094469948393e-06,
+      "loss": 3.3989,
+      "step": 5837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3461180916570853e-06,
+      "loss": 3.4781,
+      "step": 5838
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.34442715899881e-06,
+      "loss": 3.4981,
+      "step": 5839
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3427366492893243e-06,
+      "loss": 3.4123,
+      "step": 5840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.341046562797876e-06,
+      "loss": 3.4232,
+      "step": 5841
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.339356899793638e-06,
+      "loss": 3.4449,
+      "step": 5842
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.337667660545723e-06,
+      "loss": 3.437,
+      "step": 5843
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3359788453231723e-06,
+      "loss": 3.356,
+      "step": 5844
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3342904543949614e-06,
+      "loss": 3.3766,
+      "step": 5845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.332602488029997e-06,
+      "loss": 3.4012,
+      "step": 5846
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.330914946497121e-06,
+      "loss": 3.3588,
+      "step": 5847
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3292278300651017e-06,
+      "loss": 3.3662,
+      "step": 5848
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3275411390026453e-06,
+      "loss": 3.4601,
+      "step": 5849
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.325854873578388e-06,
+      "loss": 3.4467,
+      "step": 5850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.324169034060898e-06,
+      "loss": 3.4167,
+      "step": 5851
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.32248362071868e-06,
+      "loss": 3.4015,
+      "step": 5852
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.320798633820163e-06,
+      "loss": 3.3834,
+      "step": 5853
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.319114073633712e-06,
+      "loss": 3.431,
+      "step": 5854
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.317429940427626e-06,
+      "loss": 3.4823,
+      "step": 5855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.315746234470134e-06,
+      "loss": 3.371,
+      "step": 5856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.314062956029397e-06,
+      "loss": 3.3969,
+      "step": 5857
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.312380105373508e-06,
+      "loss": 3.4486,
+      "step": 5858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.310697682770493e-06,
+      "loss": 3.4592,
+      "step": 5859
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.309015688488305e-06,
+      "loss": 3.4927,
+      "step": 5860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.307334122794836e-06,
+      "loss": 3.4703,
+      "step": 5861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3056529859579047e-06,
+      "loss": 3.4379,
+      "step": 5862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.303972278245264e-06,
+      "loss": 3.3472,
+      "step": 5863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3022919999245964e-06,
+      "loss": 3.3302,
+      "step": 5864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3006121512635198e-06,
+      "loss": 3.4121,
+      "step": 5865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2989327325295776e-06,
+      "loss": 3.3966,
+      "step": 5866
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.297253743990249e-06,
+      "loss": 3.4346,
+      "step": 5867
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2955751859129437e-06,
+      "loss": 3.3285,
+      "step": 5868
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.293897058565004e-06,
+      "loss": 3.4636,
+      "step": 5869
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2922193622137017e-06,
+      "loss": 3.4368,
+      "step": 5870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.290542097126243e-06,
+      "loss": 3.3192,
+      "step": 5871
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.28886526356976e-06,
+      "loss": 3.4498,
+      "step": 5872
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2871888618113195e-06,
+      "loss": 3.5525,
+      "step": 5873
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.285512892117921e-06,
+      "loss": 3.4348,
+      "step": 5874
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.283837354756493e-06,
+      "loss": 3.3656,
+      "step": 5875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.282162249993895e-06,
+      "loss": 3.3537,
+      "step": 5876
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.280487578096919e-06,
+      "loss": 3.4356,
+      "step": 5877
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2788133393322863e-06,
+      "loss": 3.5437,
+      "step": 5878
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2771395339666513e-06,
+      "loss": 3.3996,
+      "step": 5879
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2754661622665976e-06,
+      "loss": 3.3688,
+      "step": 5880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2737932244986404e-06,
+      "loss": 3.3551,
+      "step": 5881
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2721207209292286e-06,
+      "loss": 3.4588,
+      "step": 5882
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2704486518247343e-06,
+      "loss": 3.328,
+      "step": 5883
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2687770174514674e-06,
+      "loss": 3.4623,
+      "step": 5884
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2671058180756667e-06,
+      "loss": 3.3284,
+      "step": 5885
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2654350539635008e-06,
+      "loss": 3.358,
+      "step": 5886
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.26376472538107e-06,
+      "loss": 3.4147,
+      "step": 5887
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.262094832594405e-06,
+      "loss": 3.4974,
+      "step": 5888
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.260425375869469e-06,
+      "loss": 3.3222,
+      "step": 5889
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2587563554721493e-06,
+      "loss": 3.3744,
+      "step": 5890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2570877716682706e-06,
+      "loss": 3.4357,
+      "step": 5891
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2554196247235843e-06,
+      "loss": 3.426,
+      "step": 5892
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2537519149037746e-06,
+      "loss": 3.388,
+      "step": 5893
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2520846424744545e-06,
+      "loss": 3.4863,
+      "step": 5894
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2504178077011705e-06,
+      "loss": 3.4363,
+      "step": 5895
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.248751410849392e-06,
+      "loss": 3.4779,
+      "step": 5896
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2470854521845255e-06,
+      "loss": 3.5101,
+      "step": 5897
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.245419931971905e-06,
+      "loss": 3.4253,
+      "step": 5898
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.243754850476796e-06,
+      "loss": 3.4182,
+      "step": 5899
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.242090207964393e-06,
+      "loss": 3.4421,
+      "step": 5900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2404260046998212e-06,
+      "loss": 3.5597,
+      "step": 5901
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2387622409481347e-06,
+      "loss": 3.3567,
+      "step": 5902
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2370989169743196e-06,
+      "loss": 3.4489,
+      "step": 5903
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.23543603304329e-06,
+      "loss": 3.3889,
+      "step": 5904
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2337735894198915e-06,
+      "loss": 3.4391,
+      "step": 5905
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2321115863689003e-06,
+      "loss": 3.4105,
+      "step": 5906
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2304500241550164e-06,
+      "loss": 3.3776,
+      "step": 5907
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.228788903042877e-06,
+      "loss": 3.4831,
+      "step": 5908
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.227128223297046e-06,
+      "loss": 3.3531,
+      "step": 5909
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.225467985182017e-06,
+      "loss": 3.3884,
+      "step": 5910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2238081889622136e-06,
+      "loss": 3.4995,
+      "step": 5911
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2221488349019903e-06,
+      "loss": 3.383,
+      "step": 5912
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2204899232656263e-06,
+      "loss": 3.4181,
+      "step": 5913
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.218831454317336e-06,
+      "loss": 3.4354,
+      "step": 5914
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.217173428321261e-06,
+      "loss": 3.4445,
+      "step": 5915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2155158455414717e-06,
+      "loss": 3.4062,
+      "step": 5916
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2138587062419685e-06,
+      "loss": 3.43,
+      "step": 5917
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2122020106866825e-06,
+      "loss": 3.4939,
+      "step": 5918
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2105457591394745e-06,
+      "loss": 3.411,
+      "step": 5919
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2088899518641284e-06,
+      "loss": 3.4364,
+      "step": 5920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2072345891243646e-06,
+      "loss": 3.4791,
+      "step": 5921
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2055796711838296e-06,
+      "loss": 3.4151,
+      "step": 5922
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2039251983061e-06,
+      "loss": 3.3434,
+      "step": 5923
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.20227117075468e-06,
+      "loss": 3.5508,
+      "step": 5924
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2006175887930054e-06,
+      "loss": 3.4669,
+      "step": 5925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1989644526844385e-06,
+      "loss": 3.4546,
+      "step": 5926
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1973117626922714e-06,
+      "loss": 3.5082,
+      "step": 5927
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1956595190797264e-06,
+      "loss": 3.3952,
+      "step": 5928
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1940077221099524e-06,
+      "loss": 3.3419,
+      "step": 5929
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1923563720460317e-06,
+      "loss": 3.4771,
+      "step": 5930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.190705469150967e-06,
+      "loss": 3.4216,
+      "step": 5931
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.189055013687698e-06,
+      "loss": 3.2812,
+      "step": 5932
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1874050059190895e-06,
+      "loss": 3.4117,
+      "step": 5933
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1857554461079356e-06,
+      "loss": 3.4305,
+      "step": 5934
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.184106334516959e-06,
+      "loss": 3.4132,
+      "step": 5935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1824576714088126e-06,
+      "loss": 3.3266,
+      "step": 5936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.180809457046073e-06,
+      "loss": 3.4355,
+      "step": 5937
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1791616916912508e-06,
+      "loss": 3.4,
+      "step": 5938
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1775143756067823e-06,
+      "loss": 3.4461,
+      "step": 5939
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.175867509055033e-06,
+      "loss": 3.3187,
+      "step": 5940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.174221092298297e-06,
+      "loss": 3.3934,
+      "step": 5941
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1725751255987964e-06,
+      "loss": 3.3092,
+      "step": 5942
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1709296092186827e-06,
+      "loss": 3.5284,
+      "step": 5943
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1692845434200323e-06,
+      "loss": 3.4205,
+      "step": 5944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1676399284648523e-06,
+      "loss": 3.3854,
+      "step": 5945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1659957646150797e-06,
+      "loss": 3.4749,
+      "step": 5946
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.164352052132577e-06,
+      "loss": 3.3788,
+      "step": 5947
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.162708791279135e-06,
+      "loss": 3.4098,
+      "step": 5948
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1610659823164744e-06,
+      "loss": 3.4953,
+      "step": 5949
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1594236255062416e-06,
+      "loss": 3.4392,
+      "step": 5950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1577817211100126e-06,
+      "loss": 3.4426,
+      "step": 5951
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1561402693892907e-06,
+      "loss": 3.4248,
+      "step": 5952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.154499270605508e-06,
+      "loss": 3.3564,
+      "step": 5953
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.152858725020025e-06,
+      "loss": 3.4342,
+      "step": 5954
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1512186328941247e-06,
+      "loss": 3.4847,
+      "step": 5955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.149578994489024e-06,
+      "loss": 3.3854,
+      "step": 5956
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.147939810065866e-06,
+      "loss": 3.4149,
+      "step": 5957
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1463010798857202e-06,
+      "loss": 3.498,
+      "step": 5958
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1446628042095845e-06,
+      "loss": 3.4107,
+      "step": 5959
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1430249832983874e-06,
+      "loss": 3.3989,
+      "step": 5960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.141387617412977e-06,
+      "loss": 3.4532,
+      "step": 5961
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1397507068141367e-06,
+      "loss": 3.4072,
+      "step": 5962
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1381142517625736e-06,
+      "loss": 3.4461,
+      "step": 5963
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.136478252518924e-06,
+      "loss": 3.3564,
+      "step": 5964
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1348427093437506e-06,
+      "loss": 3.3997,
+      "step": 5965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1332076224975455e-06,
+      "loss": 3.3456,
+      "step": 5966
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.131572992240723e-06,
+      "loss": 3.4305,
+      "step": 5967
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1299388188336296e-06,
+      "loss": 3.4234,
+      "step": 5968
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.128305102536537e-06,
+      "loss": 3.4534,
+      "step": 5969
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1266718436096455e-06,
+      "loss": 3.2991,
+      "step": 5970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.125039042313081e-06,
+      "loss": 3.4174,
+      "step": 5971
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1234066989068972e-06,
+      "loss": 3.3439,
+      "step": 5972
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1217748136510767e-06,
+      "loss": 3.4269,
+      "step": 5973
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1201433868055223e-06,
+      "loss": 3.3959,
+      "step": 5974
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1185124186300727e-06,
+      "loss": 3.4145,
+      "step": 5975
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.116881909384489e-06,
+      "loss": 3.3783,
+      "step": 5976
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1152518593284593e-06,
+      "loss": 3.356,
+      "step": 5977
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.113622268721601e-06,
+      "loss": 3.3006,
+      "step": 5978
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.111993137823453e-06,
+      "loss": 3.4486,
+      "step": 5979
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1103644668934857e-06,
+      "loss": 3.3511,
+      "step": 5980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.108736256191095e-06,
+      "loss": 3.4277,
+      "step": 5981
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1071085059756036e-06,
+      "loss": 3.3151,
+      "step": 5982
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1054812165062606e-06,
+      "loss": 3.4659,
+      "step": 5983
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.103854388042243e-06,
+      "loss": 3.4543,
+      "step": 5984
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.10222802084265e-06,
+      "loss": 3.4484,
+      "step": 5985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1006021151665123e-06,
+      "loss": 3.4787,
+      "step": 5986
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0989766712727848e-06,
+      "loss": 3.4558,
+      "step": 5987
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0973516894203507e-06,
+      "loss": 3.5359,
+      "step": 5988
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0957271698680164e-06,
+      "loss": 3.4727,
+      "step": 5989
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.09410311287452e-06,
+      "loss": 3.3762,
+      "step": 5990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.092479518698517e-06,
+      "loss": 3.4525,
+      "step": 5991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.090856387598598e-06,
+      "loss": 3.5198,
+      "step": 5992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0892337198332757e-06,
+      "loss": 3.3597,
+      "step": 5993
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.08761151566099e-06,
+      "loss": 3.3821,
+      "step": 5994
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0859897753401065e-06,
+      "loss": 3.4582,
+      "step": 5995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0843684991289177e-06,
+      "loss": 3.3416,
+      "step": 5996
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0827476872856427e-06,
+      "loss": 3.288,
+      "step": 5997
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0811273400684224e-06,
+      "loss": 3.3943,
+      "step": 5998
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.079507457735327e-06,
+      "loss": 3.3613,
+      "step": 5999
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0778880405443557e-06,
+      "loss": 3.3803,
+      "step": 6000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0762690887534307e-06,
+      "loss": 3.4343,
+      "step": 6001
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0746506026203974e-06,
+      "loss": 3.3609,
+      "step": 6002
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.073032582403029e-06,
+      "loss": 3.4057,
+      "step": 6003
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.071415028359026e-06,
+      "loss": 3.3014,
+      "step": 6004
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.069797940746014e-06,
+      "loss": 3.4539,
+      "step": 6005
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.068181319821544e-06,
+      "loss": 3.3964,
+      "step": 6006
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0665651658430913e-06,
+      "loss": 3.4791,
+      "step": 6007
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.064949479068062e-06,
+      "loss": 3.3678,
+      "step": 6008
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.063334259753778e-06,
+      "loss": 3.3952,
+      "step": 6009
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0617195081574966e-06,
+      "loss": 3.4012,
+      "step": 6010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.060105224536395e-06,
+      "loss": 3.4314,
+      "step": 6011
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.058491409147579e-06,
+      "loss": 3.3545,
+      "step": 6012
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0568780622480773e-06,
+      "loss": 3.4605,
+      "step": 6013
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0552651840948473e-06,
+      "loss": 3.3663,
+      "step": 6014
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0536527749447656e-06,
+      "loss": 3.4481,
+      "step": 6015
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.05204083505464e-06,
+      "loss": 3.3401,
+      "step": 6016
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0504293646812014e-06,
+      "loss": 3.4929,
+      "step": 6017
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.048818364081106e-06,
+      "loss": 3.4554,
+      "step": 6018
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0472078335109358e-06,
+      "loss": 3.4757,
+      "step": 6019
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.045597773227199e-06,
+      "loss": 3.3687,
+      "step": 6020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0439881834863234e-06,
+      "loss": 3.2913,
+      "step": 6021
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.042379064544668e-06,
+      "loss": 3.4873,
+      "step": 6022
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0407704166585116e-06,
+      "loss": 3.3095,
+      "step": 6023
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0391622400840665e-06,
+      "loss": 3.454,
+      "step": 6024
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.037554535077463e-06,
+      "loss": 3.5091,
+      "step": 6025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.035947301894754e-06,
+      "loss": 3.3282,
+      "step": 6026
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0343405407919235e-06,
+      "loss": 3.3611,
+      "step": 6027
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.032734252024877e-06,
+      "loss": 3.3648,
+      "step": 6028
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0311284358494464e-06,
+      "loss": 3.4917,
+      "step": 6029
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0295230925213866e-06,
+      "loss": 3.5154,
+      "step": 6030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0279182222963784e-06,
+      "loss": 3.3413,
+      "step": 6031
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0263138254300285e-06,
+      "loss": 3.4319,
+      "step": 6032
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0247099021778634e-06,
+      "loss": 3.3299,
+      "step": 6033
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.023106452795339e-06,
+      "loss": 3.4839,
+      "step": 6034
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0215034775378336e-06,
+      "loss": 3.5288,
+      "step": 6035
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.019900976660651e-06,
+      "loss": 3.3913,
+      "step": 6036
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0182989504190187e-06,
+      "loss": 3.4433,
+      "step": 6037
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.016697399068091e-06,
+      "loss": 3.3819,
+      "step": 6038
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0150963228629398e-06,
+      "loss": 3.5773,
+      "step": 6039
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.013495722058569e-06,
+      "loss": 3.3145,
+      "step": 6040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0118955969099034e-06,
+      "loss": 3.3646,
+      "step": 6041
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0102959476717914e-06,
+      "loss": 3.3682,
+      "step": 6042
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0086967745990076e-06,
+      "loss": 3.3633,
+      "step": 6043
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0070980779462513e-06,
+      "loss": 3.3864,
+      "step": 6044
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.00549985796814e-06,
+      "loss": 3.3652,
+      "step": 6045
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0039021149192224e-06,
+      "loss": 3.5119,
+      "step": 6046
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0023048490539674e-06,
+      "loss": 3.4884,
+      "step": 6047
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0007080606267686e-06,
+      "loss": 3.3494,
+      "step": 6048
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9991117498919484e-06,
+      "loss": 3.3756,
+      "step": 6049
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.997515917103743e-06,
+      "loss": 3.4267,
+      "step": 6050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9959205625163196e-06,
+      "loss": 3.4111,
+      "step": 6051
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.994325686383769e-06,
+      "loss": 3.3721,
+      "step": 6052
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.992731288960104e-06,
+      "loss": 3.4044,
+      "step": 6053
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9911373704992617e-06,
+      "loss": 3.4138,
+      "step": 6054
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.989543931255104e-06,
+      "loss": 3.4728,
+      "step": 6055
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.987950971481413e-06,
+      "loss": 3.4461,
+      "step": 6056
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9863584914318975e-06,
+      "loss": 3.4254,
+      "step": 6057
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9847664913601895e-06,
+      "loss": 3.3994,
+      "step": 6058
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.983174971519845e-06,
+      "loss": 3.5164,
+      "step": 6059
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.981583932164342e-06,
+      "loss": 3.4269,
+      "step": 6060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.979993373547084e-06,
+      "loss": 3.4485,
+      "step": 6061
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9784032959213967e-06,
+      "loss": 3.47,
+      "step": 6062
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.976813699540526e-06,
+      "loss": 3.4287,
+      "step": 6063
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.975224584657648e-06,
+      "loss": 3.4481,
+      "step": 6064
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9736359515258564e-06,
+      "loss": 3.3952,
+      "step": 6065
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.972047800398171e-06,
+      "loss": 3.4261,
+      "step": 6066
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9704601315275333e-06,
+      "loss": 3.5245,
+      "step": 6067
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9688729451668116e-06,
+      "loss": 3.3749,
+      "step": 6068
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9672862415687904e-06,
+      "loss": 3.4357,
+      "step": 6069
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9657000209861837e-06,
+      "loss": 3.4734,
+      "step": 6070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9641142836716258e-06,
+      "loss": 3.3436,
+      "step": 6071
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9625290298776727e-06,
+      "loss": 3.4194,
+      "step": 6072
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9609442598568105e-06,
+      "loss": 3.4088,
+      "step": 6073
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9593599738614382e-06,
+      "loss": 3.4038,
+      "step": 6074
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.957776172143884e-06,
+      "loss": 3.3232,
+      "step": 6075
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.956192854956397e-06,
+      "loss": 3.4805,
+      "step": 6076
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9546100225511495e-06,
+      "loss": 3.481,
+      "step": 6077
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9530276751802375e-06,
+      "loss": 3.393,
+      "step": 6078
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9514458130956797e-06,
+      "loss": 3.381,
+      "step": 6079
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9498644365494147e-06,
+      "loss": 3.4798,
+      "step": 6080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.948283545793305e-06,
+      "loss": 3.3683,
+      "step": 6081
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9467031410791393e-06,
+      "loss": 3.3748,
+      "step": 6082
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.945123222658625e-06,
+      "loss": 3.5485,
+      "step": 6083
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.943543790783392e-06,
+      "loss": 3.3473,
+      "step": 6084
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.941964845704997e-06,
+      "loss": 3.4602,
+      "step": 6085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9403863876749124e-06,
+      "loss": 3.277,
+      "step": 6086
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9388084169445375e-06,
+      "loss": 3.478,
+      "step": 6087
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9372309337651945e-06,
+      "loss": 3.4344,
+      "step": 6088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9356539383881253e-06,
+      "loss": 3.3894,
+      "step": 6089
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9340774310644963e-06,
+      "loss": 3.3782,
+      "step": 6090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9325014120453948e-06,
+      "loss": 3.3287,
+      "step": 6091
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9309258815818326e-06,
+      "loss": 3.3422,
+      "step": 6092
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929350839924739e-06,
+      "loss": 3.4023,
+      "step": 6093
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.927776287324969e-06,
+      "loss": 3.4641,
+      "step": 6094
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9262022240333e-06,
+      "loss": 3.3108,
+      "step": 6095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9246286503004296e-06,
+      "loss": 3.4846,
+      "step": 6096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9230555663769794e-06,
+      "loss": 3.3532,
+      "step": 6097
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9214829725134915e-06,
+      "loss": 3.3922,
+      "step": 6098
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9199108689604296e-06,
+      "loss": 3.3977,
+      "step": 6099
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9183392559681812e-06,
+      "loss": 3.4422,
+      "step": 6100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.916768133787054e-06,
+      "loss": 3.3139,
+      "step": 6101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9151975026672785e-06,
+      "loss": 3.4238,
+      "step": 6102
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.913627362859008e-06,
+      "loss": 3.3636,
+      "step": 6103
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9120577146123125e-06,
+      "loss": 3.3722,
+      "step": 6104
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.910488558177189e-06,
+      "loss": 3.4161,
+      "step": 6105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9089198938035557e-06,
+      "loss": 3.4481,
+      "step": 6106
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.90735172174125e-06,
+      "loss": 3.469,
+      "step": 6107
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.905784042240032e-06,
+      "loss": 3.4371,
+      "step": 6108
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9042168555495854e-06,
+      "loss": 3.4835,
+      "step": 6109
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.902650161919511e-06,
+      "loss": 3.51,
+      "step": 6110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9010839615993343e-06,
+      "loss": 3.4263,
+      "step": 6111
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8995182548385015e-06,
+      "loss": 3.3653,
+      "step": 6112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8979530418863807e-06,
+      "loss": 3.4043,
+      "step": 6113
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8963883229922603e-06,
+      "loss": 3.5189,
+      "step": 6114
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8948240984053512e-06,
+      "loss": 3.337,
+      "step": 6115
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8932603683747858e-06,
+      "loss": 3.3405,
+      "step": 6116
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8916971331496143e-06,
+      "loss": 3.406,
+      "step": 6117
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.890134392978812e-06,
+      "loss": 3.4942,
+      "step": 6118
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8885721481112734e-06,
+      "loss": 3.3966,
+      "step": 6119
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8870103987958155e-06,
+      "loss": 3.3948,
+      "step": 6120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8854491452811758e-06,
+      "loss": 3.3412,
+      "step": 6121
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8838883878160114e-06,
+      "loss": 3.439,
+      "step": 6122
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.882328126648903e-06,
+      "loss": 3.3399,
+      "step": 6123
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8807683620283496e-06,
+      "loss": 3.3613,
+      "step": 6124
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8792090942027734e-06,
+      "loss": 3.5001,
+      "step": 6125
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8776503234205162e-06,
+      "loss": 3.3659,
+      "step": 6126
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8760920499298425e-06,
+      "loss": 3.4431,
+      "step": 6127
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8745342739789324e-06,
+      "loss": 3.4329,
+      "step": 6128
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8729769958158922e-06,
+      "loss": 3.4029,
+      "step": 6129
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8714202156887474e-06,
+      "loss": 3.2948,
+      "step": 6130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8698639338454433e-06,
+      "loss": 3.3596,
+      "step": 6131
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8683081505338468e-06,
+      "loss": 3.4381,
+      "step": 6132
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8667528660017463e-06,
+      "loss": 3.3821,
+      "step": 6133
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8651980804968466e-06,
+      "loss": 3.3794,
+      "step": 6134
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8636437942667773e-06,
+      "loss": 3.3936,
+      "step": 6135
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.862090007559087e-06,
+      "loss": 3.3195,
+      "step": 6136
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.860536720621245e-06,
+      "loss": 3.4523,
+      "step": 6137
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8589839337006404e-06,
+      "loss": 3.44,
+      "step": 6138
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8574316470445852e-06,
+      "loss": 3.4033,
+      "step": 6139
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8558798609003053e-06,
+      "loss": 3.3127,
+      "step": 6140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8543285755149543e-06,
+      "loss": 3.3711,
+      "step": 6141
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.852777791135602e-06,
+      "loss": 3.4249,
+      "step": 6142
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.851227508009239e-06,
+      "loss": 3.3567,
+      "step": 6143
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8496777263827775e-06,
+      "loss": 3.4534,
+      "step": 6144
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.848128446503048e-06,
+      "loss": 3.3482,
+      "step": 6145
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8465796686168013e-06,
+      "loss": 3.4522,
+      "step": 6146
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8450313929707098e-06,
+      "loss": 3.4138,
+      "step": 6147
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8434836198113642e-06,
+      "loss": 3.4443,
+      "step": 6148
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.841936349385276e-06,
+      "loss": 3.5104,
+      "step": 6149
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8403895819388767e-06,
+      "loss": 3.2815,
+      "step": 6150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.838843317718519e-06,
+      "loss": 3.362,
+      "step": 6151
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8372975569704698e-06,
+      "loss": 3.3805,
+      "step": 6152
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8357522999409227e-06,
+      "loss": 3.4807,
+      "step": 6153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.834207546875988e-06,
+      "loss": 3.4927,
+      "step": 6154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8326632980216952e-06,
+      "loss": 3.3687,
+      "step": 6155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8311195536239956e-06,
+      "loss": 3.3258,
+      "step": 6156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8295763139287597e-06,
+      "loss": 3.4243,
+      "step": 6157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8280335791817733e-06,
+      "loss": 3.415,
+      "step": 6158
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8264913496287478e-06,
+      "loss": 3.3666,
+      "step": 6159
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8249496255153105e-06,
+      "loss": 3.3599,
+      "step": 6160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8234084070870107e-06,
+      "loss": 3.3176,
+      "step": 6161
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8218676945893143e-06,
+      "loss": 3.4493,
+      "step": 6162
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8203274882676103e-06,
+      "loss": 3.4068,
+      "step": 6163
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8187877883672024e-06,
+      "loss": 3.5519,
+      "step": 6164
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8172485951333164e-06,
+      "loss": 3.4091,
+      "step": 6165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.815709908811098e-06,
+      "loss": 3.4767,
+      "step": 6166
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8141717296456108e-06,
+      "loss": 3.3092,
+      "step": 6167
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8126340578818386e-06,
+      "loss": 3.3183,
+      "step": 6168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8110968937646833e-06,
+      "loss": 3.4049,
+      "step": 6169
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8095602375389686e-06,
+      "loss": 3.3609,
+      "step": 6170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8080240894494305e-06,
+      "loss": 3.5296,
+      "step": 6171
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8064884497407332e-06,
+      "loss": 3.4712,
+      "step": 6172
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8049533186574546e-06,
+      "loss": 3.4039,
+      "step": 6173
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8034186964440937e-06,
+      "loss": 3.3407,
+      "step": 6174
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8018845833450638e-06,
+      "loss": 3.3365,
+      "step": 6175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8003509796047024e-06,
+      "loss": 3.4561,
+      "step": 6176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7988178854672645e-06,
+      "loss": 3.3272,
+      "step": 6177
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.797285301176923e-06,
+      "loss": 3.3566,
+      "step": 6178
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7957532269777712e-06,
+      "loss": 3.3771,
+      "step": 6179
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7942216631138188e-06,
+      "loss": 3.4243,
+      "step": 6180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.792690609828997e-06,
+      "loss": 3.3333,
+      "step": 6181
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.791160067367152e-06,
+      "loss": 3.4106,
+      "step": 6182
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.789630035972052e-06,
+      "loss": 3.3199,
+      "step": 6183
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7881005158873826e-06,
+      "loss": 3.4125,
+      "step": 6184
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.786571507356748e-06,
+      "loss": 3.4269,
+      "step": 6185
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7850430106236704e-06,
+      "loss": 3.397,
+      "step": 6186
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7835150259315937e-06,
+      "loss": 3.4459,
+      "step": 6187
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7819875535238739e-06,
+      "loss": 3.4859,
+      "step": 6188
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7804605936437898e-06,
+      "loss": 3.4976,
+      "step": 6189
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7789341465345388e-06,
+      "loss": 3.3548,
+      "step": 6190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7774082124392356e-06,
+      "loss": 3.4412,
+      "step": 6191
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.775882791600913e-06,
+      "loss": 3.5107,
+      "step": 6192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7743578842625237e-06,
+      "loss": 3.3421,
+      "step": 6193
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7728334906669342e-06,
+      "loss": 3.4514,
+      "step": 6194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7713096110569328e-06,
+      "loss": 3.3884,
+      "step": 6195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7697862456752273e-06,
+      "loss": 3.5519,
+      "step": 6196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7682633947644412e-06,
+      "loss": 3.5243,
+      "step": 6197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766741058567118e-06,
+      "loss": 3.5064,
+      "step": 6198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7652192373257137e-06,
+      "loss": 3.5227,
+      "step": 6199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7636979312826085e-06,
+      "loss": 3.4843,
+      "step": 6200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7621771406800981e-06,
+      "loss": 3.4206,
+      "step": 6201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7606568657603968e-06,
+      "loss": 3.3175,
+      "step": 6202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.759137106765636e-06,
+      "loss": 3.5166,
+      "step": 6203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.757617863937865e-06,
+      "loss": 3.4811,
+      "step": 6204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7560991375190533e-06,
+      "loss": 3.3761,
+      "step": 6205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7545809277510828e-06,
+      "loss": 3.3988,
+      "step": 6206
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.753063234875757e-06,
+      "loss": 3.4816,
+      "step": 6207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7515460591347972e-06,
+      "loss": 3.2953,
+      "step": 6208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7500294007698415e-06,
+      "loss": 3.439,
+      "step": 6209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7485132600224458e-06,
+      "loss": 3.4193,
+      "step": 6210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.746997637134084e-06,
+      "loss": 3.4421,
+      "step": 6211
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.745482532346145e-06,
+      "loss": 3.4895,
+      "step": 6212
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7439679458999376e-06,
+      "loss": 3.3573,
+      "step": 6213
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7424538780366884e-06,
+      "loss": 3.4878,
+      "step": 6214
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7409403289975396e-06,
+      "loss": 3.3975,
+      "step": 6215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7394272990235529e-06,
+      "loss": 3.4771,
+      "step": 6216
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7379147883557068e-06,
+      "loss": 3.3078,
+      "step": 6217
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7364027972348936e-06,
+      "loss": 3.503,
+      "step": 6218
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7348913259019267e-06,
+      "loss": 3.3324,
+      "step": 6219
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7333803745975353e-06,
+      "loss": 3.3972,
+      "step": 6220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.731869943562368e-06,
+      "loss": 3.3105,
+      "step": 6221
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7303600330369897e-06,
+      "loss": 3.3829,
+      "step": 6222
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7288506432618779e-06,
+      "loss": 3.4532,
+      "step": 6223
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7273417744774323e-06,
+      "loss": 3.4363,
+      "step": 6224
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7258334269239674e-06,
+      "loss": 3.38,
+      "step": 6225
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7243256008417164e-06,
+      "loss": 3.3537,
+      "step": 6226
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7228182964708274e-06,
+      "loss": 3.3356,
+      "step": 6227
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7213115140513687e-06,
+      "loss": 3.3782,
+      "step": 6228
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.719805253823319e-06,
+      "loss": 3.3774,
+      "step": 6229
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7182995160265798e-06,
+      "loss": 3.4767,
+      "step": 6230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.716794300900968e-06,
+      "loss": 3.3354,
+      "step": 6231
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7152896086862164e-06,
+      "loss": 3.4251,
+      "step": 6232
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.713785439621975e-06,
+      "loss": 3.4073,
+      "step": 6233
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7122817939478104e-06,
+      "loss": 3.4189,
+      "step": 6234
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7107786719032071e-06,
+      "loss": 3.4276,
+      "step": 6235
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7092760737275626e-06,
+      "loss": 3.4242,
+      "step": 6236
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7077739996601938e-06,
+      "loss": 3.429,
+      "step": 6237
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7062724499403338e-06,
+      "loss": 3.4154,
+      "step": 6238
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7047714248071329e-06,
+      "loss": 3.3461,
+      "step": 6239
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7032709244996559e-06,
+      "loss": 3.3795,
+      "step": 6240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7017709492568874e-06,
+      "loss": 3.3865,
+      "step": 6241
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7002714993177222e-06,
+      "loss": 3.3352,
+      "step": 6242
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6987725749209772e-06,
+      "loss": 3.476,
+      "step": 6243
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6972741763053835e-06,
+      "loss": 3.4253,
+      "step": 6244
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.695776303709587e-06,
+      "loss": 3.4531,
+      "step": 6245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6942789573721569e-06,
+      "loss": 3.426,
+      "step": 6246
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6927821375315672e-06,
+      "loss": 3.3926,
+      "step": 6247
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6912858444262165e-06,
+      "loss": 3.44,
+      "step": 6248
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6897900782944171e-06,
+      "loss": 3.4342,
+      "step": 6249
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6882948393743964e-06,
+      "loss": 3.3622,
+      "step": 6250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.686800127904299e-06,
+      "loss": 3.3473,
+      "step": 6251
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.685305944122187e-06,
+      "loss": 3.4307,
+      "step": 6252
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6838122882660335e-06,
+      "loss": 3.4935,
+      "step": 6253
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6823191605737316e-06,
+      "loss": 3.3843,
+      "step": 6254
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6808265612830904e-06,
+      "loss": 3.2744,
+      "step": 6255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6793344906318327e-06,
+      "loss": 3.4016,
+      "step": 6256
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6778429488575993e-06,
+      "loss": 3.4186,
+      "step": 6257
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6763519361979458e-06,
+      "loss": 3.481,
+      "step": 6258
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.674861452890341e-06,
+      "loss": 3.2869,
+      "step": 6259
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6733714991721738e-06,
+      "loss": 3.5308,
+      "step": 6260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6718820752807462e-06,
+      "loss": 3.415,
+      "step": 6261
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6703931814532764e-06,
+      "loss": 3.4184,
+      "step": 6262
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6689048179268984e-06,
+      "loss": 3.3502,
+      "step": 6263
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6674169849386606e-06,
+      "loss": 3.3792,
+      "step": 6264
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.66592968272553e-06,
+      "loss": 3.3482,
+      "step": 6265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6644429115243837e-06,
+      "loss": 3.3847,
+      "step": 6266
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6629566715720192e-06,
+      "loss": 3.4318,
+      "step": 6267
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6614709631051461e-06,
+      "loss": 3.4234,
+      "step": 6268
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6599857863603908e-06,
+      "loss": 3.389,
+      "step": 6269
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.658501141574299e-06,
+      "loss": 3.4007,
+      "step": 6270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6570170289833232e-06,
+      "loss": 3.2688,
+      "step": 6271
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6555334488238372e-06,
+      "loss": 3.405,
+      "step": 6272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.654050401332128e-06,
+      "loss": 3.3362,
+      "step": 6273
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6525678867443989e-06,
+      "loss": 3.4625,
+      "step": 6274
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6510859052967664e-06,
+      "loss": 3.5231,
+      "step": 6275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.649604457225266e-06,
+      "loss": 3.4581,
+      "step": 6276
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6481235427658427e-06,
+      "loss": 3.3778,
+      "step": 6277
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6466431621543594e-06,
+      "loss": 3.4752,
+      "step": 6278
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6451633156265951e-06,
+      "loss": 3.5791,
+      "step": 6279
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6436840034182417e-06,
+      "loss": 3.3164,
+      "step": 6280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.642205225764908e-06,
+      "loss": 3.3969,
+      "step": 6281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6407269829021173e-06,
+      "loss": 3.3756,
+      "step": 6282
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.639249275065304e-06,
+      "loss": 3.5048,
+      "step": 6283
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6377721024898214e-06,
+      "loss": 3.4363,
+      "step": 6284
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.636295465410937e-06,
+      "loss": 3.3021,
+      "step": 6285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6348193640638328e-06,
+      "loss": 3.4282,
+      "step": 6286
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6333437986836036e-06,
+      "loss": 3.3476,
+      "step": 6287
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6318687695052616e-06,
+      "loss": 3.3816,
+      "step": 6288
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6303942767637333e-06,
+      "loss": 3.3745,
+      "step": 6289
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.628920320693856e-06,
+      "loss": 3.4096,
+      "step": 6290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.627446901530385e-06,
+      "loss": 3.386,
+      "step": 6291
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6259740195079903e-06,
+      "loss": 3.4139,
+      "step": 6292
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6245016748612557e-06,
+      "loss": 3.3983,
+      "step": 6293
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.623029867824678e-06,
+      "loss": 3.3307,
+      "step": 6294
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6215585986326703e-06,
+      "loss": 3.4675,
+      "step": 6295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6200878675195597e-06,
+      "loss": 3.3557,
+      "step": 6296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6186176747195858e-06,
+      "loss": 3.4187,
+      "step": 6297
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.617148020466905e-06,
+      "loss": 3.372,
+      "step": 6298
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6156789049955862e-06,
+      "loss": 3.4231,
+      "step": 6299
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.614210328539615e-06,
+      "loss": 3.3823,
+      "step": 6300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6127422913328855e-06,
+      "loss": 3.4927,
+      "step": 6301
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6112747936092122e-06,
+      "loss": 3.3794,
+      "step": 6302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6098078356023194e-06,
+      "loss": 3.4931,
+      "step": 6303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.608341417545849e-06,
+      "loss": 3.4273,
+      "step": 6304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6068755396733538e-06,
+      "loss": 3.5034,
+      "step": 6305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6054102022183033e-06,
+      "loss": 3.3182,
+      "step": 6306
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6039454054140774e-06,
+      "loss": 3.4812,
+      "step": 6307
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6024811494939723e-06,
+      "loss": 3.2927,
+      "step": 6308
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6010174346911983e-06,
+      "loss": 3.4451,
+      "step": 6309
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5995542612388793e-06,
+      "loss": 3.5961,
+      "step": 6310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.598091629370051e-06,
+      "loss": 3.4407,
+      "step": 6311
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5966295393176678e-06,
+      "loss": 3.3668,
+      "step": 6312
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5951679913145902e-06,
+      "loss": 3.4911,
+      "step": 6313
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.593706985593599e-06,
+      "loss": 3.4997,
+      "step": 6314
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5922465223873852e-06,
+      "loss": 3.3599,
+      "step": 6315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5907866019285544e-06,
+      "loss": 3.4981,
+      "step": 6316
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5893272244496265e-06,
+      "loss": 3.4276,
+      "step": 6317
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5878683901830345e-06,
+      "loss": 3.4308,
+      "step": 6318
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5864100993611237e-06,
+      "loss": 3.3758,
+      "step": 6319
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5849523522161537e-06,
+      "loss": 3.4927,
+      "step": 6320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.583495148980298e-06,
+      "loss": 3.4207,
+      "step": 6321
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5820384898856433e-06,
+      "loss": 3.3933,
+      "step": 6322
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5805823751641885e-06,
+      "loss": 3.4377,
+      "step": 6323
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5791268050478487e-06,
+      "loss": 3.3943,
+      "step": 6324
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5776717797684465e-06,
+      "loss": 3.3927,
+      "step": 6325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.576217299557724e-06,
+      "loss": 3.3369,
+      "step": 6326
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.574763364647333e-06,
+      "loss": 3.4215,
+      "step": 6327
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5733099752688401e-06,
+      "loss": 3.5893,
+      "step": 6328
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.571857131653724e-06,
+      "loss": 3.3587,
+      "step": 6329
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5704048340333784e-06,
+      "loss": 3.5071,
+      "step": 6330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5689530826391041e-06,
+      "loss": 3.4548,
+      "step": 6331
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5675018777021223e-06,
+      "loss": 3.3796,
+      "step": 6332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5660512194535638e-06,
+      "loss": 3.3808,
+      "step": 6333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5646011081244717e-06,
+      "loss": 3.3499,
+      "step": 6334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.563151543945804e-06,
+      "loss": 3.3656,
+      "step": 6335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5617025271484315e-06,
+      "loss": 3.4749,
+      "step": 6336
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.560254057963133e-06,
+      "loss": 3.3589,
+      "step": 6337
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5588061366206065e-06,
+      "loss": 3.4478,
+      "step": 6338
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5573587633514597e-06,
+      "loss": 3.4031,
+      "step": 6339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.555911938386213e-06,
+      "loss": 3.3867,
+      "step": 6340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5544656619553007e-06,
+      "loss": 3.4565,
+      "step": 6341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5530199342890696e-06,
+      "loss": 3.4184,
+      "step": 6342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5515747556177734e-06,
+      "loss": 3.3816,
+      "step": 6343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5501301261715896e-06,
+      "loss": 3.4966,
+      "step": 6344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5486860461805992e-06,
+      "loss": 3.3598,
+      "step": 6345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5472425158747988e-06,
+      "loss": 3.3614,
+      "step": 6346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5457995354840994e-06,
+      "loss": 3.5034,
+      "step": 6347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5443571052383177e-06,
+      "loss": 3.4458,
+      "step": 6348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5429152253671898e-06,
+      "loss": 3.3771,
+      "step": 6349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.541473896100361e-06,
+      "loss": 3.3078,
+      "step": 6350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.54003311766739e-06,
+      "loss": 3.4484,
+      "step": 6351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5385928902977471e-06,
+      "loss": 3.3797,
+      "step": 6352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5371532142208145e-06,
+      "loss": 3.5332,
+      "step": 6353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.535714089665889e-06,
+      "loss": 3.3102,
+      "step": 6354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5342755168621747e-06,
+      "loss": 3.4189,
+      "step": 6355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.532837496038792e-06,
+      "loss": 3.3902,
+      "step": 6356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5314000274247725e-06,
+      "loss": 3.4005,
+      "step": 6357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5299631112490587e-06,
+      "loss": 3.2847,
+      "step": 6358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.528526747740507e-06,
+      "loss": 3.4237,
+      "step": 6359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5270909371278859e-06,
+      "loss": 3.4582,
+      "step": 6360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5256556796398715e-06,
+      "loss": 3.4305,
+      "step": 6361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5242209755050558e-06,
+      "loss": 3.4665,
+      "step": 6362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5227868249519423e-06,
+      "loss": 3.3195,
+      "step": 6363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5213532282089466e-06,
+      "loss": 3.3921,
+      "step": 6364
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5199201855043943e-06,
+      "loss": 3.2477,
+      "step": 6365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5184876970665259e-06,
+      "loss": 3.3871,
+      "step": 6366
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5170557631234879e-06,
+      "loss": 3.4194,
+      "step": 6367
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5156243839033424e-06,
+      "loss": 3.4317,
+      "step": 6368
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5141935596340657e-06,
+      "loss": 3.3047,
+      "step": 6369
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5127632905435419e-06,
+      "loss": 3.4046,
+      "step": 6370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5113335768595693e-06,
+      "loss": 3.4527,
+      "step": 6371
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.509904418809852e-06,
+      "loss": 3.5265,
+      "step": 6372
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.508475816622012e-06,
+      "loss": 3.461,
+      "step": 6373
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5070477705235803e-06,
+      "loss": 3.3327,
+      "step": 6374
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5056202807419995e-06,
+      "loss": 3.3082,
+      "step": 6375
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5041933475046233e-06,
+      "loss": 3.3593,
+      "step": 6376
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5027669710387172e-06,
+      "loss": 3.3589,
+      "step": 6377
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.50134115157146e-06,
+      "loss": 3.3445,
+      "step": 6378
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.499915889329936e-06,
+      "loss": 3.4081,
+      "step": 6379
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4984911845411453e-06,
+      "loss": 3.3817,
+      "step": 6380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4970670374320001e-06,
+      "loss": 3.3575,
+      "step": 6381
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4956434482293204e-06,
+      "loss": 3.4154,
+      "step": 6382
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4942204171598396e-06,
+      "loss": 3.49,
+      "step": 6383
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4927979444502028e-06,
+      "loss": 3.2517,
+      "step": 6384
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.491376030326962e-06,
+      "loss": 3.3948,
+      "step": 6385
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.489954675016585e-06,
+      "loss": 3.5298,
+      "step": 6386
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.488533878745449e-06,
+      "loss": 3.5011,
+      "step": 6387
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4871136417398407e-06,
+      "loss": 3.3938,
+      "step": 6388
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.48569396422596e-06,
+      "loss": 3.4139,
+      "step": 6389
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4842748464299179e-06,
+      "loss": 3.478,
+      "step": 6390
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.482856288577732e-06,
+      "loss": 3.3147,
+      "step": 6391
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4814382908953334e-06,
+      "loss": 3.4452,
+      "step": 6392
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4800208536085675e-06,
+      "loss": 3.3853,
+      "step": 6393
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.478603976943186e-06,
+      "loss": 3.3551,
+      "step": 6394
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4771876611248538e-06,
+      "loss": 3.4972,
+      "step": 6395
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.475771906379142e-06,
+      "loss": 3.5235,
+      "step": 6396
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4743567129315377e-06,
+      "loss": 3.3481,
+      "step": 6397
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4729420810074357e-06,
+      "loss": 3.4652,
+      "step": 6398
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4715280108321427e-06,
+      "loss": 3.401,
+      "step": 6399
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4701145026308754e-06,
+      "loss": 3.4179,
+      "step": 6400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4687015566287627e-06,
+      "loss": 3.5446,
+      "step": 6401
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4672891730508382e-06,
+      "loss": 3.4408,
+      "step": 6402
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.465877352122052e-06,
+      "loss": 3.3873,
+      "step": 6403
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 3.4093,
+      "step": 6404
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.463055399111239e-06,
+      "loss": 3.5253,
+      "step": 6405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4616452674786601e-06,
+      "loss": 3.3682,
+      "step": 6406
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4602356993941153e-06,
+      "loss": 3.4122,
+      "step": 6407
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4588266950821055e-06,
+      "loss": 3.4973,
+      "step": 6408
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4574182547670378e-06,
+      "loss": 3.2443,
+      "step": 6409
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4560103786732337e-06,
+      "loss": 3.3268,
+      "step": 6410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4546030670249233e-06,
+      "loss": 3.4476,
+      "step": 6411
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4531963200462468e-06,
+      "loss": 3.4358,
+      "step": 6412
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4517901379612542e-06,
+      "loss": 3.5085,
+      "step": 6413
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4503845209939083e-06,
+      "loss": 3.4294,
+      "step": 6414
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4489794693680754e-06,
+      "loss": 3.3305,
+      "step": 6415
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.447574983307538e-06,
+      "loss": 3.3879,
+      "step": 6416
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4461710630359843e-06,
+      "loss": 3.4862,
+      "step": 6417
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.444767708777018e-06,
+      "loss": 3.4906,
+      "step": 6418
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4433649207541484e-06,
+      "loss": 3.508,
+      "step": 6419
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4419626991907925e-06,
+      "loss": 3.3789,
+      "step": 6420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4405610443102819e-06,
+      "loss": 3.4181,
+      "step": 6421
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4391599563358555e-06,
+      "loss": 3.5092,
+      "step": 6422
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4377594354906622e-06,
+      "loss": 3.357,
+      "step": 6423
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4363594819977606e-06,
+      "loss": 3.4086,
+      "step": 6424
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4349600960801208e-06,
+      "loss": 3.2762,
+      "step": 6425
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4335612779606173e-06,
+      "loss": 3.4266,
+      "step": 6426
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4321630278620397e-06,
+      "loss": 3.4619,
+      "step": 6427
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4307653460070848e-06,
+      "loss": 3.4078,
+      "step": 6428
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4293682326183594e-06,
+      "loss": 3.4137,
+      "step": 6429
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4279716879183786e-06,
+      "loss": 3.3855,
+      "step": 6430
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4265757121295704e-06,
+      "loss": 3.4579,
+      "step": 6431
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4251803054742658e-06,
+      "loss": 3.4692,
+      "step": 6432
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.423785468174711e-06,
+      "loss": 3.3702,
+      "step": 6433
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4223912004530593e-06,
+      "loss": 3.4126,
+      "step": 6434
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4209975025313733e-06,
+      "loss": 3.5637,
+      "step": 6435
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4196043746316252e-06,
+      "loss": 3.4309,
+      "step": 6436
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4182118169756959e-06,
+      "loss": 3.4432,
+      "step": 6437
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4168198297853774e-06,
+      "loss": 3.2517,
+      "step": 6438
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.415428413282367e-06,
+      "loss": 3.3165,
+      "step": 6439
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.414037567688274e-06,
+      "loss": 3.343,
+      "step": 6440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4126472932246149e-06,
+      "loss": 3.436,
+      "step": 6441
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4112575901128195e-06,
+      "loss": 3.5115,
+      "step": 6442
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4098684585742233e-06,
+      "loss": 3.4179,
+      "step": 6443
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4084798988300684e-06,
+      "loss": 3.4648,
+      "step": 6444
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4070919111015097e-06,
+      "loss": 3.4782,
+      "step": 6445
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4057044956096099e-06,
+      "loss": 3.411,
+      "step": 6446
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4043176525753405e-06,
+      "loss": 3.4664,
+      "step": 6447
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4029313822195817e-06,
+      "loss": 3.5005,
+      "step": 6448
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4015456847631243e-06,
+      "loss": 3.3986,
+      "step": 6449
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4001605604266622e-06,
+      "loss": 3.4805,
+      "step": 6450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3987760094308039e-06,
+      "loss": 3.3912,
+      "step": 6451
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3973920319960654e-06,
+      "loss": 3.2634,
+      "step": 6452
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3960086283428692e-06,
+      "loss": 3.3545,
+      "step": 6453
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3946257986915486e-06,
+      "loss": 3.4047,
+      "step": 6454
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.393243543262346e-06,
+      "loss": 3.4675,
+      "step": 6455
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3918618622754076e-06,
+      "loss": 3.3498,
+      "step": 6456
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3904807559507939e-06,
+      "loss": 3.4175,
+      "step": 6457
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3891002245084706e-06,
+      "loss": 3.4749,
+      "step": 6458
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3877202681683128e-06,
+      "loss": 3.428,
+      "step": 6459
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3863408871501043e-06,
+      "loss": 3.4257,
+      "step": 6460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3849620816735365e-06,
+      "loss": 3.4456,
+      "step": 6461
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3835838519582117e-06,
+      "loss": 3.5112,
+      "step": 6462
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3822061982236346e-06,
+      "loss": 3.5442,
+      "step": 6463
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3808291206892232e-06,
+      "loss": 3.3791,
+      "step": 6464
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.379452619574303e-06,
+      "loss": 3.3145,
+      "step": 6465
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3780766950981055e-06,
+      "loss": 3.4813,
+      "step": 6466
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3767013474797764e-06,
+      "loss": 3.434,
+      "step": 6467
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3753265769383605e-06,
+      "loss": 3.4681,
+      "step": 6468
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3739523836928171e-06,
+      "loss": 3.3303,
+      "step": 6469
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3725787679620117e-06,
+      "loss": 3.4059,
+      "step": 6470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.371205729964717e-06,
+      "loss": 3.4032,
+      "step": 6471
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.369833269919616e-06,
+      "loss": 3.4753,
+      "step": 6472
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3684613880452979e-06,
+      "loss": 3.3832,
+      "step": 6473
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3670900845602585e-06,
+      "loss": 3.3247,
+      "step": 6474
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3657193596829044e-06,
+      "loss": 3.3816,
+      "step": 6475
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3643492136315483e-06,
+      "loss": 3.3841,
+      "step": 6476
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3629796466244117e-06,
+      "loss": 3.4755,
+      "step": 6477
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3616106588796229e-06,
+      "loss": 3.2985,
+      "step": 6478
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3602422506152196e-06,
+      "loss": 3.4114,
+      "step": 6479
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3588744220491434e-06,
+      "loss": 3.4733,
+      "step": 6480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.357507173399248e-06,
+      "loss": 3.4059,
+      "step": 6481
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.356140504883292e-06,
+      "loss": 3.3964,
+      "step": 6482
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.354774416718943e-06,
+      "loss": 3.4442,
+      "step": 6483
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3534089091237757e-06,
+      "loss": 3.4433,
+      "step": 6484
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.352043982315273e-06,
+      "loss": 3.3121,
+      "step": 6485
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3506796365108232e-06,
+      "loss": 3.4464,
+      "step": 6486
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3493158719277232e-06,
+      "loss": 3.4703,
+      "step": 6487
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.347952688783179e-06,
+      "loss": 3.4287,
+      "step": 6488
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3465900872943017e-06,
+      "loss": 3.3257,
+      "step": 6489
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.345228067678111e-06,
+      "loss": 3.4165,
+      "step": 6490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.343866630151533e-06,
+      "loss": 3.4392,
+      "step": 6491
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3425057749314023e-06,
+      "loss": 3.3985,
+      "step": 6492
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3411455022344598e-06,
+      "loss": 3.4522,
+      "step": 6493
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3397858122773544e-06,
+      "loss": 3.3621,
+      "step": 6494
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3384267052766413e-06,
+      "loss": 3.4078,
+      "step": 6495
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.337068181448783e-06,
+      "loss": 3.4137,
+      "step": 6496
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3357102410101519e-06,
+      "loss": 3.3984,
+      "step": 6497
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3343528841770204e-06,
+      "loss": 3.3945,
+      "step": 6498
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3329961111655754e-06,
+      "loss": 3.4184,
+      "step": 6499
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3316399221919075e-06,
+      "loss": 3.2754,
+      "step": 6500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3302843174720148e-06,
+      "loss": 3.2553,
+      "step": 6501
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3289292972218015e-06,
+      "loss": 3.3681,
+      "step": 6502
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3275748616570822e-06,
+      "loss": 3.3252,
+      "step": 6503
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3262210109935719e-06,
+      "loss": 3.3472,
+      "step": 6504
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3248677454468984e-06,
+      "loss": 3.3558,
+      "step": 6505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.323515065232593e-06,
+      "loss": 3.4468,
+      "step": 6506
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.322162970566096e-06,
+      "loss": 3.4259,
+      "step": 6507
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3208114616627526e-06,
+      "loss": 3.3365,
+      "step": 6508
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3194605387378168e-06,
+      "loss": 3.4312,
+      "step": 6509
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3181102020064456e-06,
+      "loss": 3.5174,
+      "step": 6510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3167604516837062e-06,
+      "loss": 3.4032,
+      "step": 6511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3154112879845715e-06,
+      "loss": 3.4276,
+      "step": 6512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3140627111239196e-06,
+      "loss": 3.5432,
+      "step": 6513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.312714721316537e-06,
+      "loss": 3.515,
+      "step": 6514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3113673187771153e-06,
+      "loss": 3.3975,
+      "step": 6515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.310020503720254e-06,
+      "loss": 3.3115,
+      "step": 6516
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3086742763604571e-06,
+      "loss": 3.4088,
+      "step": 6517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3073286369121368e-06,
+      "loss": 3.3811,
+      "step": 6518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3059835855896109e-06,
+      "loss": 3.541,
+      "step": 6519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3046391226071043e-06,
+      "loss": 3.3558,
+      "step": 6520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3032952481787453e-06,
+      "loss": 3.4753,
+      "step": 6521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3019519625185718e-06,
+      "loss": 3.4559,
+      "step": 6522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3006092658405262e-06,
+      "loss": 3.381,
+      "step": 6523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2992671583584587e-06,
+      "loss": 3.2165,
+      "step": 6524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2979256402861234e-06,
+      "loss": 3.3895,
+      "step": 6525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2965847118371827e-06,
+      "loss": 3.391,
+      "step": 6526
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2952443732252058e-06,
+      "loss": 3.4416,
+      "step": 6527
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2939046246636616e-06,
+      "loss": 3.3528,
+      "step": 6528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.292565466365933e-06,
+      "loss": 3.4831,
+      "step": 6529
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2912268985453047e-06,
+      "loss": 3.3769,
+      "step": 6530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2898889214149685e-06,
+      "loss": 3.3786,
+      "step": 6531
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2885515351880217e-06,
+      "loss": 3.5042,
+      "step": 6532
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2872147400774686e-06,
+      "loss": 3.4626,
+      "step": 6533
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2858785362962166e-06,
+      "loss": 3.3789,
+      "step": 6534
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2845429240570818e-06,
+      "loss": 3.3876,
+      "step": 6535
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.283207903572784e-06,
+      "loss": 3.5023,
+      "step": 6536
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2818734750559514e-06,
+      "loss": 3.4482,
+      "step": 6537
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.280539638719115e-06,
+      "loss": 3.4373,
+      "step": 6538
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2792063947747146e-06,
+      "loss": 3.3572,
+      "step": 6539
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2778737434350897e-06,
+      "loss": 3.4409,
+      "step": 6540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2765416849124933e-06,
+      "loss": 3.4728,
+      "step": 6541
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.275210219419079e-06,
+      "loss": 3.4932,
+      "step": 6542
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2738793471669076e-06,
+      "loss": 3.4564,
+      "step": 6543
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2725490683679458e-06,
+      "loss": 3.3147,
+      "step": 6544
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.271219383234063e-06,
+      "loss": 3.2832,
+      "step": 6545
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2698902919770361e-06,
+      "loss": 3.4592,
+      "step": 6546
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2685617948085483e-06,
+      "loss": 3.4256,
+      "step": 6547
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2672338919401866e-06,
+      "loss": 3.3434,
+      "step": 6548
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2659065835834444e-06,
+      "loss": 3.4453,
+      "step": 6549
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2645798699497203e-06,
+      "loss": 3.3839,
+      "step": 6550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2632537512503178e-06,
+      "loss": 3.4022,
+      "step": 6551
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2619282276964446e-06,
+      "loss": 3.1857,
+      "step": 6552
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.260603299499215e-06,
+      "loss": 3.3542,
+      "step": 6553
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2592789668696482e-06,
+      "loss": 3.3186,
+      "step": 6554
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2579552300186693e-06,
+      "loss": 3.4875,
+      "step": 6555
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2566320891571066e-06,
+      "loss": 3.3572,
+      "step": 6556
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.255309544495697e-06,
+      "loss": 3.3268,
+      "step": 6557
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.253987596245077e-06,
+      "loss": 3.3703,
+      "step": 6558
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2526662446157921e-06,
+      "loss": 3.4121,
+      "step": 6559
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.251345489818292e-06,
+      "loss": 3.4071,
+      "step": 6560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2500253320629319e-06,
+      "loss": 3.3335,
+      "step": 6561
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2487057715599704e-06,
+      "loss": 3.4549,
+      "step": 6562
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2473868085195733e-06,
+      "loss": 3.346,
+      "step": 6563
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2460684431518055e-06,
+      "loss": 3.4945,
+      "step": 6564
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2447506756666455e-06,
+      "loss": 3.4308,
+      "step": 6565
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2434335062739699e-06,
+      "loss": 3.365,
+      "step": 6566
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2421169351835627e-06,
+      "loss": 3.5044,
+      "step": 6567
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2408009626051137e-06,
+      "loss": 3.2554,
+      "step": 6568
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2394855887482121e-06,
+      "loss": 3.3433,
+      "step": 6569
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2381708138223576e-06,
+      "loss": 3.4829,
+      "step": 6570
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2368566380369512e-06,
+      "loss": 3.3977,
+      "step": 6571
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.235543061601301e-06,
+      "loss": 3.4148,
+      "step": 6572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2342300847246168e-06,
+      "loss": 3.4211,
+      "step": 6573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.232917707616017e-06,
+      "loss": 3.4821,
+      "step": 6574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2316059304845174e-06,
+      "loss": 3.3396,
+      "step": 6575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2302947535390459e-06,
+      "loss": 3.4066,
+      "step": 6576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2289841769884298e-06,
+      "loss": 3.4021,
+      "step": 6577
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2276742010414034e-06,
+      "loss": 3.5049,
+      "step": 6578
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2263648259066046e-06,
+      "loss": 3.46,
+      "step": 6579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2250560517925747e-06,
+      "loss": 3.3689,
+      "step": 6580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2237478789077623e-06,
+      "loss": 3.2987,
+      "step": 6581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2224403074605146e-06,
+      "loss": 3.4708,
+      "step": 6582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2211333376590877e-06,
+      "loss": 3.3712,
+      "step": 6583
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2198269697116416e-06,
+      "loss": 3.3779,
+      "step": 6584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2185212038262379e-06,
+      "loss": 3.4608,
+      "step": 6585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2172160402108453e-06,
+      "loss": 3.378,
+      "step": 6586
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2159114790733352e-06,
+      "loss": 3.4948,
+      "step": 6587
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2146075206214807e-06,
+      "loss": 3.2873,
+      "step": 6588
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2133041650629613e-06,
+      "loss": 3.4489,
+      "step": 6589
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.212001412605363e-06,
+      "loss": 3.4799,
+      "step": 6590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2106992634561716e-06,
+      "loss": 3.2899,
+      "step": 6591
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.20939771782278e-06,
+      "loss": 3.3318,
+      "step": 6592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2080967759124802e-06,
+      "loss": 3.4235,
+      "step": 6593
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.206796437932472e-06,
+      "loss": 3.5652,
+      "step": 6594
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.205496704089859e-06,
+      "loss": 3.4857,
+      "step": 6595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2041975745916474e-06,
+      "loss": 3.421,
+      "step": 6596
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2028990496447473e-06,
+      "loss": 3.3783,
+      "step": 6597
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2016011294559744e-06,
+      "loss": 3.3953,
+      "step": 6598
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2003038142320434e-06,
+      "loss": 3.3495,
+      "step": 6599
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1990071041795765e-06,
+      "loss": 3.4645,
+      "step": 6600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1977109995050995e-06,
+      "loss": 3.4055,
+      "step": 6601
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1964155004150402e-06,
+      "loss": 3.4,
+      "step": 6602
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1951206071157306e-06,
+      "loss": 3.3919,
+      "step": 6603
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1938263198134087e-06,
+      "loss": 3.4915,
+      "step": 6604
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1925326387142095e-06,
+      "loss": 3.4357,
+      "step": 6605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1912395640241776e-06,
+      "loss": 3.5707,
+      "step": 6606
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1899470959492583e-06,
+      "loss": 3.3248,
+      "step": 6607
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1886552346953018e-06,
+      "loss": 3.4595,
+      "step": 6608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1873639804680599e-06,
+      "loss": 3.3336,
+      "step": 6609
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.186073333473189e-06,
+      "loss": 3.4881,
+      "step": 6610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1847832939162497e-06,
+      "loss": 3.4259,
+      "step": 6611
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.183493862002702e-06,
+      "loss": 3.394,
+      "step": 6612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1822050379379124e-06,
+      "loss": 3.4263,
+      "step": 6613
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1809168219271488e-06,
+      "loss": 3.3757,
+      "step": 6614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1796292141755861e-06,
+      "loss": 3.4249,
+      "step": 6615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.178342214888299e-06,
+      "loss": 3.3424,
+      "step": 6616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.177055824270264e-06,
+      "loss": 3.3236,
+      "step": 6617
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.175770042526363e-06,
+      "loss": 3.3989,
+      "step": 6618
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1744848698613808e-06,
+      "loss": 3.311,
+      "step": 6619
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1732003064800046e-06,
+      "loss": 3.5013,
+      "step": 6620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1719163525868248e-06,
+      "loss": 3.358,
+      "step": 6621
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.170633008386336e-06,
+      "loss": 3.4334,
+      "step": 6622
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1693502740829315e-06,
+      "loss": 3.368,
+      "step": 6623
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.168068149880912e-06,
+      "loss": 3.4193,
+      "step": 6624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1667866359844792e-06,
+      "loss": 3.2667,
+      "step": 6625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1655057325977375e-06,
+      "loss": 3.4371,
+      "step": 6626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1642254399246943e-06,
+      "loss": 3.4804,
+      "step": 6627
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1629457581692616e-06,
+      "loss": 3.4135,
+      "step": 6628
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1616666875352494e-06,
+      "loss": 3.3384,
+      "step": 6629
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.160388228226374e-06,
+      "loss": 3.4003,
+      "step": 6630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1591103804462538e-06,
+      "loss": 3.4343,
+      "step": 6631
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1578331443984103e-06,
+      "loss": 3.326,
+      "step": 6632
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1565565202862661e-06,
+      "loss": 3.4494,
+      "step": 6633
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1552805083131468e-06,
+      "loss": 3.3898,
+      "step": 6634
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.154005108682283e-06,
+      "loss": 3.3955,
+      "step": 6635
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1527303215968016e-06,
+      "loss": 3.4774,
+      "step": 6636
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1514561472597386e-06,
+      "loss": 3.4117,
+      "step": 6637
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1501825858740273e-06,
+      "loss": 3.3185,
+      "step": 6638
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1489096376425101e-06,
+      "loss": 3.4239,
+      "step": 6639
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.147637302767924e-06,
+      "loss": 3.4032,
+      "step": 6640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1463655814529118e-06,
+      "loss": 3.3532,
+      "step": 6641
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1450944739000197e-06,
+      "loss": 3.332,
+      "step": 6642
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.143823980311694e-06,
+      "loss": 3.4676,
+      "step": 6643
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1425541008902852e-06,
+      "loss": 3.3859,
+      "step": 6644
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1412848358380442e-06,
+      "loss": 3.3624,
+      "step": 6645
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1400161853571263e-06,
+      "loss": 3.4437,
+      "step": 6646
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1387481496495845e-06,
+      "loss": 3.4459,
+      "step": 6647
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1374807289173783e-06,
+      "loss": 3.1728,
+      "step": 6648
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1362139233623675e-06,
+      "loss": 3.4929,
+      "step": 6649
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.134947733186315e-06,
+      "loss": 3.4441,
+      "step": 6650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1336821585908848e-06,
+      "loss": 3.4617,
+      "step": 6651
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.132417199777643e-06,
+      "loss": 3.3137,
+      "step": 6652
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1311528569480558e-06,
+      "loss": 3.3567,
+      "step": 6653
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1298891303034942e-06,
+      "loss": 3.3936,
+      "step": 6654
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1286260200452303e-06,
+      "loss": 3.4782,
+      "step": 6655
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.127363526374437e-06,
+      "loss": 3.3703,
+      "step": 6656
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.12610164949219e-06,
+      "loss": 3.3125,
+      "step": 6657
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1248403895994675e-06,
+      "loss": 3.3783,
+      "step": 6658
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1235797468971455e-06,
+      "loss": 3.4115,
+      "step": 6659
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1223197215860065e-06,
+      "loss": 3.5142,
+      "step": 6660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.121060313866732e-06,
+      "loss": 3.3894,
+      "step": 6661
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1198015239399063e-06,
+      "loss": 3.4208,
+      "step": 6662
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1185433520060147e-06,
+      "loss": 3.3371,
+      "step": 6663
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1172857982654445e-06,
+      "loss": 3.4301,
+      "step": 6664
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1160288629184834e-06,
+      "loss": 3.412,
+      "step": 6665
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1147725461653225e-06,
+      "loss": 3.4648,
+      "step": 6666
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1135168482060521e-06,
+      "loss": 3.3254,
+      "step": 6667
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1122617692406667e-06,
+      "loss": 3.42,
+      "step": 6668
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1110073094690593e-06,
+      "loss": 3.4404,
+      "step": 6669
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1097534690910283e-06,
+      "loss": 3.2705,
+      "step": 6670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1085002483062667e-06,
+      "loss": 3.4769,
+      "step": 6671
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1072476473143755e-06,
+      "loss": 3.483,
+      "step": 6672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.105995666314854e-06,
+      "loss": 3.4964,
+      "step": 6673
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1047443055071033e-06,
+      "loss": 3.3443,
+      "step": 6674
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.103493565090425e-06,
+      "loss": 3.4229,
+      "step": 6675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1022434452640252e-06,
+      "loss": 3.3532,
+      "step": 6676
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1009939462270042e-06,
+      "loss": 3.2762,
+      "step": 6677
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.09974506817837e-06,
+      "loss": 3.4506,
+      "step": 6678
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.098496811317029e-06,
+      "loss": 3.3649,
+      "step": 6679
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0972491758417892e-06,
+      "loss": 3.4161,
+      "step": 6680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0960021619513595e-06,
+      "loss": 3.3153,
+      "step": 6681
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.094755769844351e-06,
+      "loss": 3.3376,
+      "step": 6682
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0935099997192722e-06,
+      "loss": 3.4386,
+      "step": 6683
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.092264851774536e-06,
+      "loss": 3.3134,
+      "step": 6684
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.091020326208455e-06,
+      "loss": 3.4205,
+      "step": 6685
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.089776423219243e-06,
+      "loss": 3.3693,
+      "step": 6686
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0885331430050139e-06,
+      "loss": 3.3829,
+      "step": 6687
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0872904857637839e-06,
+      "loss": 3.4033,
+      "step": 6688
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0860484516934684e-06,
+      "loss": 3.3227,
+      "step": 6689
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0848070409918837e-06,
+      "loss": 3.4532,
+      "step": 6690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0835662538567482e-06,
+      "loss": 3.4706,
+      "step": 6691
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0823260904856791e-06,
+      "loss": 3.4508,
+      "step": 6692
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0810865510761975e-06,
+      "loss": 3.3465,
+      "step": 6693
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0798476358257198e-06,
+      "loss": 3.3914,
+      "step": 6694
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0786093449315672e-06,
+      "loss": 3.424,
+      "step": 6695
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0773716785909599e-06,
+      "loss": 3.4288,
+      "step": 6696
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.07613463700102e-06,
+      "loss": 3.4146,
+      "step": 6697
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0748982203587688e-06,
+      "loss": 3.448,
+      "step": 6698
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.073662428861128e-06,
+      "loss": 3.5209,
+      "step": 6699
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0724272627049215e-06,
+      "loss": 3.432,
+      "step": 6700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0711927220868694e-06,
+      "loss": 3.4937,
+      "step": 6701
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0699588072035972e-06,
+      "loss": 3.4017,
+      "step": 6702
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0687255182516277e-06,
+      "loss": 3.4357,
+      "step": 6703
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.067492855427385e-06,
+      "loss": 3.3214,
+      "step": 6704
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0662608189271938e-06,
+      "loss": 3.4727,
+      "step": 6705
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.065029408947279e-06,
+      "loss": 3.3532,
+      "step": 6706
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0637986256837634e-06,
+      "loss": 3.3704,
+      "step": 6707
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0625684693326727e-06,
+      "loss": 3.3987,
+      "step": 6708
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0613389400899322e-06,
+      "loss": 3.5369,
+      "step": 6709
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.060110038151367e-06,
+      "loss": 3.3771,
+      "step": 6710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.058881763712702e-06,
+      "loss": 3.3842,
+      "step": 6711
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0576541169695625e-06,
+      "loss": 3.3945,
+      "step": 6712
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0564270981174746e-06,
+      "loss": 3.4634,
+      "step": 6713
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0552007073518628e-06,
+      "loss": 3.382,
+      "step": 6714
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.053974944868053e-06,
+      "loss": 3.4602,
+      "step": 6715
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0527498108612694e-06,
+      "loss": 3.4421,
+      "step": 6716
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0515253055266394e-06,
+      "loss": 3.3455,
+      "step": 6717
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.050301429059185e-06,
+      "loss": 3.5071,
+      "step": 6718
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0490781816538326e-06,
+      "loss": 3.459,
+      "step": 6719
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0478555635054061e-06,
+      "loss": 3.3909,
+      "step": 6720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.046633574808631e-06,
+      "loss": 3.4169,
+      "step": 6721
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0454122157581304e-06,
+      "loss": 3.4017,
+      "step": 6722
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0441914865484288e-06,
+      "loss": 3.502,
+      "step": 6723
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0429713873739505e-06,
+      "loss": 3.424,
+      "step": 6724
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.041751918429017e-06,
+      "loss": 3.3892,
+      "step": 6725
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0405330799078523e-06,
+      "loss": 3.4068,
+      "step": 6726
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0393148720045777e-06,
+      "loss": 3.3412,
+      "step": 6727
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0380972949132167e-06,
+      "loss": 3.4136,
+      "step": 6728
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0368803488276895e-06,
+      "loss": 3.539,
+      "step": 6729
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0356640339418195e-06,
+      "loss": 3.4707,
+      "step": 6730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0344483504493248e-06,
+      "loss": 3.4533,
+      "step": 6731
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0332332985438248e-06,
+      "loss": 3.3139,
+      "step": 6732
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0320188784188405e-06,
+      "loss": 3.3597,
+      "step": 6733
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0308050902677907e-06,
+      "loss": 3.2683,
+      "step": 6734
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0295919342839928e-06,
+      "loss": 3.433,
+      "step": 6735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0283794106606654e-06,
+      "loss": 3.4491,
+      "step": 6736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.027167519590922e-06,
+      "loss": 3.3523,
+      "step": 6737
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0259562612677815e-06,
+      "loss": 3.3952,
+      "step": 6738
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0247456358841584e-06,
+      "loss": 3.4737,
+      "step": 6739
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0235356436328675e-06,
+      "loss": 3.3367,
+      "step": 6740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0223262847066224e-06,
+      "loss": 3.439,
+      "step": 6741
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0211175592980337e-06,
+      "loss": 3.4854,
+      "step": 6742
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0199094675996145e-06,
+      "loss": 3.4283,
+      "step": 6743
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0187020098037759e-06,
+      "loss": 3.3942,
+      "step": 6744
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.017495186102827e-06,
+      "loss": 3.3575,
+      "step": 6745
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0162889966889772e-06,
+      "loss": 3.491,
+      "step": 6746
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.015083441754336e-06,
+      "loss": 3.5038,
+      "step": 6747
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0138785214909065e-06,
+      "loss": 3.3902,
+      "step": 6748
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0126742360905967e-06,
+      "loss": 3.3319,
+      "step": 6749
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0114705857452106e-06,
+      "loss": 3.4417,
+      "step": 6750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0102675706464521e-06,
+      "loss": 3.4372,
+      "step": 6751
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0090651909859228e-06,
+      "loss": 3.4699,
+      "step": 6752
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0078634469551246e-06,
+      "loss": 3.4743,
+      "step": 6753
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0066623387454582e-06,
+      "loss": 3.3986,
+      "step": 6754
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0054618665482197e-06,
+      "loss": 3.4184,
+      "step": 6755
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0042620305546069e-06,
+      "loss": 3.3926,
+      "step": 6756
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0030628309557166e-06,
+      "loss": 3.3546,
+      "step": 6757
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0018642679425433e-06,
+      "loss": 3.5327,
+      "step": 6758
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0006663417059804e-06,
+      "loss": 3.3064,
+      "step": 6759
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.9946905243682e-07,
+      "loss": 3.4626,
+      "step": 6760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.982724003257493e-07,
+      "loss": 3.287,
+      "step": 6761
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.970763855633613e-07,
+      "loss": 3.3412,
+      "step": 6762
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.958810083401411e-07,
+      "loss": 3.3856,
+      "step": 6763
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.946862688464753e-07,
+      "loss": 3.394,
+      "step": 6764
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.934921672726489e-07,
+      "loss": 3.2521,
+      "step": 6765
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.922987038088428e-07,
+      "loss": 3.4287,
+      "step": 6766
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.911058786451383e-07,
+      "loss": 3.4299,
+      "step": 6767
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.899136919715147e-07,
+      "loss": 3.3669,
+      "step": 6768
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.887221439778505e-07,
+      "loss": 3.2808,
+      "step": 6769
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.875312348539213e-07,
+      "loss": 3.3184,
+      "step": 6770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.863409647894018e-07,
+      "loss": 3.4303,
+      "step": 6771
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.851513339738627e-07,
+      "loss": 3.4004,
+      "step": 6772
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.83962342596776e-07,
+      "loss": 3.3766,
+      "step": 6773
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.827739908475099e-07,
+      "loss": 3.4234,
+      "step": 6774
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.815862789153314e-07,
+      "loss": 3.3803,
+      "step": 6775
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.803992069894053e-07,
+      "loss": 3.3875,
+      "step": 6776
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.79212775258796e-07,
+      "loss": 3.4207,
+      "step": 6777
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.780269839124617e-07,
+      "loss": 3.3897,
+      "step": 6778
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.768418331392638e-07,
+      "loss": 3.3671,
+      "step": 6779
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.756573231279582e-07,
+      "loss": 3.3952,
+      "step": 6780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.744734540672006e-07,
+      "loss": 3.424,
+      "step": 6781
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.732902261455435e-07,
+      "loss": 3.3516,
+      "step": 6782
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.721076395514384e-07,
+      "loss": 3.4951,
+      "step": 6783
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.709256944732343e-07,
+      "loss": 3.5093,
+      "step": 6784
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.697443910991755e-07,
+      "loss": 3.4974,
+      "step": 6785
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.68563729617406e-07,
+      "loss": 3.4361,
+      "step": 6786
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.673837102159712e-07,
+      "loss": 3.4179,
+      "step": 6787
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.662043330828086e-07,
+      "loss": 3.4772,
+      "step": 6788
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.650255984057578e-07,
+      "loss": 3.3989,
+      "step": 6789
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.63847506372551e-07,
+      "loss": 3.446,
+      "step": 6790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.62670057170822e-07,
+      "loss": 3.2874,
+      "step": 6791
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.61493250988102e-07,
+      "loss": 3.3655,
+      "step": 6792
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.603170880118185e-07,
+      "loss": 3.3584,
+      "step": 6793
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.591415684292975e-07,
+      "loss": 3.4063,
+      "step": 6794
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.579666924277625e-07,
+      "loss": 3.3479,
+      "step": 6795
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.567924601943323e-07,
+      "loss": 3.3493,
+      "step": 6796
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.556188719160265e-07,
+      "loss": 3.4083,
+      "step": 6797
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.544459277797597e-07,
+      "loss": 3.3731,
+      "step": 6798
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.532736279723453e-07,
+      "loss": 3.3866,
+      "step": 6799
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.521019726804937e-07,
+      "loss": 3.3944,
+      "step": 6800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.509309620908136e-07,
+      "loss": 3.4276,
+      "step": 6801
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.497605963898076e-07,
+      "loss": 3.3942,
+      "step": 6802
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.48590875763879e-07,
+      "loss": 3.362,
+      "step": 6803
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.474218003993275e-07,
+      "loss": 3.3311,
+      "step": 6804
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.462533704823496e-07,
+      "loss": 3.4729,
+      "step": 6805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.450855861990394e-07,
+      "loss": 3.446,
+      "step": 6806
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.439184477353886e-07,
+      "loss": 3.2811,
+      "step": 6807
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.427519552772857e-07,
+      "loss": 3.2807,
+      "step": 6808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.415861090105138e-07,
+      "loss": 3.4532,
+      "step": 6809
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.404209091207555e-07,
+      "loss": 3.505,
+      "step": 6810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.392563557935936e-07,
+      "loss": 3.3331,
+      "step": 6811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.380924492145033e-07,
+      "loss": 3.3375,
+      "step": 6812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.369291895688565e-07,
+      "loss": 3.3471,
+      "step": 6813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.357665770419244e-07,
+      "loss": 3.3641,
+      "step": 6814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.346046118188745e-07,
+      "loss": 3.4797,
+      "step": 6815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.334432940847715e-07,
+      "loss": 3.4383,
+      "step": 6816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.322826240245764e-07,
+      "loss": 3.3358,
+      "step": 6817
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.311226018231467e-07,
+      "loss": 3.4498,
+      "step": 6818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.299632276652399e-07,
+      "loss": 3.4132,
+      "step": 6819
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.288045017355035e-07,
+      "loss": 3.3831,
+      "step": 6820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.276464242184885e-07,
+      "loss": 3.4463,
+      "step": 6821
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.264889952986389e-07,
+      "loss": 3.4677,
+      "step": 6822
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.253322151602972e-07,
+      "loss": 3.3839,
+      "step": 6823
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.241760839877023e-07,
+      "loss": 3.441,
+      "step": 6824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.230206019649896e-07,
+      "loss": 3.3449,
+      "step": 6825
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.21865769276189e-07,
+      "loss": 3.3132,
+      "step": 6826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.207115861052301e-07,
+      "loss": 3.384,
+      "step": 6827
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.195580526359371e-07,
+      "loss": 3.3829,
+      "step": 6828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.184051690520324e-07,
+      "loss": 3.4138,
+      "step": 6829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.172529355371329e-07,
+      "loss": 3.4383,
+      "step": 6830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.161013522747553e-07,
+      "loss": 3.3358,
+      "step": 6831
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.149504194483066e-07,
+      "loss": 3.428,
+      "step": 6832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.138001372410965e-07,
+      "loss": 3.3386,
+      "step": 6833
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.126505058363283e-07,
+      "loss": 3.4807,
+      "step": 6834
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.115015254171e-07,
+      "loss": 3.3571,
+      "step": 6835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.10353196166412e-07,
+      "loss": 3.3854,
+      "step": 6836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.092055182671533e-07,
+      "loss": 3.4386,
+      "step": 6837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.08058491902114e-07,
+      "loss": 3.4364,
+      "step": 6838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.069121172539791e-07,
+      "loss": 3.3014,
+      "step": 6839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.057663945053291e-07,
+      "loss": 3.5069,
+      "step": 6840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.046213238386425e-07,
+      "loss": 3.4569,
+      "step": 6841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.034769054362918e-07,
+      "loss": 3.3665,
+      "step": 6842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.023331394805491e-07,
+      "loss": 3.4104,
+      "step": 6843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.011900261535767e-07,
+      "loss": 3.3678,
+      "step": 6844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.00047565637438e-07,
+      "loss": 3.3801,
+      "step": 6845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.989057581140909e-07,
+      "loss": 3.2593,
+      "step": 6846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.977646037653886e-07,
+      "loss": 3.4059,
+      "step": 6847
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.966241027730821e-07,
+      "loss": 3.4284,
+      "step": 6848
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.954842553188176e-07,
+      "loss": 3.4342,
+      "step": 6849
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.943450615841343e-07,
+      "loss": 3.3976,
+      "step": 6850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.93206521750471e-07,
+      "loss": 3.5037,
+      "step": 6851
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.920686359991615e-07,
+      "loss": 3.4606,
+      "step": 6852
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.909314045114343e-07,
+      "loss": 3.3637,
+      "step": 6853
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.897948274684154e-07,
+      "loss": 3.3745,
+      "step": 6854
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.886589050511257e-07,
+      "loss": 3.3275,
+      "step": 6855
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.8752363744048e-07,
+      "loss": 3.3553,
+      "step": 6856
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.863890248172918e-07,
+      "loss": 3.2944,
+      "step": 6857
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.852550673622684e-07,
+      "loss": 3.3349,
+      "step": 6858
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.841217652560141e-07,
+      "loss": 3.4689,
+      "step": 6859
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.82989118679028e-07,
+      "loss": 3.4204,
+      "step": 6860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.818571278117044e-07,
+      "loss": 3.4453,
+      "step": 6861
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.807257928343338e-07,
+      "loss": 3.4485,
+      "step": 6862
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.795951139271025e-07,
+      "loss": 3.3008,
+      "step": 6863
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.784650912700909e-07,
+      "loss": 3.5341,
+      "step": 6864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.773357250432774e-07,
+      "loss": 3.3468,
+      "step": 6865
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.762070154265346e-07,
+      "loss": 3.4202,
+      "step": 6866
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.750789625996276e-07,
+      "loss": 3.4532,
+      "step": 6867
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.739515667422211e-07,
+      "loss": 3.4181,
+      "step": 6868
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.728248280338741e-07,
+      "loss": 3.4708,
+      "step": 6869
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.716987466540394e-07,
+      "loss": 3.4097,
+      "step": 6870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.705733227820673e-07,
+      "loss": 3.3426,
+      "step": 6871
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.694485565972016e-07,
+      "loss": 3.3933,
+      "step": 6872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.68324448278583e-07,
+      "loss": 3.4849,
+      "step": 6873
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.672009980052442e-07,
+      "loss": 3.4183,
+      "step": 6874
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.660782059561168e-07,
+      "loss": 3.4422,
+      "step": 6875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.649560723100259e-07,
+      "loss": 3.4623,
+      "step": 6876
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.638345972456919e-07,
+      "loss": 3.4775,
+      "step": 6877
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.627137809417302e-07,
+      "loss": 3.3619,
+      "step": 6878
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.615936235766531e-07,
+      "loss": 3.4207,
+      "step": 6879
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.604741253288629e-07,
+      "loss": 3.41,
+      "step": 6880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.593552863766624e-07,
+      "loss": 3.4755,
+      "step": 6881
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.582371068982465e-07,
+      "loss": 3.3986,
+      "step": 6882
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.571195870717064e-07,
+      "loss": 3.3209,
+      "step": 6883
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.560027270750276e-07,
+      "loss": 3.3818,
+      "step": 6884
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.548865270860906e-07,
+      "loss": 3.5,
+      "step": 6885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.537709872826705e-07,
+      "loss": 3.4416,
+      "step": 6886
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.526561078424378e-07,
+      "loss": 3.3488,
+      "step": 6887
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.515418889429572e-07,
+      "loss": 3.4001,
+      "step": 6888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.50428330761689e-07,
+      "loss": 3.373,
+      "step": 6889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.493154334759884e-07,
+      "loss": 3.4472,
+      "step": 6890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.482031972631033e-07,
+      "loss": 3.3578,
+      "step": 6891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.470916223001779e-07,
+      "loss": 3.3473,
+      "step": 6892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.459807087642518e-07,
+      "loss": 3.3705,
+      "step": 6893
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.44870456832258e-07,
+      "loss": 3.4492,
+      "step": 6894
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.437608666810243e-07,
+      "loss": 3.4874,
+      "step": 6895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.426519384872733e-07,
+      "loss": 3.4091,
+      "step": 6896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.415436724276244e-07,
+      "loss": 3.3109,
+      "step": 6897
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.404360686785856e-07,
+      "loss": 3.4336,
+      "step": 6898
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.393291274165655e-07,
+      "loss": 3.3404,
+      "step": 6899
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.382228488178639e-07,
+      "loss": 3.3589,
+      "step": 6900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.371172330586769e-07,
+      "loss": 3.3047,
+      "step": 6901
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.360122803150938e-07,
+      "loss": 3.4091,
+      "step": 6902
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.349079907630991e-07,
+      "loss": 3.433,
+      "step": 6903
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.338043645785698e-07,
+      "loss": 3.3445,
+      "step": 6904
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.327014019372792e-07,
+      "loss": 3.4531,
+      "step": 6905
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.315991030148951e-07,
+      "loss": 3.4772,
+      "step": 6906
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.304974679869782e-07,
+      "loss": 3.4203,
+      "step": 6907
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.293964970289841e-07,
+      "loss": 3.3599,
+      "step": 6908
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.282961903162629e-07,
+      "loss": 3.4388,
+      "step": 6909
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.27196548024059e-07,
+      "loss": 3.4265,
+      "step": 6910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.260975703275104e-07,
+      "loss": 3.456,
+      "step": 6911
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.249992574016491e-07,
+      "loss": 3.4023,
+      "step": 6912
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.239016094214019e-07,
+      "loss": 3.4393,
+      "step": 6913
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.228046265615908e-07,
+      "loss": 3.4637,
+      "step": 6914
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.21708308996928e-07,
+      "loss": 3.5158,
+      "step": 6915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.206126569020234e-07,
+      "loss": 3.3929,
+      "step": 6916
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.195176704513796e-07,
+      "loss": 3.3671,
+      "step": 6917
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.184233498193939e-07,
+      "loss": 3.3689,
+      "step": 6918
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.173296951803561e-07,
+      "loss": 3.3734,
+      "step": 6919
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.162367067084526e-07,
+      "loss": 3.4872,
+      "step": 6920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.151443845777591e-07,
+      "loss": 3.4654,
+      "step": 6921
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.14052728962249e-07,
+      "loss": 3.369,
+      "step": 6922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.129617400357898e-07,
+      "loss": 3.4392,
+      "step": 6923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.118714179721404e-07,
+      "loss": 3.3606,
+      "step": 6924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.107817629449543e-07,
+      "loss": 3.4257,
+      "step": 6925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.096927751277805e-07,
+      "loss": 3.3446,
+      "step": 6926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.086044546940602e-07,
+      "loss": 3.2693,
+      "step": 6927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.07516801817127e-07,
+      "loss": 3.4113,
+      "step": 6928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.0642981667021e-07,
+      "loss": 3.3417,
+      "step": 6929
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.053434994264319e-07,
+      "loss": 3.5019,
+      "step": 6930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.042578502588089e-07,
+      "loss": 3.404,
+      "step": 6931
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.031728693402502e-07,
+      "loss": 3.4429,
+      "step": 6932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.020885568435594e-07,
+      "loss": 3.443,
+      "step": 6933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.010049129414333e-07,
+      "loss": 3.4917,
+      "step": 6934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.99921937806461e-07,
+      "loss": 3.4296,
+      "step": 6935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.988396316111275e-07,
+      "loss": 3.3378,
+      "step": 6936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.977579945278091e-07,
+      "loss": 3.3678,
+      "step": 6937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.966770267287783e-07,
+      "loss": 3.368,
+      "step": 6938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.955967283861965e-07,
+      "loss": 3.4413,
+      "step": 6939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.945170996721213e-07,
+      "loss": 3.4215,
+      "step": 6940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.934381407585045e-07,
+      "loss": 3.3796,
+      "step": 6941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.923598518171905e-07,
+      "loss": 3.4097,
+      "step": 6942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.912822330199149e-07,
+      "loss": 3.4511,
+      "step": 6943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.902052845383112e-07,
+      "loss": 3.3096,
+      "step": 6944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.891290065439e-07,
+      "loss": 3.4234,
+      "step": 6945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.880533992081002e-07,
+      "loss": 3.4688,
+      "step": 6946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.869784627022215e-07,
+      "loss": 3.4189,
+      "step": 6947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.859041971974668e-07,
+      "loss": 3.394,
+      "step": 6948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.848306028649339e-07,
+      "loss": 3.3799,
+      "step": 6949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.837576798756125e-07,
+      "loss": 3.3498,
+      "step": 6950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.826854284003837e-07,
+      "loss": 3.5026,
+      "step": 6951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.816138486100245e-07,
+      "loss": 3.3836,
+      "step": 6952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.805429406752025e-07,
+      "loss": 3.3675,
+      "step": 6953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.794727047664807e-07,
+      "loss": 3.3606,
+      "step": 6954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.784031410543135e-07,
+      "loss": 3.3744,
+      "step": 6955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.773342497090486e-07,
+      "loss": 3.4786,
+      "step": 6956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.762660309009273e-07,
+      "loss": 3.3558,
+      "step": 6957
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.751984848000804e-07,
+      "loss": 3.4144,
+      "step": 6958
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.741316115765368e-07,
+      "loss": 3.4281,
+      "step": 6959
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.730654114002156e-07,
+      "loss": 3.4756,
+      "step": 6960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.719998844409276e-07,
+      "loss": 3.4497,
+      "step": 6961
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.7093503086838e-07,
+      "loss": 3.4441,
+      "step": 6962
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.698708508521669e-07,
+      "loss": 3.4257,
+      "step": 6963
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.6880734456178e-07,
+      "loss": 3.3864,
+      "step": 6964
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.677445121666021e-07,
+      "loss": 3.4152,
+      "step": 6965
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.666823538359098e-07,
+      "loss": 3.3907,
+      "step": 6966
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.656208697388695e-07,
+      "loss": 3.448,
+      "step": 6967
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.645600600445452e-07,
+      "loss": 3.4228,
+      "step": 6968
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.634999249218866e-07,
+      "loss": 3.3469,
+      "step": 6969
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.62440464539741e-07,
+      "loss": 3.5341,
+      "step": 6970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.613816790668477e-07,
+      "loss": 3.3347,
+      "step": 6971
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.603235686718374e-07,
+      "loss": 3.4145,
+      "step": 6972
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.592661335232338e-07,
+      "loss": 3.406,
+      "step": 6973
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.58209373789453e-07,
+      "loss": 3.4835,
+      "step": 6974
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.571532896388028e-07,
+      "loss": 3.4568,
+      "step": 6975
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.560978812394837e-07,
+      "loss": 3.4004,
+      "step": 6976
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.550431487595894e-07,
+      "loss": 3.4843,
+      "step": 6977
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.539890923671061e-07,
+      "loss": 3.4411,
+      "step": 6978
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.529357122299108e-07,
+      "loss": 3.4401,
+      "step": 6979
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.518830085157735e-07,
+      "loss": 3.3883,
+      "step": 6980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.508309813923587e-07,
+      "loss": 3.3902,
+      "step": 6981
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.497796310272182e-07,
+      "loss": 3.4455,
+      "step": 6982
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.487289575877981e-07,
+      "loss": 3.3687,
+      "step": 6983
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.476789612414414e-07,
+      "loss": 3.4826,
+      "step": 6984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.466296421553782e-07,
+      "loss": 3.32,
+      "step": 6985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.455810004967301e-07,
+      "loss": 3.4009,
+      "step": 6986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.445330364325131e-07,
+      "loss": 3.3544,
+      "step": 6987
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.434857501296356e-07,
+      "loss": 3.4103,
+      "step": 6988
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.42439141754896e-07,
+      "loss": 3.4104,
+      "step": 6989
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.413932114749872e-07,
+      "loss": 3.4425,
+      "step": 6990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.403479594564916e-07,
+      "loss": 3.358,
+      "step": 6991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.393033858658871e-07,
+      "loss": 3.3458,
+      "step": 6992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.382594908695384e-07,
+      "loss": 3.4157,
+      "step": 6993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.372162746337059e-07,
+      "loss": 3.3617,
+      "step": 6994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.361737373245415e-07,
+      "loss": 3.4002,
+      "step": 6995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.351318791080881e-07,
+      "loss": 3.3365,
+      "step": 6996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.340907001502806e-07,
+      "loss": 3.4962,
+      "step": 6997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.330502006169471e-07,
+      "loss": 3.4014,
+      "step": 6998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.320103806738043e-07,
+      "loss": 3.3397,
+      "step": 6999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.309712404864639e-07,
+      "loss": 3.3601,
+      "step": 7000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.299327802204276e-07,
+      "loss": 3.3394,
+      "step": 7001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.288950000410893e-07,
+      "loss": 3.3613,
+      "step": 7002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.278579001137348e-07,
+      "loss": 3.3294,
+      "step": 7003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.268214806035423e-07,
+      "loss": 3.4779,
+      "step": 7004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.257857416755787e-07,
+      "loss": 3.5047,
+      "step": 7005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.247506834948054e-07,
+      "loss": 3.4114,
+      "step": 7006
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.237163062260732e-07,
+      "loss": 3.407,
+      "step": 7007
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.226826100341283e-07,
+      "loss": 3.4852,
+      "step": 7008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.216495950836056e-07,
+      "loss": 3.4472,
+      "step": 7009
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.206172615390294e-07,
+      "loss": 3.3636,
+      "step": 7010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.195856095648185e-07,
+      "loss": 3.441,
+      "step": 7011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.185546393252835e-07,
+      "loss": 3.4885,
+      "step": 7012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.175243509846247e-07,
+      "loss": 3.3414,
+      "step": 7013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.164947447069343e-07,
+      "loss": 3.3723,
+      "step": 7014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.15465820656196e-07,
+      "loss": 3.4517,
+      "step": 7015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.144375789962865e-07,
+      "loss": 3.3981,
+      "step": 7016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.134100198909699e-07,
+      "loss": 3.2772,
+      "step": 7017
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.123831435039047e-07,
+      "loss": 3.4283,
+      "step": 7018
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.113569499986401e-07,
+      "loss": 3.4121,
+      "step": 7019
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.103314395386158e-07,
+      "loss": 3.3936,
+      "step": 7020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.093066122871633e-07,
+      "loss": 3.336,
+      "step": 7021
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.082824684075068e-07,
+      "loss": 3.4483,
+      "step": 7022
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.072590080627572e-07,
+      "loss": 3.336,
+      "step": 7023
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.062362314159211e-07,
+      "loss": 3.4277,
+      "step": 7024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.052141386298944e-07,
+      "loss": 3.4149,
+      "step": 7025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.041927298674633e-07,
+      "loss": 3.5287,
+      "step": 7026
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.031720052913071e-07,
+      "loss": 3.3059,
+      "step": 7027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.021519650639952e-07,
+      "loss": 3.3525,
+      "step": 7028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.011326093479859e-07,
+      "loss": 3.4493,
+      "step": 7029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.001139383056315e-07,
+      "loss": 3.432,
+      "step": 7030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.990959520991742e-07,
+      "loss": 3.4713,
+      "step": 7031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.980786508907456e-07,
+      "loss": 3.3196,
+      "step": 7032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.970620348423729e-07,
+      "loss": 3.3679,
+      "step": 7033
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.960461041159677e-07,
+      "loss": 3.4042,
+      "step": 7034
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.950308588733367e-07,
+      "loss": 3.5068,
+      "step": 7035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.940162992761762e-07,
+      "loss": 3.3295,
+      "step": 7036
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.93002425486074e-07,
+      "loss": 3.3371,
+      "step": 7037
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.919892376645076e-07,
+      "loss": 3.4682,
+      "step": 7038
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.909767359728475e-07,
+      "loss": 3.4737,
+      "step": 7039
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.899649205723507e-07,
+      "loss": 3.4737,
+      "step": 7040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.889537916241679e-07,
+      "loss": 3.4261,
+      "step": 7041
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.879433492893412e-07,
+      "loss": 3.3733,
+      "step": 7042
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.869335937288007e-07,
+      "loss": 3.3637,
+      "step": 7043
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.859245251033697e-07,
+      "loss": 3.4734,
+      "step": 7044
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8491614357376e-07,
+      "loss": 3.4508,
+      "step": 7045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.839084493005771e-07,
+      "loss": 3.434,
+      "step": 7046
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.829014424443114e-07,
+      "loss": 3.4148,
+      "step": 7047
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.818951231653493e-07,
+      "loss": 3.392,
+      "step": 7048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.808894916239656e-07,
+      "loss": 3.3671,
+      "step": 7049
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.798845479803246e-07,
+      "loss": 3.3732,
+      "step": 7050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.788802923944832e-07,
+      "loss": 3.3376,
+      "step": 7051
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.778767250263884e-07,
+      "loss": 3.3352,
+      "step": 7052
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.768738460358742e-07,
+      "loss": 3.3227,
+      "step": 7053
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.758716555826683e-07,
+      "loss": 3.3623,
+      "step": 7054
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.748701538263891e-07,
+      "loss": 3.3146,
+      "step": 7055
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.738693409265423e-07,
+      "loss": 3.3623,
+      "step": 7056
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.728692170425289e-07,
+      "loss": 3.4662,
+      "step": 7057
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.718697823336339e-07,
+      "loss": 3.4943,
+      "step": 7058
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.708710369590371e-07,
+      "loss": 3.4337,
+      "step": 7059
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.698729810778065e-07,
+      "loss": 3.4975,
+      "step": 7060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.688756148489017e-07,
+      "loss": 3.444,
+      "step": 7061
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.678789384311707e-07,
+      "loss": 3.5383,
+      "step": 7062
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.668829519833536e-07,
+      "loss": 3.432,
+      "step": 7063
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.658876556640781e-07,
+      "loss": 3.3959,
+      "step": 7064
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.64893049631864e-07,
+      "loss": 3.3559,
+      "step": 7065
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.638991340451212e-07,
+      "loss": 3.427,
+      "step": 7066
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.629059090621482e-07,
+      "loss": 3.4449,
+      "step": 7067
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.619133748411344e-07,
+      "loss": 3.3548,
+      "step": 7068
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.609215315401607e-07,
+      "loss": 3.403,
+      "step": 7069
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.599303793171946e-07,
+      "loss": 3.4567,
+      "step": 7070
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.589399183300954e-07,
+      "loss": 3.3853,
+      "step": 7071
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.579501487366125e-07,
+      "loss": 3.4955,
+      "step": 7072
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.569610706943852e-07,
+      "loss": 3.2642,
+      "step": 7073
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.559726843609421e-07,
+      "loss": 3.394,
+      "step": 7074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.549849898937022e-07,
+      "loss": 3.5509,
+      "step": 7075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.539979874499747e-07,
+      "loss": 3.433,
+      "step": 7076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.53011677186956e-07,
+      "loss": 3.4469,
+      "step": 7077
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.52026059261735e-07,
+      "loss": 3.4222,
+      "step": 7078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.510411338312894e-07,
+      "loss": 3.2163,
+      "step": 7079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.500569010524871e-07,
+      "loss": 3.4659,
+      "step": 7080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.490733610820854e-07,
+      "loss": 3.4173,
+      "step": 7081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.480905140767302e-07,
+      "loss": 3.2601,
+      "step": 7082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.471083601929589e-07,
+      "loss": 3.478,
+      "step": 7083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.461268995871967e-07,
+      "loss": 3.3833,
+      "step": 7084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.451461324157604e-07,
+      "loss": 3.3751,
+      "step": 7085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.441660588348541e-07,
+      "loss": 3.5429,
+      "step": 7086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.431866790005737e-07,
+      "loss": 3.2894,
+      "step": 7087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.422079930689024e-07,
+      "loss": 3.3883,
+      "step": 7088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.412300011957134e-07,
+      "loss": 3.4769,
+      "step": 7089
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.402527035367711e-07,
+      "loss": 3.413,
+      "step": 7090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.392761002477277e-07,
+      "loss": 3.4457,
+      "step": 7091
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.383001914841252e-07,
+      "loss": 3.3496,
+      "step": 7092
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.373249774013962e-07,
+      "loss": 3.4483,
+      "step": 7093
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.36350458154859e-07,
+      "loss": 3.4001,
+      "step": 7094
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.353766338997253e-07,
+      "loss": 3.3922,
+      "step": 7095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.344035047910941e-07,
+      "loss": 3.397,
+      "step": 7096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.334310709839542e-07,
+      "loss": 3.3849,
+      "step": 7097
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.324593326331834e-07,
+      "loss": 3.4269,
+      "step": 7098
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.314882898935492e-07,
+      "loss": 3.4689,
+      "step": 7099
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.305179429197094e-07,
+      "loss": 3.4251,
+      "step": 7100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.295482918662066e-07,
+      "loss": 3.3988,
+      "step": 7101
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.285793368874771e-07,
+      "loss": 3.4276,
+      "step": 7102
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.276110781378448e-07,
+      "loss": 3.3993,
+      "step": 7103
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.266435157715222e-07,
+      "loss": 3.4686,
+      "step": 7104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.256766499426126e-07,
+      "loss": 3.546,
+      "step": 7105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.247104808051058e-07,
+      "loss": 3.382,
+      "step": 7106
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.237450085128826e-07,
+      "loss": 3.359,
+      "step": 7107
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.227802332197125e-07,
+      "loss": 3.3371,
+      "step": 7108
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.218161550792535e-07,
+      "loss": 3.2978,
+      "step": 7109
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.208527742450526e-07,
+      "loss": 3.4455,
+      "step": 7110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.198900908705468e-07,
+      "loss": 3.4208,
+      "step": 7111
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.18928105109059e-07,
+      "loss": 3.4382,
+      "step": 7112
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.179668171138048e-07,
+      "loss": 3.4332,
+      "step": 7113
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.170062270378862e-07,
+      "loss": 3.4688,
+      "step": 7114
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.16046335034296e-07,
+      "loss": 3.3159,
+      "step": 7115
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.150871412559134e-07,
+      "loss": 3.3839,
+      "step": 7116
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.14128645855509e-07,
+      "loss": 3.4991,
+      "step": 7117
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.131708489857391e-07,
+      "loss": 3.4377,
+      "step": 7118
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.12213750799151e-07,
+      "loss": 3.398,
+      "step": 7119
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.112573514481807e-07,
+      "loss": 3.443,
+      "step": 7120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.103016510851518e-07,
+      "loss": 3.3627,
+      "step": 7121
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.09346649862278e-07,
+      "loss": 3.4029,
+      "step": 7122
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.083923479316606e-07,
+      "loss": 3.4892,
+      "step": 7123
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.074387454452891e-07,
+      "loss": 3.3999,
+      "step": 7124
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.064858425550419e-07,
+      "loss": 3.46,
+      "step": 7125
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.055336394126876e-07,
+      "loss": 3.44,
+      "step": 7126
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.045821361698811e-07,
+      "loss": 3.4134,
+      "step": 7127
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.036313329781668e-07,
+      "loss": 3.4414,
+      "step": 7128
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.026812299889778e-07,
+      "loss": 3.5126,
+      "step": 7129
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.01731827353636e-07,
+      "loss": 3.3441,
+      "step": 7130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.007831252233503e-07,
+      "loss": 3.455,
+      "step": 7131
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9983512374922e-07,
+      "loss": 3.3894,
+      "step": 7132
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.988878230822309e-07,
+      "loss": 3.404,
+      "step": 7133
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.979412233732578e-07,
+      "loss": 3.4111,
+      "step": 7134
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.969953247730665e-07,
+      "loss": 3.3817,
+      "step": 7135
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.96050127432305e-07,
+      "loss": 3.43,
+      "step": 7136
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.951056315015158e-07,
+      "loss": 3.4461,
+      "step": 7137
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.941618371311264e-07,
+      "loss": 3.4239,
+      "step": 7138
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.932187444714533e-07,
+      "loss": 3.5107,
+      "step": 7139
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.922763536727023e-07,
+      "loss": 3.4552,
+      "step": 7140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.913346648849666e-07,
+      "loss": 3.3455,
+      "step": 7141
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.903936782582253e-07,
+      "loss": 3.3834,
+      "step": 7142
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.894533939423497e-07,
+      "loss": 3.3866,
+      "step": 7143
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.885138120870965e-07,
+      "loss": 3.4269,
+      "step": 7144
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.875749328421121e-07,
+      "loss": 3.3644,
+      "step": 7145
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.866367563569298e-07,
+      "loss": 3.3774,
+      "step": 7146
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.856992827809726e-07,
+      "loss": 3.4167,
+      "step": 7147
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.847625122635481e-07,
+      "loss": 3.3991,
+      "step": 7148
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.838264449538561e-07,
+      "loss": 3.3682,
+      "step": 7149
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.828910810009814e-07,
+      "loss": 3.2856,
+      "step": 7150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.819564205538985e-07,
+      "loss": 3.2845,
+      "step": 7151
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.810224637614692e-07,
+      "loss": 3.3667,
+      "step": 7152
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.800892107724438e-07,
+      "loss": 3.3437,
+      "step": 7153
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.791566617354599e-07,
+      "loss": 3.4086,
+      "step": 7154
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.782248167990406e-07,
+      "loss": 3.4665,
+      "step": 7155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.772936761116027e-07,
+      "loss": 3.3946,
+      "step": 7156
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.763632398214464e-07,
+      "loss": 3.2755,
+      "step": 7157
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.754335080767609e-07,
+      "loss": 3.3651,
+      "step": 7158
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.745044810256222e-07,
+      "loss": 3.3948,
+      "step": 7159
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.735761588159949e-07,
+      "loss": 3.3916,
+      "step": 7160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.726485415957323e-07,
+      "loss": 3.3924,
+      "step": 7161
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.717216295125743e-07,
+      "loss": 3.3575,
+      "step": 7162
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.707954227141482e-07,
+      "loss": 3.2808,
+      "step": 7163
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.698699213479697e-07,
+      "loss": 3.3842,
+      "step": 7164
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.68945125561443e-07,
+      "loss": 3.3135,
+      "step": 7165
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.680210355018572e-07,
+      "loss": 3.4486,
+      "step": 7166
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.670976513163911e-07,
+      "loss": 3.4823,
+      "step": 7167
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.661749731521104e-07,
+      "loss": 3.2932,
+      "step": 7168
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.652530011559692e-07,
+      "loss": 3.4489,
+      "step": 7169
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.643317354748079e-07,
+      "loss": 3.564,
+      "step": 7170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.634111762553563e-07,
+      "loss": 3.3164,
+      "step": 7171
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.624913236442287e-07,
+      "loss": 3.3041,
+      "step": 7172
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.615721777879285e-07,
+      "loss": 3.441,
+      "step": 7173
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.606537388328481e-07,
+      "loss": 3.397,
+      "step": 7174
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.597360069252645e-07,
+      "loss": 3.4856,
+      "step": 7175
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.588189822113438e-07,
+      "loss": 3.4402,
+      "step": 7176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.579026648371399e-07,
+      "loss": 3.4914,
+      "step": 7177
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.569870549485917e-07,
+      "loss": 3.394,
+      "step": 7178
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.560721526915263e-07,
+      "loss": 3.3454,
+      "step": 7179
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.551579582116612e-07,
+      "loss": 3.4503,
+      "step": 7180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.542444716545975e-07,
+      "loss": 3.4529,
+      "step": 7181
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.533316931658256e-07,
+      "loss": 3.4467,
+      "step": 7182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.524196228907203e-07,
+      "loss": 3.4683,
+      "step": 7183
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.515082609745465e-07,
+      "loss": 3.3327,
+      "step": 7184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.505976075624553e-07,
+      "loss": 3.4722,
+      "step": 7185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.496876627994851e-07,
+      "loss": 3.3765,
+      "step": 7186
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.487784268305613e-07,
+      "loss": 3.4442,
+      "step": 7187
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.478698998004967e-07,
+      "loss": 3.3593,
+      "step": 7188
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.469620818539916e-07,
+      "loss": 3.5173,
+      "step": 7189
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.460549731356312e-07,
+      "loss": 3.3565,
+      "step": 7190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.451485737898892e-07,
+      "loss": 3.4574,
+      "step": 7191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.442428839611275e-07,
+      "loss": 3.4125,
+      "step": 7192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.433379037935932e-07,
+      "loss": 3.4023,
+      "step": 7193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.424336334314212e-07,
+      "loss": 3.2766,
+      "step": 7194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.41530073018634e-07,
+      "loss": 3.3941,
+      "step": 7195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.406272226991383e-07,
+      "loss": 3.3941,
+      "step": 7196
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.397250826167305e-07,
+      "loss": 3.5626,
+      "step": 7197
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.388236529150936e-07,
+      "loss": 3.5134,
+      "step": 7198
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.379229337377967e-07,
+      "loss": 3.4375,
+      "step": 7199
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.370229252282949e-07,
+      "loss": 3.4401,
+      "step": 7200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.361236275299331e-07,
+      "loss": 3.4304,
+      "step": 7201
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.352250407859389e-07,
+      "loss": 3.3708,
+      "step": 7202
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.343271651394294e-07,
+      "loss": 3.4561,
+      "step": 7203
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.334300007334065e-07,
+      "loss": 3.3568,
+      "step": 7204
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.325335477107629e-07,
+      "loss": 3.4882,
+      "step": 7205
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.316378062142751e-07,
+      "loss": 3.3013,
+      "step": 7206
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.307427763866041e-07,
+      "loss": 3.3888,
+      "step": 7207
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.298484583703006e-07,
+      "loss": 3.383,
+      "step": 7208
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.28954852307802e-07,
+      "loss": 3.4302,
+      "step": 7209
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.28061958341431e-07,
+      "loss": 3.4022,
+      "step": 7210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.271697766133971e-07,
+      "loss": 3.4221,
+      "step": 7211
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.262783072657984e-07,
+      "loss": 3.3399,
+      "step": 7212
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.253875504406147e-07,
+      "loss": 3.4071,
+      "step": 7213
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.244975062797176e-07,
+      "loss": 3.376,
+      "step": 7214
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.236081749248623e-07,
+      "loss": 3.4276,
+      "step": 7215
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.227195565176906e-07,
+      "loss": 3.3848,
+      "step": 7216
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.218316511997329e-07,
+      "loss": 3.4498,
+      "step": 7217
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.209444591124029e-07,
+      "loss": 3.4387,
+      "step": 7218
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.200579803970041e-07,
+      "loss": 3.4418,
+      "step": 7219
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.191722151947227e-07,
+      "loss": 3.3515,
+      "step": 7220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.18287163646633e-07,
+      "loss": 3.3502,
+      "step": 7221
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.174028258936969e-07,
+      "loss": 3.446,
+      "step": 7222
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.165192020767612e-07,
+      "loss": 3.4132,
+      "step": 7223
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.156362923365587e-07,
+      "loss": 3.3422,
+      "step": 7224
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.147540968137105e-07,
+      "loss": 3.3513,
+      "step": 7225
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1387261564872e-07,
+      "loss": 3.3687,
+      "step": 7226
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.129918489819807e-07,
+      "loss": 3.4859,
+      "step": 7227
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.121117969537704e-07,
+      "loss": 3.3446,
+      "step": 7228
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.112324597042533e-07,
+      "loss": 3.3838,
+      "step": 7229
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.103538373734828e-07,
+      "loss": 3.3527,
+      "step": 7230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.094759301013918e-07,
+      "loss": 3.3524,
+      "step": 7231
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.085987380278057e-07,
+      "loss": 3.377,
+      "step": 7232
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.077222612924321e-07,
+      "loss": 3.4744,
+      "step": 7233
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.068465000348666e-07,
+      "loss": 3.4323,
+      "step": 7234
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.059714543945904e-07,
+      "loss": 3.3726,
+      "step": 7235
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.050971245109709e-07,
+      "loss": 3.4455,
+      "step": 7236
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.042235105232601e-07,
+      "loss": 3.4183,
+      "step": 7237
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.033506125705978e-07,
+      "loss": 3.298,
+      "step": 7238
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.024784307920089e-07,
+      "loss": 3.421,
+      "step": 7239
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.016069653264044e-07,
+      "loss": 3.428,
+      "step": 7240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.007362163125812e-07,
+      "loss": 3.2721,
+      "step": 7241
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.998661838892232e-07,
+      "loss": 3.465,
+      "step": 7242
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.98996868194897e-07,
+      "loss": 3.506,
+      "step": 7243
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.981282693680584e-07,
+      "loss": 3.4522,
+      "step": 7244
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.972603875470472e-07,
+      "loss": 3.4012,
+      "step": 7245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.963932228700896e-07,
+      "loss": 3.4287,
+      "step": 7246
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.955267754752974e-07,
+      "loss": 3.5571,
+      "step": 7247
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.946610455006689e-07,
+      "loss": 3.4049,
+      "step": 7248
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.93796033084088e-07,
+      "loss": 3.4285,
+      "step": 7249
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.929317383633225e-07,
+      "loss": 3.4331,
+      "step": 7250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.920681614760269e-07,
+      "loss": 3.3366,
+      "step": 7251
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.91205302559743e-07,
+      "loss": 3.3746,
+      "step": 7252
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.903431617518955e-07,
+      "loss": 3.3993,
+      "step": 7253
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.894817391897988e-07,
+      "loss": 3.4316,
+      "step": 7254
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.88621035010648e-07,
+      "loss": 3.4168,
+      "step": 7255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.877610493515262e-07,
+      "loss": 3.4248,
+      "step": 7256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.869017823494026e-07,
+      "loss": 3.3612,
+      "step": 7257
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.860432341411308e-07,
+      "loss": 3.3076,
+      "step": 7258
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.851854048634513e-07,
+      "loss": 3.4921,
+      "step": 7259
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.843282946529887e-07,
+      "loss": 3.3863,
+      "step": 7260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.834719036462532e-07,
+      "loss": 3.4534,
+      "step": 7261
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.826162319796401e-07,
+      "loss": 3.4437,
+      "step": 7262
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.817612797894322e-07,
+      "loss": 3.4543,
+      "step": 7263
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.80907047211796e-07,
+      "loss": 3.4749,
+      "step": 7264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.800535343827834e-07,
+      "loss": 3.4047,
+      "step": 7265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.792007414383331e-07,
+      "loss": 3.4221,
+      "step": 7266
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.783486685142663e-07,
+      "loss": 3.4368,
+      "step": 7267
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.77497315746292e-07,
+      "loss": 3.3085,
+      "step": 7268
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.76646683270004e-07,
+      "loss": 3.4034,
+      "step": 7269
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.757967712208811e-07,
+      "loss": 3.4275,
+      "step": 7270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7494757973428784e-07,
+      "loss": 3.3312,
+      "step": 7271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.740991089454727e-07,
+      "loss": 3.4289,
+      "step": 7272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7325135898957153e-07,
+      "loss": 3.4682,
+      "step": 7273
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7240433000160233e-07,
+      "loss": 3.4576,
+      "step": 7274
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7155802211647115e-07,
+      "loss": 3.3596,
+      "step": 7275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7071243546896795e-07,
+      "loss": 3.3773,
+      "step": 7276
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.698675701937677e-07,
+      "loss": 3.3122,
+      "step": 7277
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.690234264254312e-07,
+      "loss": 3.3233,
+      "step": 7278
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6818000429840305e-07,
+      "loss": 3.4578,
+      "step": 7279
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.673373039470147e-07,
+      "loss": 3.4628,
+      "step": 7280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6649532550548103e-07,
+      "loss": 3.3152,
+      "step": 7281
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6565406910790256e-07,
+      "loss": 3.4173,
+      "step": 7282
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.648135348882654e-07,
+      "loss": 3.362,
+      "step": 7283
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.639737229804403e-07,
+      "loss": 3.2941,
+      "step": 7284
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.631346335181819e-07,
+      "loss": 3.5706,
+      "step": 7285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6229626663512993e-07,
+      "loss": 3.3614,
+      "step": 7286
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6145862246481154e-07,
+      "loss": 3.4289,
+      "step": 7287
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6062170114063556e-07,
+      "loss": 3.3595,
+      "step": 7288
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5978550279589805e-07,
+      "loss": 3.3785,
+      "step": 7289
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5895002756377926e-07,
+      "loss": 3.5062,
+      "step": 7290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5811527557734214e-07,
+      "loss": 3.2883,
+      "step": 7291
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5728124696953815e-07,
+      "loss": 3.3316,
+      "step": 7292
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.564479418732004e-07,
+      "loss": 3.3888,
+      "step": 7293
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.556153604210484e-07,
+      "loss": 3.4283,
+      "step": 7294
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.547835027456865e-07,
+      "loss": 3.3399,
+      "step": 7295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.539523689796044e-07,
+      "loss": 3.2946,
+      "step": 7296
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.531219592551728e-07,
+      "loss": 3.437,
+      "step": 7297
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.522922737046509e-07,
+      "loss": 3.3857,
+      "step": 7298
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.514633124601814e-07,
+      "loss": 3.1689,
+      "step": 7299
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5063507565379195e-07,
+      "loss": 3.4308,
+      "step": 7300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.498075634173943e-07,
+      "loss": 3.4122,
+      "step": 7301
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.489807758827852e-07,
+      "loss": 3.6108,
+      "step": 7302
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4815471318164494e-07,
+      "loss": 3.3637,
+      "step": 7303
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.473293754455399e-07,
+      "loss": 3.4708,
+      "step": 7304
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4650476280592045e-07,
+      "loss": 3.2812,
+      "step": 7305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.456808753941205e-07,
+      "loss": 3.439,
+      "step": 7306
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.448577133413601e-07,
+      "loss": 3.3597,
+      "step": 7307
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.440352767787437e-07,
+      "loss": 3.5025,
+      "step": 7308
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4321356583725715e-07,
+      "loss": 3.5063,
+      "step": 7309
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4239258064777413e-07,
+      "loss": 3.3472,
+      "step": 7310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.415723213410517e-07,
+      "loss": 3.3647,
+      "step": 7311
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4075278804773134e-07,
+      "loss": 3.3957,
+      "step": 7312
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.399339808983388e-07,
+      "loss": 3.408,
+      "step": 7313
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3911590002328476e-07,
+      "loss": 3.3756,
+      "step": 7314
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3829854555286213e-07,
+      "loss": 3.3999,
+      "step": 7315
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.374819176172501e-07,
+      "loss": 3.4542,
+      "step": 7316
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.366660163465125e-07,
+      "loss": 3.4848,
+      "step": 7317
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.358508418705959e-07,
+      "loss": 3.3598,
+      "step": 7318
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3503639431933196e-07,
+      "loss": 3.4377,
+      "step": 7319
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.342226738224381e-07,
+      "loss": 3.3656,
+      "step": 7320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3340968050951114e-07,
+      "loss": 3.3766,
+      "step": 7321
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3259741451003643e-07,
+      "loss": 3.3689,
+      "step": 7322
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.317858759533833e-07,
+      "loss": 3.3977,
+      "step": 7323
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3097506496880325e-07,
+      "loss": 3.4213,
+      "step": 7324
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.301649816854331e-07,
+      "loss": 3.3819,
+      "step": 7325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.29355626232294e-07,
+      "loss": 3.4092,
+      "step": 7326
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2854699873828845e-07,
+      "loss": 3.3939,
+      "step": 7327
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.277390993322078e-07,
+      "loss": 3.3517,
+      "step": 7328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2693192814272424e-07,
+      "loss": 3.3999,
+      "step": 7329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2612548529839425e-07,
+      "loss": 3.4213,
+      "step": 7330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2531977092765964e-07,
+      "loss": 3.4388,
+      "step": 7331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.245147851588438e-07,
+      "loss": 3.4544,
+      "step": 7332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.237105281201559e-07,
+      "loss": 3.3273,
+      "step": 7333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2290699993968897e-07,
+      "loss": 3.3581,
+      "step": 7334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.221042007454196e-07,
+      "loss": 3.3164,
+      "step": 7335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2130213066520763e-07,
+      "loss": 3.4941,
+      "step": 7336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2050078982679875e-07,
+      "loss": 3.4528,
+      "step": 7337
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.197001783578214e-07,
+      "loss": 3.2756,
+      "step": 7338
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.189002963857858e-07,
+      "loss": 3.47,
+      "step": 7339
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.181011440380889e-07,
+      "loss": 3.2908,
+      "step": 7340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1730272144201057e-07,
+      "loss": 3.2203,
+      "step": 7341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.165050287247141e-07,
+      "loss": 3.4422,
+      "step": 7342
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.157080660132468e-07,
+      "loss": 3.4734,
+      "step": 7343
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.149118334345403e-07,
+      "loss": 3.3625,
+      "step": 7344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1411633111540785e-07,
+      "loss": 3.3665,
+      "step": 7345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1332155918254845e-07,
+      "loss": 3.4217,
+      "step": 7346
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1252751776254373e-07,
+      "loss": 3.3126,
+      "step": 7347
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1173420698186027e-07,
+      "loss": 3.2068,
+      "step": 7348
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1094162696684705e-07,
+      "loss": 3.3433,
+      "step": 7349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.10149777843738e-07,
+      "loss": 3.5453,
+      "step": 7350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.093586597386473e-07,
+      "loss": 3.4051,
+      "step": 7351
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.085682727775753e-07,
+      "loss": 3.3384,
+      "step": 7352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0777861708640787e-07,
+      "loss": 3.3647,
+      "step": 7353
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0698969279091115e-07,
+      "loss": 3.3712,
+      "step": 7354
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.062015000167363e-07,
+      "loss": 3.36,
+      "step": 7355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.054140388894162e-07,
+      "loss": 3.5207,
+      "step": 7356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0462730953436945e-07,
+      "loss": 3.3522,
+      "step": 7357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0384131207689634e-07,
+      "loss": 3.5275,
+      "step": 7358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0305604664218233e-07,
+      "loss": 3.2978,
+      "step": 7359
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.022715133552951e-07,
+      "loss": 3.3686,
+      "step": 7360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0148771234118587e-07,
+      "loss": 3.4203,
+      "step": 7361
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.007046437246903e-07,
+      "loss": 3.481,
+      "step": 7362
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9992230763052476e-07,
+      "loss": 3.3825,
+      "step": 7363
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9914070418329123e-07,
+      "loss": 3.5054,
+      "step": 7364
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9835983350747507e-07,
+      "loss": 3.3799,
+      "step": 7365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9757969572744417e-07,
+      "loss": 3.4284,
+      "step": 7366
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.96800290967449e-07,
+      "loss": 3.4683,
+      "step": 7367
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.960216193516253e-07,
+      "loss": 3.448,
+      "step": 7368
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9524368100399e-07,
+      "loss": 3.4144,
+      "step": 7369
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.94466476048444e-07,
+      "loss": 3.4749,
+      "step": 7370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.936900046087716e-07,
+      "loss": 3.3915,
+      "step": 7371
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9291426680863997e-07,
+      "loss": 3.3203,
+      "step": 7372
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9213926277160084e-07,
+      "loss": 3.3458,
+      "step": 7373
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9136499262108716e-07,
+      "loss": 3.3892,
+      "step": 7374
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9059145648041474e-07,
+      "loss": 3.464,
+      "step": 7375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.898186544727833e-07,
+      "loss": 3.389,
+      "step": 7376
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8904658672127726e-07,
+      "loss": 3.4229,
+      "step": 7377
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.882752533488621e-07,
+      "loss": 3.4276,
+      "step": 7378
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.875046544783878e-07,
+      "loss": 3.3227,
+      "step": 7379
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8673479023258464e-07,
+      "loss": 3.3882,
+      "step": 7380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.859656607340678e-07,
+      "loss": 3.372,
+      "step": 7381
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8519726610533603e-07,
+      "loss": 3.4239,
+      "step": 7382
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.844296064687697e-07,
+      "loss": 3.4008,
+      "step": 7383
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.836626819466338e-07,
+      "loss": 3.3998,
+      "step": 7384
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8289649266107454e-07,
+      "loss": 3.533,
+      "step": 7385
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.821310387341215e-07,
+      "loss": 3.4261,
+      "step": 7386
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.813663202876866e-07,
+      "loss": 3.3505,
+      "step": 7387
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8060233744356634e-07,
+      "loss": 3.341,
+      "step": 7388
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.798390903234389e-07,
+      "loss": 3.3941,
+      "step": 7389
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.790765790488654e-07,
+      "loss": 3.4117,
+      "step": 7390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7831480374128926e-07,
+      "loss": 3.4454,
+      "step": 7391
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.775537645220384e-07,
+      "loss": 3.3856,
+      "step": 7392
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7679346151232087e-07,
+      "loss": 3.3055,
+      "step": 7393
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.760338948332293e-07,
+      "loss": 3.3272,
+      "step": 7394
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7527506460573904e-07,
+      "loss": 3.3531,
+      "step": 7395
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7451697095070736e-07,
+      "loss": 3.4429,
+      "step": 7396
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.737596139888755e-07,
+      "loss": 3.3724,
+      "step": 7397
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.730029938408658e-07,
+      "loss": 3.4091,
+      "step": 7398
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7224711062718355e-07,
+      "loss": 3.3778,
+      "step": 7399
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7149196446821745e-07,
+      "loss": 3.3612,
+      "step": 7400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.707375554842374e-07,
+      "loss": 3.4424,
+      "step": 7401
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.699838837953984e-07,
+      "loss": 3.4569,
+      "step": 7402
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6923094952173723e-07,
+      "loss": 3.4039,
+      "step": 7403
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.684787527831707e-07,
+      "loss": 3.459,
+      "step": 7404
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6772729369950024e-07,
+      "loss": 3.3944,
+      "step": 7405
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6697657239040953e-07,
+      "loss": 3.4516,
+      "step": 7406
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6622658897546514e-07,
+      "loss": 3.4021,
+      "step": 7407
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6547734357411603e-07,
+      "loss": 3.421,
+      "step": 7408
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.647288363056933e-07,
+      "loss": 3.3736,
+      "step": 7409
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.639810672894095e-07,
+      "loss": 3.3745,
+      "step": 7410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.632340366443615e-07,
+      "loss": 3.4031,
+      "step": 7411
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6248774448952695e-07,
+      "loss": 3.3831,
+      "step": 7412
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.617421909437668e-07,
+      "loss": 3.5055,
+      "step": 7413
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6099737612582455e-07,
+      "loss": 3.4321,
+      "step": 7414
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6025330015432625e-07,
+      "loss": 3.4003,
+      "step": 7415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.595099631477783e-07,
+      "loss": 3.3895,
+      "step": 7416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.58767365224571e-07,
+      "loss": 3.3715,
+      "step": 7417
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5802550650297694e-07,
+      "loss": 3.2886,
+      "step": 7418
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5728438710115163e-07,
+      "loss": 3.37,
+      "step": 7419
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.565440071371307e-07,
+      "loss": 3.3265,
+      "step": 7420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.558043667288336e-07,
+      "loss": 3.4764,
+      "step": 7421
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5506546599406276e-07,
+      "loss": 3.3896,
+      "step": 7422
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.543273050505003e-07,
+      "loss": 3.426,
+      "step": 7423
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.53589884015712e-07,
+      "loss": 3.4241,
+      "step": 7424
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5285320300714574e-07,
+      "loss": 3.2848,
+      "step": 7425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.521172621421326e-07,
+      "loss": 3.3032,
+      "step": 7426
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5138206153788437e-07,
+      "loss": 3.3458,
+      "step": 7427
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.506476013114946e-07,
+      "loss": 3.3524,
+      "step": 7428
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4991388157993967e-07,
+      "loss": 3.5164,
+      "step": 7429
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4918090246007785e-07,
+      "loss": 3.3232,
+      "step": 7430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4844866406865017e-07,
+      "loss": 3.5019,
+      "step": 7431
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.477171665222784e-07,
+      "loss": 3.3965,
+      "step": 7432
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.469864099374681e-07,
+      "loss": 3.4278,
+      "step": 7433
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.462563944306047e-07,
+      "loss": 3.306,
+      "step": 7434
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.455271201179561e-07,
+      "loss": 3.3584,
+      "step": 7435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.447985871156734e-07,
+      "loss": 3.3192,
+      "step": 7436
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4407079553978875e-07,
+      "loss": 3.3201,
+      "step": 7437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.433437455062166e-07,
+      "loss": 3.3069,
+      "step": 7438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4261743713075335e-07,
+      "loss": 3.3191,
+      "step": 7439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.418918705290758e-07,
+      "loss": 3.3717,
+      "step": 7440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4116704581674374e-07,
+      "loss": 3.4141,
+      "step": 7441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.404429631091999e-07,
+      "loss": 3.4906,
+      "step": 7442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.397196225217675e-07,
+      "loss": 3.3061,
+      "step": 7443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3899702416965166e-07,
+      "loss": 3.4383,
+      "step": 7444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3827516816793914e-07,
+      "loss": 3.2854,
+      "step": 7445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3755405463159974e-07,
+      "loss": 3.3197,
+      "step": 7446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.368336836754826e-07,
+      "loss": 3.4538,
+      "step": 7447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.361140554143211e-07,
+      "loss": 3.403,
+      "step": 7448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.353951699627289e-07,
+      "loss": 3.4152,
+      "step": 7449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3467702743520123e-07,
+      "loss": 3.3469,
+      "step": 7450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3395962794611715e-07,
+      "loss": 3.3178,
+      "step": 7451
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.332429716097341e-07,
+      "loss": 3.4074,
+      "step": 7452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.32527058540193e-07,
+      "loss": 3.4316,
+      "step": 7453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3181188885151706e-07,
+      "loss": 3.5114,
+      "step": 7454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.310974626576091e-07,
+      "loss": 3.5755,
+      "step": 7455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.303837800722554e-07,
+      "loss": 3.4147,
+      "step": 7456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2967084120912316e-07,
+      "loss": 3.3958,
+      "step": 7457
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.289586461817601e-07,
+      "loss": 3.3638,
+      "step": 7458
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2824719510359703e-07,
+      "loss": 3.3943,
+      "step": 7459
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2753648808794505e-07,
+      "loss": 3.3767,
+      "step": 7460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.26826525247998e-07,
+      "loss": 3.3804,
+      "step": 7461
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.261173066968304e-07,
+      "loss": 3.3572,
+      "step": 7462
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2540883254739864e-07,
+      "loss": 3.4592,
+      "step": 7463
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.247011029125391e-07,
+      "loss": 3.3896,
+      "step": 7464
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2399411790497157e-07,
+      "loss": 3.4289,
+      "step": 7465
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2328787763729606e-07,
+      "loss": 3.3858,
+      "step": 7466
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2258238222199424e-07,
+      "loss": 3.4117,
+      "step": 7467
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2187763177143007e-07,
+      "loss": 3.3809,
+      "step": 7468
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2117362639784765e-07,
+      "loss": 3.4321,
+      "step": 7469
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.204703662133724e-07,
+      "loss": 3.3788,
+      "step": 7470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.197678513300112e-07,
+      "loss": 3.4104,
+      "step": 7471
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.190660818596525e-07,
+      "loss": 3.4067,
+      "step": 7472
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.183650579140668e-07,
+      "loss": 3.3415,
+      "step": 7473
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1766477960490485e-07,
+      "loss": 3.2817,
+      "step": 7474
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1696524704369846e-07,
+      "loss": 3.4477,
+      "step": 7475
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1626646034186084e-07,
+      "loss": 3.4302,
+      "step": 7476
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.155684196106873e-07,
+      "loss": 3.417,
+      "step": 7477
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1487112496135287e-07,
+      "loss": 3.3252,
+      "step": 7478
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.141745765049153e-07,
+      "loss": 3.4119,
+      "step": 7479
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.134787743523121e-07,
+      "loss": 3.4218,
+      "step": 7480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1278371861436397e-07,
+      "loss": 3.4018,
+      "step": 7481
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1208940940176955e-07,
+      "loss": 3.3897,
+      "step": 7482
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1139584682511106e-07,
+      "loss": 3.4143,
+      "step": 7483
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1070303099485055e-07,
+      "loss": 3.4859,
+      "step": 7484
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1001096202133263e-07,
+      "loss": 3.4276,
+      "step": 7485
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0931964001478134e-07,
+      "loss": 3.4455,
+      "step": 7486
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0862906508530355e-07,
+      "loss": 3.3819,
+      "step": 7487
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.079392373428841e-07,
+      "loss": 3.3858,
+      "step": 7488
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.072501568973918e-07,
+      "loss": 3.3358,
+      "step": 7489
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.065618238585749e-07,
+      "loss": 3.4126,
+      "step": 7490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.058742383360641e-07,
+      "loss": 3.3943,
+      "step": 7491
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.05187400439369e-07,
+      "loss": 3.426,
+      "step": 7492
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0450131027788265e-07,
+      "loss": 3.351,
+      "step": 7493
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.038159679608754e-07,
+      "loss": 3.3486,
+      "step": 7494
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.031313735975022e-07,
+      "loss": 3.4148,
+      "step": 7495
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0244752729679636e-07,
+      "loss": 3.401,
+      "step": 7496
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0176442916767303e-07,
+      "loss": 3.4144,
+      "step": 7497
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.01082079318929e-07,
+      "loss": 3.4005,
+      "step": 7498
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0040047785924065e-07,
+      "loss": 3.3869,
+      "step": 7499
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.997196248971651e-07,
+      "loss": 3.3812,
+      "step": 7500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.990395205411406e-07,
+      "loss": 3.5644,
+      "step": 7501
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9836016489948715e-07,
+      "loss": 3.4756,
+      "step": 7502
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9768155808040433e-07,
+      "loss": 3.4908,
+      "step": 7503
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9700370019197287e-07,
+      "loss": 3.2875,
+      "step": 7504
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.963265913421531e-07,
+      "loss": 3.4641,
+      "step": 7505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.956502316387877e-07,
+      "loss": 3.3847,
+      "step": 7506
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9497462118959984e-07,
+      "loss": 3.3237,
+      "step": 7507
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.942997601021924e-07,
+      "loss": 3.4844,
+      "step": 7508
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9362564848404884e-07,
+      "loss": 3.5295,
+      "step": 7509
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.929522864425355e-07,
+      "loss": 3.4518,
+      "step": 7510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9227967408489653e-07,
+      "loss": 3.3704,
+      "step": 7511
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.916078115182575e-07,
+      "loss": 3.4458,
+      "step": 7512
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9093669884962537e-07,
+      "loss": 3.3103,
+      "step": 7513
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9026633618588706e-07,
+      "loss": 3.3652,
+      "step": 7514
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.895967236338104e-07,
+      "loss": 3.3463,
+      "step": 7515
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.889278613000429e-07,
+      "loss": 3.2919,
+      "step": 7516
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.882597492911149e-07,
+      "loss": 3.4319,
+      "step": 7517
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.875923877134329e-07,
+      "loss": 3.2988,
+      "step": 7518
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8692577667328803e-07,
+      "loss": 3.4988,
+      "step": 7519
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.862599162768498e-07,
+      "loss": 3.3754,
+      "step": 7520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8559480663016945e-07,
+      "loss": 3.2751,
+      "step": 7521
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8493044783917733e-07,
+      "loss": 3.4258,
+      "step": 7522
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8426684000968594e-07,
+      "loss": 3.4143,
+      "step": 7523
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8360398324738415e-07,
+      "loss": 3.4357,
+      "step": 7524
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8294187765784687e-07,
+      "loss": 3.4054,
+      "step": 7525
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.822805233465259e-07,
+      "loss": 3.3421,
+      "step": 7526
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8161992041875364e-07,
+      "loss": 3.4188,
+      "step": 7527
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8096006897974425e-07,
+      "loss": 3.3418,
+      "step": 7528
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8030096913458984e-07,
+      "loss": 3.3877,
+      "step": 7529
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.796426209882647e-07,
+      "loss": 3.3589,
+      "step": 7530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7898502464562274e-07,
+      "loss": 3.3732,
+      "step": 7531
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.783281802113985e-07,
+      "loss": 3.461,
+      "step": 7532
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.776720877902067e-07,
+      "loss": 3.5424,
+      "step": 7533
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7701674748654206e-07,
+      "loss": 3.3891,
+      "step": 7534
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7636215940477996e-07,
+      "loss": 3.4106,
+      "step": 7535
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7570832364917423e-07,
+      "loss": 3.5159,
+      "step": 7536
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.75055240323861e-07,
+      "loss": 3.3089,
+      "step": 7537
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7440290953285597e-07,
+      "loss": 3.4474,
+      "step": 7538
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7375133138005493e-07,
+      "loss": 3.4055,
+      "step": 7539
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7310050596923323e-07,
+      "loss": 3.4456,
+      "step": 7540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7245043340404733e-07,
+      "loss": 3.4366,
+      "step": 7541
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7180111378803276e-07,
+      "loss": 3.3861,
+      "step": 7542
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7115254722460515e-07,
+      "loss": 3.3285,
+      "step": 7543
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7050473381706186e-07,
+      "loss": 3.4876,
+      "step": 7544
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.698576736685782e-07,
+      "loss": 3.4052,
+      "step": 7545
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.692113668822105e-07,
+      "loss": 3.4911,
+      "step": 7546
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6856581356089593e-07,
+      "loss": 3.4763,
+      "step": 7547
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.679210138074484e-07,
+      "loss": 3.4391,
+      "step": 7548
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.672769677245668e-07,
+      "loss": 3.4696,
+      "step": 7549
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6663367541482575e-07,
+      "loss": 3.3317,
+      "step": 7550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.659911369806817e-07,
+      "loss": 3.4041,
+      "step": 7551
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.653493525244721e-07,
+      "loss": 3.434,
+      "step": 7552
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.647083221484104e-07,
+      "loss": 3.4734,
+      "step": 7553
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.640680459545941e-07,
+      "loss": 3.3574,
+      "step": 7554
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6342852404499795e-07,
+      "loss": 3.2931,
+      "step": 7555
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6278975652147875e-07,
+      "loss": 3.3452,
+      "step": 7556
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.621517434857712e-07,
+      "loss": 3.4739,
+      "step": 7557
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.615144850394913e-07,
+      "loss": 3.3522,
+      "step": 7558
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.608779812841333e-07,
+      "loss": 3.2895,
+      "step": 7559
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.602422323210724e-07,
+      "loss": 3.3662,
+      "step": 7560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.59607238251563e-07,
+      "loss": 3.5142,
+      "step": 7561
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5897299917674033e-07,
+      "loss": 3.4572,
+      "step": 7562
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5833951519761814e-07,
+      "loss": 3.4248,
+      "step": 7563
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.577067864150906e-07,
+      "loss": 3.457,
+      "step": 7564
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.570748129299322e-07,
+      "loss": 3.4397,
+      "step": 7565
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5644359484279455e-07,
+      "loss": 3.4536,
+      "step": 7566
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.558131322542118e-07,
+      "loss": 3.4133,
+      "step": 7567
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.551834252645963e-07,
+      "loss": 3.4417,
+      "step": 7568
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5455447397424114e-07,
+      "loss": 3.4039,
+      "step": 7569
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.539262784833174e-07,
+      "loss": 3.4109,
+      "step": 7570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5329883889187836e-07,
+      "loss": 3.4106,
+      "step": 7571
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5267215529985346e-07,
+      "loss": 3.3959,
+      "step": 7572
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.520462278070535e-07,
+      "loss": 3.4229,
+      "step": 7573
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.514210565131708e-07,
+      "loss": 3.3482,
+      "step": 7574
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.507966415177743e-07,
+      "loss": 3.4111,
+      "step": 7575
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.501729829203142e-07,
+      "loss": 3.4973,
+      "step": 7576
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.49550080820119e-07,
+      "loss": 3.2924,
+      "step": 7577
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4892793531639647e-07,
+      "loss": 3.3551,
+      "step": 7578
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.483065465082363e-07,
+      "loss": 3.4536,
+      "step": 7579
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.476859144946053e-07,
+      "loss": 3.5728,
+      "step": 7580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4706603937435016e-07,
+      "loss": 3.358,
+      "step": 7581
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4644692124619895e-07,
+      "loss": 3.421,
+      "step": 7582
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4582856020875565e-07,
+      "loss": 3.4916,
+      "step": 7583
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.452109563605065e-07,
+      "loss": 3.4273,
+      "step": 7584
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4459410979981666e-07,
+      "loss": 3.4468,
+      "step": 7585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4397802062492927e-07,
+      "loss": 3.4192,
+      "step": 7586
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4336268893396853e-07,
+      "loss": 3.4128,
+      "step": 7587
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.427481148249383e-07,
+      "loss": 3.3779,
+      "step": 7588
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.421342983957192e-07,
+      "loss": 3.4343,
+      "step": 7589
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.41521239744073e-07,
+      "loss": 3.397,
+      "step": 7590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4090893896764056e-07,
+      "loss": 3.4182,
+      "step": 7591
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.402973961639432e-07,
+      "loss": 3.3203,
+      "step": 7592
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3968661143037864e-07,
+      "loss": 3.3972,
+      "step": 7593
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3907658486422734e-07,
+      "loss": 3.3475,
+      "step": 7594
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3846731656264666e-07,
+      "loss": 3.4734,
+      "step": 7595
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3785880662267223e-07,
+      "loss": 3.4178,
+      "step": 7596
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.372510551412227e-07,
+      "loss": 3.4675,
+      "step": 7597
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3664406221509118e-07,
+      "loss": 3.3318,
+      "step": 7598
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3603782794095532e-07,
+      "loss": 3.5486,
+      "step": 7599
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3543235241536788e-07,
+      "loss": 3.4699,
+      "step": 7600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3482763573476118e-07,
+      "loss": 3.471,
+      "step": 7601
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.342236779954482e-07,
+      "loss": 3.3186,
+      "step": 7602
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3362047929361975e-07,
+      "loss": 3.5454,
+      "step": 7603
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.330180397253473e-07,
+      "loss": 3.4216,
+      "step": 7604
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3241635938657968e-07,
+      "loss": 3.4976,
+      "step": 7605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3181543837314635e-07,
+      "loss": 3.4874,
+      "step": 7606
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.312152767807535e-07,
+      "loss": 3.4086,
+      "step": 7607
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3061587470498915e-07,
+      "loss": 3.3946,
+      "step": 7608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3001723224131856e-07,
+      "loss": 3.4925,
+      "step": 7609
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.294193494850866e-07,
+      "loss": 3.3373,
+      "step": 7610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2882222653151766e-07,
+      "loss": 3.4285,
+      "step": 7611
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.282258634757145e-07,
+      "loss": 3.4852,
+      "step": 7612
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2763026041265735e-07,
+      "loss": 3.4352,
+      "step": 7613
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2703541743720913e-07,
+      "loss": 3.4748,
+      "step": 7614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2644133464410744e-07,
+      "loss": 3.4845,
+      "step": 7615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.258480121279727e-07,
+      "loss": 3.4658,
+      "step": 7616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2525544998330152e-07,
+      "loss": 3.3741,
+      "step": 7617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2466364830447007e-07,
+      "loss": 3.3632,
+      "step": 7618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2407260718573522e-07,
+      "loss": 3.3712,
+      "step": 7619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2348232672122937e-07,
+      "loss": 3.4619,
+      "step": 7620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2289280700496573e-07,
+      "loss": 3.2666,
+      "step": 7621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2230404813083582e-07,
+      "loss": 3.355,
+      "step": 7622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2171605019261245e-07,
+      "loss": 3.5397,
+      "step": 7623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2112881328394287e-07,
+      "loss": 3.3902,
+      "step": 7624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2054233749835673e-07,
+      "loss": 3.4483,
+      "step": 7625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.199566229292599e-07,
+      "loss": 3.3912,
+      "step": 7626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1937166966993938e-07,
+      "loss": 3.4479,
+      "step": 7627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1878747781355847e-07,
+      "loss": 3.3329,
+      "step": 7628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.182040474531616e-07,
+      "loss": 3.5309,
+      "step": 7629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.176213786816711e-07,
+      "loss": 3.4717,
+      "step": 7630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1703947159188553e-07,
+      "loss": 3.4024,
+      "step": 7631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1645832627648633e-07,
+      "loss": 3.5139,
+      "step": 7632
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1587794282803053e-07,
+      "loss": 3.3885,
+      "step": 7633
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.152983213389559e-07,
+      "loss": 3.4759,
+      "step": 7634
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1471946190157632e-07,
+      "loss": 3.5176,
+      "step": 7635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.141413646080881e-07,
+      "loss": 3.4063,
+      "step": 7636
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1356402955056087e-07,
+      "loss": 3.4584,
+      "step": 7637
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1298745682094779e-07,
+      "loss": 3.3831,
+      "step": 7638
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.124116465110787e-07,
+      "loss": 3.3416,
+      "step": 7639
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1183659871266082e-07,
+      "loss": 3.4223,
+      "step": 7640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1126231351728254e-07,
+      "loss": 3.4058,
+      "step": 7641
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1068879101640848e-07,
+      "loss": 3.4283,
+      "step": 7642
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1011603130138224e-07,
+      "loss": 3.4422,
+      "step": 7643
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0954403446342753e-07,
+      "loss": 3.3129,
+      "step": 7644
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0897280059364367e-07,
+      "loss": 3.3439,
+      "step": 7645
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0840232978301178e-07,
+      "loss": 3.5011,
+      "step": 7646
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.078326221223892e-07,
+      "loss": 3.4224,
+      "step": 7647
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.072636777025122e-07,
+      "loss": 3.426,
+      "step": 7648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0669549661399556e-07,
+      "loss": 3.3838,
+      "step": 7649
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0612807894733356e-07,
+      "loss": 3.4094,
+      "step": 7650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0556142479289608e-07,
+      "loss": 3.3105,
+      "step": 7651
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.049955342409349e-07,
+      "loss": 3.3761,
+      "step": 7652
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0443040738157781e-07,
+      "loss": 3.3862,
+      "step": 7653
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0386604430483237e-07,
+      "loss": 3.3822,
+      "step": 7654
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0330244510058217e-07,
+      "loss": 3.2889,
+      "step": 7655
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.027396098585921e-07,
+      "loss": 3.4714,
+      "step": 7656
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.021775386685032e-07,
+      "loss": 3.3799,
+      "step": 7657
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0161623161983612e-07,
+      "loss": 3.4473,
+      "step": 7658
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0105568880198877e-07,
+      "loss": 3.3909,
+      "step": 7659
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0049591030423977e-07,
+      "loss": 3.4241,
+      "step": 7660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9993689621574164e-07,
+      "loss": 3.3957,
+      "step": 7661
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9937864662552874e-07,
+      "loss": 3.2595,
+      "step": 7662
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9882116162251274e-07,
+      "loss": 3.3652,
+      "step": 7663
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9826444129548317e-07,
+      "loss": 3.4101,
+      "step": 7664
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9770848573310797e-07,
+      "loss": 3.4894,
+      "step": 7665
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.971532950239341e-07,
+      "loss": 3.432,
+      "step": 7666
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9659886925638417e-07,
+      "loss": 3.3578,
+      "step": 7667
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9604520851876196e-07,
+      "loss": 3.4425,
+      "step": 7668
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9549231289924753e-07,
+      "loss": 3.4504,
+      "step": 7669
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9494018248589986e-07,
+      "loss": 3.4491,
+      "step": 7670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9438881736665582e-07,
+      "loss": 3.2313,
+      "step": 7671
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.938382176293313e-07,
+      "loss": 3.4269,
+      "step": 7672
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.932883833616178e-07,
+      "loss": 3.4571,
+      "step": 7673
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.927393146510881e-07,
+      "loss": 3.3286,
+      "step": 7674
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9219101158518993e-07,
+      "loss": 3.3471,
+      "step": 7675
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.916434742512524e-07,
+      "loss": 3.3378,
+      "step": 7676
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9109670273648018e-07,
+      "loss": 3.3339,
+      "step": 7677
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9055069712795582e-07,
+      "loss": 3.4632,
+      "step": 7678
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9000545751264144e-07,
+      "loss": 3.4676,
+      "step": 7679
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8946098397737644e-07,
+      "loss": 3.3203,
+      "step": 7680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8891727660887816e-07,
+      "loss": 3.4659,
+      "step": 7681
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.883743354937423e-07,
+      "loss": 3.4205,
+      "step": 7682
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878321607184419e-07,
+      "loss": 3.4848,
+      "step": 7683
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8729075236932903e-07,
+      "loss": 3.4679,
+      "step": 7684
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8675011053263192e-07,
+      "loss": 3.3877,
+      "step": 7685
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.862102352944578e-07,
+      "loss": 3.3431,
+      "step": 7686
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.856711267407918e-07,
+      "loss": 3.4421,
+      "step": 7687
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8513278495749743e-07,
+      "loss": 3.4444,
+      "step": 7688
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.84595210030315e-07,
+      "loss": 3.5642,
+      "step": 7689
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.840584020448638e-07,
+      "loss": 3.421,
+      "step": 7690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8352236108663935e-07,
+      "loss": 3.4979,
+      "step": 7691
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.829870872410161e-07,
+      "loss": 3.4255,
+      "step": 7692
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8245258059324754e-07,
+      "loss": 3.4634,
+      "step": 7693
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8191884122846226e-07,
+      "loss": 3.3561,
+      "step": 7694
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.813858692316689e-07,
+      "loss": 3.3999,
+      "step": 7695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8085366468775289e-07,
+      "loss": 3.5136,
+      "step": 7696
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8032222768147755e-07,
+      "loss": 3.1945,
+      "step": 7697
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7979155829748407e-07,
+      "loss": 3.5053,
+      "step": 7698
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7926165662029094e-07,
+      "loss": 3.4351,
+      "step": 7699
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.787325227342951e-07,
+      "loss": 3.5026,
+      "step": 7700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.782041567237708e-07,
+      "loss": 3.2816,
+      "step": 7701
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7767655867287016e-07,
+      "loss": 3.4425,
+      "step": 7702
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.771497286656221e-07,
+      "loss": 3.5331,
+      "step": 7703
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7662366678593502e-07,
+      "loss": 3.3583,
+      "step": 7704
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7609837311759358e-07,
+      "loss": 3.3641,
+      "step": 7705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7557384774425977e-07,
+      "loss": 3.3736,
+      "step": 7706
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.750500907494751e-07,
+      "loss": 3.512,
+      "step": 7707
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7452710221665725e-07,
+      "loss": 3.4291,
+      "step": 7708
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7400488222910073e-07,
+      "loss": 3.4303,
+      "step": 7709
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7348343086997954e-07,
+      "loss": 3.4543,
+      "step": 7710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7296274822234392e-07,
+      "loss": 3.3546,
+      "step": 7711
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7244283436912257e-07,
+      "loss": 3.3707,
+      "step": 7712
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7192368939312142e-07,
+      "loss": 3.3981,
+      "step": 7713
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7140531337702383e-07,
+      "loss": 3.6013,
+      "step": 7714
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.708877064033898e-07,
+      "loss": 3.3606,
+      "step": 7715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7037086855465902e-07,
+      "loss": 3.4188,
+      "step": 7716
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.698547999131467e-07,
+      "loss": 3.329,
+      "step": 7717
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6933950056104598e-07,
+      "loss": 3.3875,
+      "step": 7718
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6882497058042846e-07,
+      "loss": 3.4664,
+      "step": 7719
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6831121005324303e-07,
+      "loss": 3.3842,
+      "step": 7720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.677982190613131e-07,
+      "loss": 3.3418,
+      "step": 7721
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6728599768634447e-07,
+      "loss": 3.3825,
+      "step": 7722
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6677454600991683e-07,
+      "loss": 3.4115,
+      "step": 7723
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6626386411348783e-07,
+      "loss": 3.3659,
+      "step": 7724
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6575395207839462e-07,
+      "loss": 3.4963,
+      "step": 7725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.652448099858478e-07,
+      "loss": 3.4711,
+      "step": 7726
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.647364379169386e-07,
+      "loss": 3.4257,
+      "step": 7727
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6422883595263505e-07,
+      "loss": 3.4197,
+      "step": 7728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6372200417378136e-07,
+      "loss": 3.46,
+      "step": 7729
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6321594266110074e-07,
+      "loss": 3.4541,
+      "step": 7730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6271065149519204e-07,
+      "loss": 3.3144,
+      "step": 7731
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6220613075653201e-07,
+      "loss": 3.4723,
+      "step": 7732
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6170238052547526e-07,
+      "loss": 3.3852,
+      "step": 7733
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6119940088225317e-07,
+      "loss": 3.2998,
+      "step": 7734
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.606971919069744e-07,
+      "loss": 3.4473,
+      "step": 7735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6019575367962559e-07,
+      "loss": 3.3952,
+      "step": 7736
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.596950862800689e-07,
+      "loss": 3.3579,
+      "step": 7737
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5919518978804614e-07,
+      "loss": 3.5634,
+      "step": 7738
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5869606428317362e-07,
+      "loss": 3.442,
+      "step": 7739
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.581977098449472e-07,
+      "loss": 3.5395,
+      "step": 7740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5770012655273836e-07,
+      "loss": 3.4554,
+      "step": 7741
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5720331448579707e-07,
+      "loss": 3.3589,
+      "step": 7742
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.567072737232489e-07,
+      "loss": 3.3339,
+      "step": 7743
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.56212004344099e-07,
+      "loss": 3.391,
+      "step": 7744
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5571750642722595e-07,
+      "loss": 3.4093,
+      "step": 7745
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5522378005138894e-07,
+      "loss": 3.4082,
+      "step": 7746
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5473082529522342e-07,
+      "loss": 3.5055,
+      "step": 7747
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.542386422372405e-07,
+      "loss": 3.4218,
+      "step": 7748
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5374723095583022e-07,
+      "loss": 3.4076,
+      "step": 7749
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5325659152925833e-07,
+      "loss": 3.4716,
+      "step": 7750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5276672403566783e-07,
+      "loss": 3.4732,
+      "step": 7751
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5227762855308026e-07,
+      "loss": 3.3721,
+      "step": 7752
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5178930515939162e-07,
+      "loss": 3.4654,
+      "step": 7753
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5130175393237744e-07,
+      "loss": 3.431,
+      "step": 7754
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5081497494968956e-07,
+      "loss": 3.3293,
+      "step": 7755
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.503289682888548e-07,
+      "loss": 3.4706,
+      "step": 7756
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4984373402728014e-07,
+      "loss": 3.4433,
+      "step": 7757
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4935927224224768e-07,
+      "loss": 3.4034,
+      "step": 7758
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4887558301091676e-07,
+      "loss": 3.4156,
+      "step": 7759
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4839266641032357e-07,
+      "loss": 3.384,
+      "step": 7760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4791052251738214e-07,
+      "loss": 3.3514,
+      "step": 7761
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4742915140888213e-07,
+      "loss": 3.4523,
+      "step": 7762
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4694855316149058e-07,
+      "loss": 3.3401,
+      "step": 7763
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4646872785175182e-07,
+      "loss": 3.38,
+      "step": 7764
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4598967555608746e-07,
+      "loss": 3.4026,
+      "step": 7765
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.455113963507948e-07,
+      "loss": 3.4772,
+      "step": 7766
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.450338903120485e-07,
+      "loss": 3.4808,
+      "step": 7767
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4455715751590048e-07,
+      "loss": 3.421,
+      "step": 7768
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4408119803827947e-07,
+      "loss": 3.4781,
+      "step": 7769
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4360601195498924e-07,
+      "loss": 3.3108,
+      "step": 7770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4313159934171373e-07,
+      "loss": 3.4123,
+      "step": 7771
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.426579602740108e-07,
+      "loss": 3.4794,
+      "step": 7772
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4218509482731735e-07,
+      "loss": 3.414,
+      "step": 7773
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.417130030769448e-07,
+      "loss": 3.4607,
+      "step": 7774
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4124168509808246e-07,
+      "loss": 3.3532,
+      "step": 7775
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4077114096579635e-07,
+      "loss": 3.4258,
+      "step": 7776
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4030137075503048e-07,
+      "loss": 3.4201,
+      "step": 7777
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3983237454060272e-07,
+      "loss": 3.4035,
+      "step": 7778
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3936415239721112e-07,
+      "loss": 3.3231,
+      "step": 7779
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.388967043994266e-07,
+      "loss": 3.2033,
+      "step": 7780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3843003062170014e-07,
+      "loss": 3.3986,
+      "step": 7781
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.379641311383584e-07,
+      "loss": 3.4329,
+      "step": 7782
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.374990060236031e-07,
+      "loss": 3.42,
+      "step": 7783
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3703465535151505e-07,
+      "loss": 3.5217,
+      "step": 7784
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3657107919605118e-07,
+      "loss": 3.3225,
+      "step": 7785
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3610827763104295e-07,
+      "loss": 3.407,
+      "step": 7786
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3564625073020088e-07,
+      "loss": 3.3575,
+      "step": 7787
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3518499856711108e-07,
+      "loss": 3.4294,
+      "step": 7788
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.347245212152365e-07,
+      "loss": 3.3611,
+      "step": 7789
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.342648187479162e-07,
+      "loss": 3.3988,
+      "step": 7790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3380589123836717e-07,
+      "loss": 3.2921,
+      "step": 7791
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.333477387596821e-07,
+      "loss": 3.2919,
+      "step": 7792
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.328903613848287e-07,
+      "loss": 3.3864,
+      "step": 7793
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3243375918665314e-07,
+      "loss": 3.2467,
+      "step": 7794
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3197793223787947e-07,
+      "loss": 3.3812,
+      "step": 7795
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3152288061110518e-07,
+      "loss": 3.3528,
+      "step": 7796
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3106860437880509e-07,
+      "loss": 3.3955,
+      "step": 7797
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3061510361333186e-07,
+      "loss": 3.3753,
+      "step": 7798
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3016237838691382e-07,
+      "loss": 3.3308,
+      "step": 7799
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2971042877165496e-07,
+      "loss": 3.5375,
+      "step": 7800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.292592548395377e-07,
+      "loss": 3.3772,
+      "step": 7801
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2880885666241905e-07,
+      "loss": 3.4014,
+      "step": 7802
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2835923431203433e-07,
+      "loss": 3.4117,
+      "step": 7803
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2791038785999243e-07,
+      "loss": 3.4236,
+      "step": 7804
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2746231737778113e-07,
+      "loss": 3.406,
+      "step": 7805
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2701502293676394e-07,
+      "loss": 3.2412,
+      "step": 7806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2656850460818104e-07,
+      "loss": 3.3224,
+      "step": 7807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.261227624631478e-07,
+      "loss": 3.3958,
+      "step": 7808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2567779657265854e-07,
+      "loss": 3.4694,
+      "step": 7809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2523360700758047e-07,
+      "loss": 3.3089,
+      "step": 7810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2479019383865977e-07,
+      "loss": 3.3675,
+      "step": 7811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.243475571365177e-07,
+      "loss": 3.4169,
+      "step": 7812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.239056969716529e-07,
+      "loss": 3.5092,
+      "step": 7813
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2346461341443962e-07,
+      "loss": 3.4334,
+      "step": 7814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.230243065351283e-07,
+      "loss": 3.3103,
+      "step": 7815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2258477640384569e-07,
+      "loss": 3.5906,
+      "step": 7816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.221460230905952e-07,
+      "loss": 3.4087,
+      "step": 7817
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2170804666525648e-07,
+      "loss": 3.4449,
+      "step": 7818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2127084719758486e-07,
+      "loss": 3.3733,
+      "step": 7819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2083442475721353e-07,
+      "loss": 3.5281,
+      "step": 7820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.203987794136502e-07,
+      "loss": 3.2465,
+      "step": 7821
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1996391123627883e-07,
+      "loss": 3.3853,
+      "step": 7822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1952982029436067e-07,
+      "loss": 3.2858,
+      "step": 7823
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1909650665703265e-07,
+      "loss": 3.3779,
+      "step": 7824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.186639703933079e-07,
+      "loss": 3.4271,
+      "step": 7825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1823221157207576e-07,
+      "loss": 3.4725,
+      "step": 7826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1780123026210233e-07,
+      "loss": 3.3483,
+      "step": 7827
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1737102653202825e-07,
+      "loss": 3.5195,
+      "step": 7828
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1694160045037206e-07,
+      "loss": 3.419,
+      "step": 7829
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1651295208552738e-07,
+      "loss": 3.415,
+      "step": 7830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1608508150576459e-07,
+      "loss": 3.3868,
+      "step": 7831
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1565798877922973e-07,
+      "loss": 3.5135,
+      "step": 7832
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1523167397394618e-07,
+      "loss": 3.4988,
+      "step": 7833
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1480613715781074e-07,
+      "loss": 3.4395,
+      "step": 7834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1438137839859919e-07,
+      "loss": 3.4104,
+      "step": 7835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.139573977639613e-07,
+      "loss": 3.3665,
+      "step": 7836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1353419532142474e-07,
+      "loss": 3.3474,
+      "step": 7837
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1311177113839167e-07,
+      "loss": 3.4401,
+      "step": 7838
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1269012528214108e-07,
+      "loss": 3.3334,
+      "step": 7839
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1226925781982756e-07,
+      "loss": 3.4452,
+      "step": 7840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1184916881848251e-07,
+      "loss": 3.3449,
+      "step": 7841
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1142985834501186e-07,
+      "loss": 3.4094,
+      "step": 7842
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1101132646619994e-07,
+      "loss": 3.3882,
+      "step": 7843
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1059357324870456e-07,
+      "loss": 3.3736,
+      "step": 7844
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1017659875906084e-07,
+      "loss": 3.5452,
+      "step": 7845
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0976040306367952e-07,
+      "loss": 3.4442,
+      "step": 7846
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.093449862288476e-07,
+      "loss": 3.3038,
+      "step": 7847
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0893034832072825e-07,
+      "loss": 3.5568,
+      "step": 7848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0851648940535919e-07,
+      "loss": 3.4822,
+      "step": 7849
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0810340954865662e-07,
+      "loss": 3.4943,
+      "step": 7850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0769110881640955e-07,
+      "loss": 3.3798,
+      "step": 7851
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0727958727428489e-07,
+      "loss": 3.4254,
+      "step": 7852
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0686884498782523e-07,
+      "loss": 3.4727,
+      "step": 7853
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.064588820224488e-07,
+      "loss": 3.3431,
+      "step": 7854
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0604969844345003e-07,
+      "loss": 3.3954,
+      "step": 7855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.056412943159979e-07,
+      "loss": 3.4702,
+      "step": 7856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0523366970513982e-07,
+      "loss": 3.4667,
+      "step": 7857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0482682467579608e-07,
+      "loss": 3.3445,
+      "step": 7858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0442075929276485e-07,
+      "loss": 3.4491,
+      "step": 7859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0401547362071939e-07,
+      "loss": 3.3619,
+      "step": 7860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.036109677242092e-07,
+      "loss": 3.4112,
+      "step": 7861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0320724166765882e-07,
+      "loss": 3.3998,
+      "step": 7862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0280429551536964e-07,
+      "loss": 3.4102,
+      "step": 7863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.024021293315175e-07,
+      "loss": 3.4092,
+      "step": 7864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0200074318015452e-07,
+      "loss": 3.3922,
+      "step": 7865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0160013712521011e-07,
+      "loss": 3.4572,
+      "step": 7866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0120031123048712e-07,
+      "loss": 3.4145,
+      "step": 7867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0080126555966574e-07,
+      "loss": 3.4071,
+      "step": 7868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0040300017630123e-07,
+      "loss": 3.3108,
+      "step": 7869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.000055151438245e-07,
+      "loss": 3.3565,
+      "step": 7870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.960881052554216e-08,
+      "loss": 3.2815,
+      "step": 7871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.9212886384637e-08,
+      "loss": 3.3973,
+      "step": 7872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.881774278416744e-08,
+      "loss": 3.3192,
+      "step": 7873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.842337978706707e-08,
+      "loss": 3.4378,
+      "step": 7874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.802979745614505e-08,
+      "loss": 3.3234,
+      "step": 7875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.763699585408737e-08,
+      "loss": 3.3737,
+      "step": 7876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.724497504345398e-08,
+      "loss": 3.2994,
+      "step": 7877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.685373508668272e-08,
+      "loss": 3.2419,
+      "step": 7878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.64632760460843e-08,
+      "loss": 3.4389,
+      "step": 7879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.607359798384785e-08,
+      "loss": 3.4027,
+      "step": 7880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.568470096203597e-08,
+      "loss": 3.3361,
+      "step": 7881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.529658504258743e-08,
+      "loss": 3.3988,
+      "step": 7882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.490925028731723e-08,
+      "loss": 3.3571,
+      "step": 7883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.452269675791603e-08,
+      "loss": 3.4547,
+      "step": 7884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.413692451594847e-08,
+      "loss": 3.2855,
+      "step": 7885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.37519336228565e-08,
+      "loss": 3.4714,
+      "step": 7886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.336772413995777e-08,
+      "loss": 3.3803,
+      "step": 7887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.298429612844329e-08,
+      "loss": 3.4271,
+      "step": 7888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.260164964938146e-08,
+      "loss": 3.4712,
+      "step": 7889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.221978476371574e-08,
+      "loss": 3.3733,
+      "step": 7890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.183870153226582e-08,
+      "loss": 3.3564,
+      "step": 7891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.145840001572537e-08,
+      "loss": 3.2959,
+      "step": 7892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.107888027466483e-08,
+      "loss": 3.3893,
+      "step": 7893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.070014236952973e-08,
+      "loss": 3.4788,
+      "step": 7894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.032218636064072e-08,
+      "loss": 3.4995,
+      "step": 7895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.994501230819464e-08,
+      "loss": 3.3655,
+      "step": 7896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.956862027226287e-08,
+      "loss": 3.3868,
+      "step": 7897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.919301031279415e-08,
+      "loss": 3.4519,
+      "step": 7898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.881818248960894e-08,
+      "loss": 3.6034,
+      "step": 7899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.844413686240727e-08,
+      "loss": 3.4271,
+      "step": 7900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.807087349076205e-08,
+      "loss": 3.4791,
+      "step": 7901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.769839243412237e-08,
+      "loss": 3.3031,
+      "step": 7902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.732669375181302e-08,
+      "loss": 3.3997,
+      "step": 7903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.69557775030344e-08,
+      "loss": 3.3027,
+      "step": 7904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.65856437468604e-08,
+      "loss": 3.388,
+      "step": 7905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.621629254224218e-08,
+      "loss": 3.3801,
+      "step": 7906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.584772394800545e-08,
+      "loss": 3.4706,
+      "step": 7907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.547993802285215e-08,
+      "loss": 3.4459,
+      "step": 7908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.511293482535821e-08,
+      "loss": 3.3347,
+      "step": 7909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.474671441397631e-08,
+      "loss": 3.2916,
+      "step": 7910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.438127684703423e-08,
+      "loss": 3.4349,
+      "step": 7911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.40166221827332e-08,
+      "loss": 3.4048,
+      "step": 7912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.365275047915233e-08,
+      "loss": 3.3092,
+      "step": 7913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.328966179424413e-08,
+      "loss": 3.3369,
+      "step": 7914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.292735618583735e-08,
+      "loss": 3.3169,
+      "step": 7915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.256583371163585e-08,
+      "loss": 3.4458,
+      "step": 7916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.22050944292191e-08,
+      "loss": 3.2538,
+      "step": 7917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.184513839604113e-08,
+      "loss": 3.4773,
+      "step": 7918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.148596566943168e-08,
+      "loss": 3.4841,
+      "step": 7919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.112757630659551e-08,
+      "loss": 3.4439,
+      "step": 7920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.076997036461254e-08,
+      "loss": 3.3898,
+      "step": 7921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.041314790043886e-08,
+      "loss": 3.4687,
+      "step": 7922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.005710897090402e-08,
+      "loss": 3.4763,
+      "step": 7923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.970185363271432e-08,
+      "loss": 3.3033,
+      "step": 7924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.934738194245062e-08,
+      "loss": 3.4886,
+      "step": 7925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.899369395656831e-08,
+      "loss": 3.4015,
+      "step": 7926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.86407897314001e-08,
+      "loss": 3.4135,
+      "step": 7927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.828866932315216e-08,
+      "loss": 3.3368,
+      "step": 7928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.793733278790461e-08,
+      "loss": 3.427,
+      "step": 7929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.758678018161603e-08,
+      "loss": 3.371,
+      "step": 7930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.723701156011787e-08,
+      "loss": 3.408,
+      "step": 7931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.688802697911668e-08,
+      "loss": 3.3978,
+      "step": 7932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.653982649419466e-08,
+      "loss": 3.3513,
+      "step": 7933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.619241016081081e-08,
+      "loss": 3.4604,
+      "step": 7934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.584577803429527e-08,
+      "loss": 3.2984,
+      "step": 7935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.549993016985669e-08,
+      "loss": 3.4684,
+      "step": 7936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.515486662257765e-08,
+      "loss": 3.4452,
+      "step": 7937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.481058744741587e-08,
+      "loss": 3.3096,
+      "step": 7938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.446709269920416e-08,
+      "loss": 3.1855,
+      "step": 7939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.41243824326504e-08,
+      "loss": 3.4169,
+      "step": 7940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.378245670233764e-08,
+      "loss": 3.4271,
+      "step": 7941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.344131556272282e-08,
+      "loss": 3.3226,
+      "step": 7942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.310095906813975e-08,
+      "loss": 3.3679,
+      "step": 7943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.276138727279669e-08,
+      "loss": 3.3421,
+      "step": 7944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.242260023077651e-08,
+      "loss": 3.41,
+      "step": 7945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.208459799603774e-08,
+      "loss": 3.4236,
+      "step": 7946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.174738062241227e-08,
+      "loss": 3.3716,
+      "step": 7947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.141094816360883e-08,
+      "loss": 3.4376,
+      "step": 7948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.107530067321011e-08,
+      "loss": 3.3283,
+      "step": 7949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.074043820467503e-08,
+      "loss": 3.4679,
+      "step": 7950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.040636081133589e-08,
+      "loss": 3.4764,
+      "step": 7951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.007306854640072e-08,
+      "loss": 3.4006,
+      "step": 7952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.97405614629526e-08,
+      "loss": 3.5007,
+      "step": 7953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.940883961394917e-08,
+      "loss": 3.4058,
+      "step": 7954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.907790305222318e-08,
+      "loss": 3.4628,
+      "step": 7955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.8747751830483e-08,
+      "loss": 3.4142,
+      "step": 7956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.841838600131046e-08,
+      "loss": 3.4166,
+      "step": 7957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.808980561716416e-08,
+      "loss": 3.3949,
+      "step": 7958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.776201073037502e-08,
+      "loss": 3.4999,
+      "step": 7959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.743500139315185e-08,
+      "loss": 3.316,
+      "step": 7960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.710877765757629e-08,
+      "loss": 3.3468,
+      "step": 7961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.678333957560513e-08,
+      "loss": 3.5731,
+      "step": 7962
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.645868719907078e-08,
+      "loss": 3.2653,
+      "step": 7963
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.613482057968023e-08,
+      "loss": 3.3444,
+      "step": 7964
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.581173976901556e-08,
+      "loss": 3.3558,
+      "step": 7965
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.548944481853225e-08,
+      "loss": 3.4561,
+      "step": 7966
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.516793577956204e-08,
+      "loss": 3.4573,
+      "step": 7967
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.484721270331118e-08,
+      "loss": 3.2988,
+      "step": 7968
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.452727564086215e-08,
+      "loss": 3.4515,
+      "step": 7969
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.420812464316861e-08,
+      "loss": 3.4318,
+      "step": 7970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.388975976106216e-08,
+      "loss": 3.4265,
+      "step": 7971
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.357218104524832e-08,
+      "loss": 3.4198,
+      "step": 7972
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.32553885463072e-08,
+      "loss": 3.4423,
+      "step": 7973
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.293938231469398e-08,
+      "loss": 3.3143,
+      "step": 7974
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.262416240073843e-08,
+      "loss": 3.3755,
+      "step": 7975
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.230972885464536e-08,
+      "loss": 3.4564,
+      "step": 7976
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.199608172649363e-08,
+      "loss": 3.3946,
+      "step": 7977
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.168322106623715e-08,
+      "loss": 3.4007,
+      "step": 7978
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.13711469237055e-08,
+      "loss": 3.4221,
+      "step": 7979
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.10598593486017e-08,
+      "loss": 3.422,
+      "step": 7980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.074935839050389e-08,
+      "loss": 3.3476,
+      "step": 7981
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.043964409886583e-08,
+      "loss": 3.307,
+      "step": 7982
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.013071652301416e-08,
+      "loss": 3.4101,
+      "step": 7983
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.982257571215178e-08,
+      "loss": 3.4514,
+      "step": 7984
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9515221715356066e-08,
+      "loss": 3.454,
+      "step": 7985
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.920865458157787e-08,
+      "loss": 3.3374,
+      "step": 7986
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.890287435964481e-08,
+      "loss": 3.3672,
+      "step": 7987
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8597881098257924e-08,
+      "loss": 3.3953,
+      "step": 7988
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.829367484599169e-08,
+      "loss": 3.4232,
+      "step": 7989
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.79902556512979e-08,
+      "loss": 3.4597,
+      "step": 7990
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7687623562500686e-08,
+      "loss": 3.503,
+      "step": 7991
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.738577862780037e-08,
+      "loss": 3.356,
+      "step": 7992
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.708472089527128e-08,
+      "loss": 3.4159,
+      "step": 7993
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.6784450412862293e-08,
+      "loss": 3.5006,
+      "step": 7994
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.64849672283968e-08,
+      "loss": 3.5287,
+      "step": 7995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.618627138957333e-08,
+      "loss": 3.457,
+      "step": 7996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.588836294396438e-08,
+      "loss": 3.4127,
+      "step": 7997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.5591241939018104e-08,
+      "loss": 3.3667,
+      "step": 7998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.529490842205498e-08,
+      "loss": 3.4134,
+      "step": 7999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.499936244027337e-08,
+      "loss": 3.4386,
+      "step": 8000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.47046040407434e-08,
+      "loss": 3.3108,
+      "step": 8001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.4410633270410294e-08,
+      "loss": 3.2099,
+      "step": 8002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.411745017609493e-08,
+      "loss": 3.4179,
+      "step": 8003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.382505480449274e-08,
+      "loss": 3.4056,
+      "step": 8004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3533447202171466e-08,
+      "loss": 3.2968,
+      "step": 8005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.324262741557673e-08,
+      "loss": 3.4353,
+      "step": 8006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.295259549102594e-08,
+      "loss": 3.5065,
+      "step": 8007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.266335147471158e-08,
+      "loss": 3.3754,
+      "step": 8008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.237489541270235e-08,
+      "loss": 3.3715,
+      "step": 8009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.208722735093874e-08,
+      "loss": 3.3623,
+      "step": 8010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.180034733523853e-08,
+      "loss": 3.3652,
+      "step": 8011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.151425541129185e-08,
+      "loss": 3.4808,
+      "step": 8012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.122895162466446e-08,
+      "loss": 3.3269,
+      "step": 8013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.094443602079613e-08,
+      "loss": 3.3664,
+      "step": 8014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.066070864500061e-08,
+      "loss": 3.3847,
+      "step": 8015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0377769542467314e-08,
+      "loss": 3.4012,
+      "step": 8016
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.009561875825964e-08,
+      "loss": 3.4039,
+      "step": 8017
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.981425633731496e-08,
+      "loss": 3.3541,
+      "step": 8018
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.953368232444522e-08,
+      "loss": 3.4925,
+      "step": 8019
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.925389676433745e-08,
+      "loss": 3.3985,
+      "step": 8020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.897489970155267e-08,
+      "loss": 3.4107,
+      "step": 8021
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.869669118052589e-08,
+      "loss": 3.4022,
+      "step": 8022
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.841927124556722e-08,
+      "loss": 3.3113,
+      "step": 8023
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.814263994086077e-08,
+      "loss": 3.4383,
+      "step": 8024
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7866797310465176e-08,
+      "loss": 3.4913,
+      "step": 8025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7591743398313076e-08,
+      "loss": 3.4045,
+      "step": 8026
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7317478248212754e-08,
+      "loss": 3.4405,
+      "step": 8027
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.704400190384484e-08,
+      "loss": 3.3065,
+      "step": 8028
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.67713144087667e-08,
+      "loss": 3.4687,
+      "step": 8029
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6499415806408044e-08,
+      "loss": 3.4532,
+      "step": 8030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.622830614007423e-08,
+      "loss": 3.3833,
+      "step": 8031
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5957985452943496e-08,
+      "loss": 3.3192,
+      "step": 8032
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5688453788070296e-08,
+      "loss": 3.4034,
+      "step": 8033
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.541971118838196e-08,
+      "loss": 3.3202,
+      "step": 8034
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.51517576966809e-08,
+      "loss": 3.4995,
+      "step": 8035
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.48845933556441e-08,
+      "loss": 3.2906,
+      "step": 8036
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4618218207821975e-08,
+      "loss": 3.3597,
+      "step": 8037
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4352632295638906e-08,
+      "loss": 3.3193,
+      "step": 8038
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.408783566139552e-08,
+      "loss": 3.4026,
+      "step": 8039
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3823828347265285e-08,
+      "loss": 3.2749,
+      "step": 8040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.356061039529569e-08,
+      "loss": 3.3597,
+      "step": 8041
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.329818184740986e-08,
+      "loss": 3.3806,
+      "step": 8042
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.303654274540325e-08,
+      "loss": 3.4084,
+      "step": 8043
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2775693130948094e-08,
+      "loss": 3.4715,
+      "step": 8044
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2515633045588366e-08,
+      "loss": 3.3815,
+      "step": 8045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.225636253074372e-08,
+      "loss": 3.4452,
+      "step": 8046
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.199788162770779e-08,
+      "loss": 3.5075,
+      "step": 8047
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1740190377648206e-08,
+      "loss": 3.4397,
+      "step": 8048
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.148328882160768e-08,
+      "loss": 3.3444,
+      "step": 8049
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.122717700050183e-08,
+      "loss": 3.3187,
+      "step": 8050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.097185495512135e-08,
+      "loss": 3.4688,
+      "step": 8051
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.071732272613149e-08,
+      "loss": 3.348,
+      "step": 8052
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.046358035407039e-08,
+      "loss": 3.4959,
+      "step": 8053
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.021062787935181e-08,
+      "loss": 3.3937,
+      "step": 8054
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.995846534226244e-08,
+      "loss": 3.3584,
+      "step": 8055
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.970709278296458e-08,
+      "loss": 3.3613,
+      "step": 8056
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.945651024149288e-08,
+      "loss": 3.4083,
+      "step": 8057
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9206717757758197e-08,
+      "loss": 3.4766,
+      "step": 8058
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.895771537154425e-08,
+      "loss": 3.3657,
+      "step": 8059
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8709503122509873e-08,
+      "loss": 3.405,
+      "step": 8060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8462081050185676e-08,
+      "loss": 3.4367,
+      "step": 8061
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.821544919398012e-08,
+      "loss": 3.3144,
+      "step": 8062
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.796960759317236e-08,
+      "loss": 3.4204,
+      "step": 8063
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.772455628691829e-08,
+      "loss": 3.3781,
+      "step": 8064
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7480295314246127e-08,
+      "loss": 3.2701,
+      "step": 8065
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7236824714059205e-08,
+      "loss": 3.4626,
+      "step": 8066
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.699414452513483e-08,
+      "loss": 3.5175,
+      "step": 8067
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.675225478612432e-08,
+      "loss": 3.5534,
+      "step": 8068
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6511155535552954e-08,
+      "loss": 3.5147,
+      "step": 8069
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.627084681181947e-08,
+      "loss": 3.4063,
+      "step": 8070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6031328653198807e-08,
+      "loss": 3.316,
+      "step": 8071
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.579260109783766e-08,
+      "loss": 3.3135,
+      "step": 8072
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.555466418375841e-08,
+      "loss": 3.4006,
+      "step": 8073
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.531751794885574e-08,
+      "loss": 3.3727,
+      "step": 8074
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.508116243090109e-08,
+      "loss": 3.3748,
+      "step": 8075
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4845597667537144e-08,
+      "loss": 3.3721,
+      "step": 8076
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.461082369628332e-08,
+      "loss": 3.5018,
+      "step": 8077
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4376840554530275e-08,
+      "loss": 3.4439,
+      "step": 8078
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.41436482795443e-08,
+      "loss": 3.4244,
+      "step": 8079
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391124690846625e-08,
+      "loss": 3.4437,
+      "step": 8080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3679636478310404e-08,
+      "loss": 3.4106,
+      "step": 8081
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.344881702596392e-08,
+      "loss": 3.3418,
+      "step": 8082
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.321878858819017e-08,
+      "loss": 3.4144,
+      "step": 8083
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2989551201624836e-08,
+      "loss": 3.4459,
+      "step": 8084
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2761104902778175e-08,
+      "loss": 3.4068,
+      "step": 8085
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2533449728034405e-08,
+      "loss": 3.4235,
+      "step": 8086
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2306585713652306e-08,
+      "loss": 3.4956,
+      "step": 8087
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2080512895763525e-08,
+      "loss": 3.455,
+      "step": 8088
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.185523131037482e-08,
+      "loss": 3.4831,
+      "step": 8089
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.163074099336583e-08,
+      "loss": 3.4267,
+      "step": 8090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.140704198049127e-08,
+      "loss": 3.4372,
+      "step": 8091
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.118413430737932e-08,
+      "loss": 3.4601,
+      "step": 8092
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0962018009532134e-08,
+      "loss": 3.4468,
+      "step": 8093
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0740693122325304e-08,
+      "loss": 3.4616,
+      "step": 8094
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.052015968100952e-08,
+      "loss": 3.3392,
+      "step": 8095
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.030041772070835e-08,
+      "loss": 3.4707,
+      "step": 8096
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0081467276419915e-08,
+      "loss": 3.437,
+      "step": 8097
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.986330838301632e-08,
+      "loss": 3.3996,
+      "step": 8098
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9645941075242546e-08,
+      "loss": 3.3422,
+      "step": 8099
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9429365387719232e-08,
+      "loss": 3.3979,
+      "step": 8100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9213581354940456e-08,
+      "loss": 3.4497,
+      "step": 8101
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.899858901127206e-08,
+      "loss": 3.4206,
+      "step": 8102
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8784388390956652e-08,
+      "loss": 3.4608,
+      "step": 8103
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.857097952810972e-08,
+      "loss": 3.4381,
+      "step": 8104
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8358362456720745e-08,
+      "loss": 3.4738,
+      "step": 8105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.814653721065208e-08,
+      "loss": 3.4175,
+      "step": 8106
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7935503823641186e-08,
+      "loss": 3.4225,
+      "step": 8107
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7725262329298953e-08,
+      "loss": 3.4155,
+      "step": 8108
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7515812761110262e-08,
+      "loss": 3.4816,
+      "step": 8109
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7307155152433986e-08,
+      "loss": 3.4851,
+      "step": 8110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7099289536502426e-08,
+      "loss": 3.3863,
+      "step": 8111
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.689221594642244e-08,
+      "loss": 3.5065,
+      "step": 8112
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.66859344151732e-08,
+      "loss": 3.3909,
+      "step": 8113
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6480444975610086e-08,
+      "loss": 3.4932,
+      "step": 8114
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6275747660460817e-08,
+      "loss": 3.4139,
+      "step": 8115
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6071842502326526e-08,
+      "loss": 3.3624,
+      "step": 8116
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.586872953368347e-08,
+      "loss": 3.5497,
+      "step": 8117
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5666408786880758e-08,
+      "loss": 3.4364,
+      "step": 8118
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.546488029414207e-08,
+      "loss": 3.3373,
+      "step": 8119
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5264144087564503e-08,
+      "loss": 3.4188,
+      "step": 8120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.506420019911804e-08,
+      "loss": 3.4197,
+      "step": 8121
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4865048660648873e-08,
+      "loss": 3.2898,
+      "step": 8122
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4666689503874964e-08,
+      "loss": 3.4246,
+      "step": 8123
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4469122760388264e-08,
+      "loss": 3.3179,
+      "step": 8124
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.427234846165527e-08,
+      "loss": 3.4384,
+      "step": 8125
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4076366639015914e-08,
+      "loss": 3.5161,
+      "step": 8126
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3881177323683557e-08,
+      "loss": 3.3351,
+      "step": 8127
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3686780546746114e-08,
+      "loss": 3.3713,
+      "step": 8128
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.349317633916437e-08,
+      "loss": 3.4068,
+      "step": 8129
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3300364731774217e-08,
+      "loss": 3.4003,
+      "step": 8130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.310834575528387e-08,
+      "loss": 3.4543,
+      "step": 8131
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2917119440275524e-08,
+      "loss": 3.4497,
+      "step": 8132
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.272668581720594e-08,
+      "loss": 3.4097,
+      "step": 8133
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2537044916405292e-08,
+      "loss": 3.4535,
+      "step": 8134
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.23481967680772e-08,
+      "loss": 3.4174,
+      "step": 8135
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.216014140229872e-08,
+      "loss": 3.3883,
+      "step": 8136
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1972878849021994e-08,
+      "loss": 3.5505,
+      "step": 8137
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1786409138072063e-08,
+      "loss": 3.4319,
+      "step": 8138
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1600732299146275e-08,
+      "loss": 3.3937,
+      "step": 8139
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.141584836181876e-08,
+      "loss": 3.4345,
+      "step": 8140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1231757355534287e-08,
+      "loss": 3.3636,
+      "step": 8141
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.104845930961441e-08,
+      "loss": 3.322,
+      "step": 8142
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0865954253250776e-08,
+      "loss": 3.4081,
+      "step": 8143
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0684242215511797e-08,
+      "loss": 3.351,
+      "step": 8144
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0503323225338768e-08,
+      "loss": 3.3865,
+      "step": 8145
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0323197311545305e-08,
+      "loss": 3.354,
+      "step": 8146
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0143864502820686e-08,
+      "loss": 3.5087,
+      "step": 8147
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.996532482772595e-08,
+      "loss": 3.3387,
+      "step": 8148
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9787578314698352e-08,
+      "loss": 3.406,
+      "step": 8149
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.96106249920458e-08,
+      "loss": 3.3828,
+      "step": 8150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.943446488795242e-08,
+      "loss": 3.4264,
+      "step": 8151
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.92590980304741e-08,
+      "loss": 3.4771,
+      "step": 8152
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9084524447541275e-08,
+      "loss": 3.3842,
+      "step": 8153
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8910744166958926e-08,
+      "loss": 3.4555,
+      "step": 8154
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8737757216404362e-08,
+      "loss": 3.4453,
+      "step": 8155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8565563623428318e-08,
+      "loss": 3.4769,
+      "step": 8156
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.839416341545608e-08,
+      "loss": 3.3339,
+      "step": 8157
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.822355661978692e-08,
+      "loss": 3.4236,
+      "step": 8158
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.805374326359244e-08,
+      "loss": 3.4158,
+      "step": 8159
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7884723373918223e-08,
+      "loss": 3.3998,
+      "step": 8160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.771649697768496e-08,
+      "loss": 3.4498,
+      "step": 8161
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7549064101684553e-08,
+      "loss": 3.3527,
+      "step": 8162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.738242477258456e-08,
+      "loss": 3.4591,
+      "step": 8163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7216579016925415e-08,
+      "loss": 3.3919,
+      "step": 8164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.705152686112044e-08,
+      "loss": 3.3753,
+      "step": 8165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.688726833145804e-08,
+      "loss": 3.3691,
+      "step": 8166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6723803454098408e-08,
+      "loss": 3.364,
+      "step": 8167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6561132255077382e-08,
+      "loss": 3.3379,
+      "step": 8168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6399254760302575e-08,
+      "loss": 3.3841,
+      "step": 8169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6238170995556135e-08,
+      "loss": 3.3876,
+      "step": 8170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6077880986493655e-08,
+      "loss": 3.3919,
+      "step": 8171
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5918384758644157e-08,
+      "loss": 3.4443,
+      "step": 8172
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575968233741121e-08,
+      "loss": 3.4609,
+      "step": 8173
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5601773748070147e-08,
+      "loss": 3.5124,
+      "step": 8174
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5444659015770858e-08,
+      "loss": 3.5762,
+      "step": 8175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5288338165536654e-08,
+      "loss": 3.268,
+      "step": 8176
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.51328112222654e-08,
+      "loss": 3.4103,
+      "step": 8177
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.497807821072672e-08,
+      "loss": 3.4663,
+      "step": 8178
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4824139155564798e-08,
+      "loss": 3.4069,
+      "step": 8179
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4670994081297796e-08,
+      "loss": 3.3627,
+      "step": 8180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4518643012316203e-08,
+      "loss": 3.4693,
+      "step": 8181
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4367085972885608e-08,
+      "loss": 3.4819,
+      "step": 8182
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4216322987142816e-08,
+      "loss": 3.2474,
+      "step": 8183
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4066354079101396e-08,
+      "loss": 3.4753,
+      "step": 8184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3917179272645576e-08,
+      "loss": 3.2755,
+      "step": 8185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3768798591534127e-08,
+      "loss": 3.4449,
+      "step": 8186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3621212059399259e-08,
+      "loss": 3.3885,
+      "step": 8187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.347441969974772e-08,
+      "loss": 3.425,
+      "step": 8188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3328421535958037e-08,
+      "loss": 3.3321,
+      "step": 8189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3183217591283825e-08,
+      "loss": 3.4775,
+      "step": 8190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3038807888851035e-08,
+      "loss": 3.2959,
+      "step": 8191
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2895192451659599e-08,
+      "loss": 3.4792,
+      "step": 8192
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.275237130258289e-08,
+      "loss": 3.3617,
+      "step": 8193
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2610344464367707e-08,
+      "loss": 3.4194,
+      "step": 8194
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.246911195963485e-08,
+      "loss": 3.4572,
+      "step": 8195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2328673810877989e-08,
+      "loss": 3.4757,
+      "step": 8196
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2189030040464233e-08,
+      "loss": 3.4116,
+      "step": 8197
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2050180670634681e-08,
+      "loss": 3.4191,
+      "step": 8198
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1912125723503864e-08,
+      "loss": 3.4702,
+      "step": 8199
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.177486522105864e-08,
+      "loss": 3.3372,
+      "step": 8200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.163839918516152e-08,
+      "loss": 3.4461,
+      "step": 8201
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1502727637546229e-08,
+      "loss": 3.3331,
+      "step": 8202
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1367850599821595e-08,
+      "loss": 3.4687,
+      "step": 8203
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1233768093468766e-08,
+      "loss": 3.3942,
+      "step": 8204
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.110048013984344e-08,
+      "loss": 3.4709,
+      "step": 8205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0967986760173633e-08,
+      "loss": 3.3539,
+      "step": 8206
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0836287975561355e-08,
+      "loss": 3.4324,
+      "step": 8207
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0705383806982606e-08,
+      "loss": 3.3552,
+      "step": 8208
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0575274275285708e-08,
+      "loss": 3.3519,
+      "step": 8209
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0445959401193529e-08,
+      "loss": 3.3618,
+      "step": 8210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.031743920530126e-08,
+      "loss": 3.4136,
+      "step": 8211
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0189713708078086e-08,
+      "loss": 3.3689,
+      "step": 8212
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0062782929867177e-08,
+      "loss": 3.3964,
+      "step": 8213
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.936646890884582e-09,
+      "loss": 3.441,
+      "step": 8214
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.811305611219235e-09,
+      "loss": 3.3949,
+      "step": 8215
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.686759110834609e-09,
+      "loss": 3.3575,
+      "step": 8216
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.563007409566504e-09,
+      "loss": 3.3193,
+      "step": 8217
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.440050527124711e-09,
+      "loss": 3.355,
+      "step": 8218
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.317888483093008e-09,
+      "loss": 3.3872,
+      "step": 8219
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.196521296927496e-09,
+      "loss": 3.4089,
+      "step": 8220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.075948987958272e-09,
+      "loss": 3.4066,
+      "step": 8221
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.956171575388306e-09,
+      "loss": 3.3725,
+      "step": 8222
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.837189078294561e-09,
+      "loss": 3.3477,
+      "step": 8223
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.719001515627434e-09,
+      "loss": 3.4581,
+      "step": 8224
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.601608906210202e-09,
+      "loss": 3.3819,
+      "step": 8225
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.485011268739574e-09,
+      "loss": 3.4068,
+      "step": 8226
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.369208621786806e-09,
+      "loss": 3.5763,
+      "step": 8227
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.254200983794369e-09,
+      "loss": 3.403,
+      "step": 8228
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.139988373079832e-09,
+      "loss": 3.4857,
+      "step": 8229
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.0265708078342e-09,
+      "loss": 3.4159,
+      "step": 8230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.9139483061208e-09,
+      "loss": 3.4093,
+      "step": 8231
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.802120885876397e-09,
+      "loss": 3.4125,
+      "step": 8232
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.691088564912852e-09,
+      "loss": 3.3163,
+      "step": 8233
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.58085136091269e-09,
+      "loss": 3.3879,
+      "step": 8234
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.471409291434084e-09,
+      "loss": 3.4573,
+      "step": 8235
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.3627623739075395e-09,
+      "loss": 3.4715,
+      "step": 8236
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.2549106256364356e-09,
+      "loss": 3.43,
+      "step": 8237
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.147854063799253e-09,
+      "loss": 3.4454,
+      "step": 8238
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.041592705445688e-09,
+      "loss": 3.3952,
+      "step": 8239
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.936126567500534e-09,
+      "loss": 3.4089,
+      "step": 8240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.831455666760911e-09,
+      "loss": 3.4148,
+      "step": 8241
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.727580019897928e-09,
+      "loss": 3.4229,
+      "step": 8242
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.624499643455018e-09,
+      "loss": 3.3706,
+      "step": 8243
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.5222145538501595e-09,
+      "loss": 3.4597,
+      "step": 8244
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.420724767374209e-09,
+      "loss": 3.3249,
+      "step": 8245
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.320030300190349e-09,
+      "loss": 3.292,
+      "step": 8246
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.220131168337418e-09,
+      "loss": 3.4143,
+      "step": 8247
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.12102738772491e-09,
+      "loss": 3.3427,
+      "step": 8248
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.022718974137976e-09,
+      "loss": 3.4325,
+      "step": 8249
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9252059432335365e-09,
+      "loss": 3.319,
+      "step": 8250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8284883105419465e-09,
+      "loss": 3.444,
+      "step": 8251
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.732566091468106e-09,
+      "loss": 3.276,
+      "step": 8252
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.637439301288683e-09,
+      "loss": 3.4565,
+      "step": 8253
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.543107955154892e-09,
+      "loss": 3.4252,
+      "step": 8254
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.44957206809027e-09,
+      "loss": 3.3634,
+      "step": 8255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3568316549929e-09,
+      "loss": 3.4579,
+      "step": 8256
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.2648867306320794e-09,
+      "loss": 3.4523,
+      "step": 8257
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.17373730965276e-09,
+      "loss": 3.3542,
+      "step": 8258
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.083383406572218e-09,
+      "loss": 3.4906,
+      "step": 8259
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9938250357806085e-09,
+      "loss": 3.3621,
+      "step": 8260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.905062211541523e-09,
+      "loss": 3.4919,
+      "step": 8261
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.81709494799254e-09,
+      "loss": 3.3501,
+      "step": 8262
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.729923259143565e-09,
+      "loss": 3.3257,
+      "step": 8263
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.643547158878492e-09,
+      "loss": 3.4009,
+      "step": 8264
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.557966660954649e-09,
+      "loss": 3.4683,
+      "step": 8265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.473181779002245e-09,
+      "loss": 3.3731,
+      "step": 8266
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.389192526524366e-09,
+      "loss": 3.4924,
+      "step": 8267
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.305998916898646e-09,
+      "loss": 3.4643,
+      "step": 8268
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.223600963374485e-09,
+      "loss": 3.4433,
+      "step": 8269
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1419986790752765e-09,
+      "loss": 3.4085,
+      "step": 8270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.061192076997844e-09,
+      "loss": 3.3463,
+      "step": 8271
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.981181170011894e-09,
+      "loss": 3.4705,
+      "step": 8272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.901965970861121e-09,
+      "loss": 3.2838,
+      "step": 8273
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8235464921621e-09,
+      "loss": 3.4694,
+      "step": 8274
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.745922746403729e-09,
+      "loss": 3.4077,
+      "step": 8275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.669094745950008e-09,
+      "loss": 3.3939,
+      "step": 8276
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.593062503036704e-09,
+      "loss": 3.4166,
+      "step": 8277
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5178260297730195e-09,
+      "loss": 3.3797,
+      "step": 8278
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4433853381421465e-09,
+      "loss": 3.4329,
+      "step": 8279
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3697404400001575e-09,
+      "loss": 3.4526,
+      "step": 8280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2968913470765586e-09,
+      "loss": 3.3994,
+      "step": 8281
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2248380709731797e-09,
+      "loss": 3.4642,
+      "step": 8282
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.153580623166952e-09,
+      "loss": 3.387,
+      "step": 8283
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0831190150054646e-09,
+      "loss": 3.5336,
+      "step": 8284
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0134532577125174e-09,
+      "loss": 3.2728,
+      "step": 8285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.944583362382569e-09,
+      "loss": 3.349,
+      "step": 8286
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8765093399846235e-09,
+      "loss": 3.2833,
+      "step": 8287
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.809231201361673e-09,
+      "loss": 3.4315,
+      "step": 8288
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.74274895722737e-09,
+      "loss": 3.4781,
+      "step": 8289
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6770626181715776e-09,
+      "loss": 3.4081,
+      "step": 8290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6121721946559264e-09,
+      "loss": 3.3867,
+      "step": 8291
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.548077697014373e-09,
+      "loss": 3.4778,
+      "step": 8292
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4847791354565278e-09,
+      "loss": 3.3814,
+      "step": 8293
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.422276520062661e-09,
+      "loss": 3.3776,
+      "step": 8294
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3605698607881423e-09,
+      "loss": 3.4405,
+      "step": 8295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.299659167460666e-09,
+      "loss": 3.3542,
+      "step": 8296
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2395444497813614e-09,
+      "loss": 3.3249,
+      "step": 8297
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1802257173247908e-09,
+      "loss": 3.4581,
+      "step": 8298
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.121702979538398e-09,
+      "loss": 3.3343,
+      "step": 8299
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.06397624574306e-09,
+      "loss": 3.3993,
+      "step": 8300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.007045525132534e-09,
+      "loss": 3.4418,
+      "step": 8301
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.950910826774566e-09,
+      "loss": 3.4099,
+      "step": 8302
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.895572159609227e-09,
+      "loss": 3.4717,
+      "step": 8303
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8410295324505778e-09,
+      "loss": 3.3021,
+      "step": 8304
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7872829539855586e-09,
+      "loss": 3.34,
+      "step": 8305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7343324327739885e-09,
+      "loss": 3.3848,
+      "step": 8306
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.682177977249122e-09,
+      "loss": 3.3917,
+      "step": 8307
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6308195957182028e-09,
+      "loss": 3.3595,
+      "step": 8308
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5802572963596884e-09,
+      "loss": 3.4168,
+      "step": 8309
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5304910872282475e-09,
+      "loss": 3.3943,
+      "step": 8310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4815209762486516e-09,
+      "loss": 3.4186,
+      "step": 8311
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4333469712207725e-09,
+      "loss": 3.4346,
+      "step": 8312
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.385969079817917e-09,
+      "loss": 3.3764,
+      "step": 8313
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3393873095846055e-09,
+      "loss": 3.3506,
+      "step": 8314
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2936016679410134e-09,
+      "loss": 3.3948,
+      "step": 8315
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2486121621785307e-09,
+      "loss": 3.4885,
+      "step": 8316
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2044187994625368e-09,
+      "loss": 3.5409,
+      "step": 8317
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1610215868324004e-09,
+      "loss": 3.4464,
+      "step": 8318
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1184205311992601e-09,
+      "loss": 3.4383,
+      "step": 8319
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.076615639348244e-09,
+      "loss": 3.3581,
+      "step": 8320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0356069179373596e-09,
+      "loss": 3.4161,
+      "step": 8321
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.95394373499159e-10,
+      "loss": 3.4743,
+      "step": 8322
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.559780124362982e-10,
+      "loss": 3.3437,
+      "step": 8323
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.173578410281992e-10,
+      "loss": 3.4017,
+      "step": 8324
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.79533865425497e-10,
+      "loss": 3.4383,
+      "step": 8325
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.42506091651707e-10,
+      "loss": 3.5089,
+      "step": 8326
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.062745256043337e-10,
+      "loss": 3.2988,
+      "step": 8327
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.708391730543164e-10,
+      "loss": 3.3976,
+      "step": 8328
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.362000396454738e-10,
+      "loss": 3.3829,
+      "step": 8329
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.023571308945043e-10,
+      "loss": 3.4708,
+      "step": 8330
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.693104521909854e-10,
+      "loss": 3.4175,
+      "step": 8331
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.370600087990397e-10,
+      "loss": 3.3792,
+      "step": 8332
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.05605805855114e-10,
+      "loss": 3.4594,
+      "step": 8333
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.749478483679794e-10,
+      "loss": 3.4299,
+      "step": 8334
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.450861412215069e-10,
+      "loss": 3.3801,
+      "step": 8335
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.16020689170782e-10,
+      "loss": 3.3483,
+      "step": 8336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.877514968459895e-10,
+      "loss": 3.4736,
+      "step": 8337
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6027856874852896e-10,
+      "loss": 3.2372,
+      "step": 8338
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.336019092548993e-10,
+      "loss": 3.3684,
+      "step": 8339
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0772152261336906e-10,
+      "loss": 3.4146,
+      "step": 8340
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8263741294619625e-10,
+      "loss": 3.4745,
+      "step": 8341
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.583495842479634e-10,
+      "loss": 3.3149,
+      "step": 8342
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3485804038668747e-10,
+      "loss": 3.3731,
+      "step": 8343
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1216278510493027e-10,
+      "loss": 3.3974,
+      "step": 8344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9026382201646775e-10,
+      "loss": 3.3537,
+      "step": 8345
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6916115460962065e-10,
+      "loss": 3.3844,
+      "step": 8346
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4885478624503414e-10,
+      "loss": 3.4351,
+      "step": 8347
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2934472015734287e-10,
+      "loss": 3.4902,
+      "step": 8348
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.10630959452951e-10,
+      "loss": 3.3864,
+      "step": 8349
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9271350711336235e-10,
+      "loss": 3.4516,
+      "step": 8350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7559236599129503e-10,
+      "loss": 3.3746,
+      "step": 8351
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5926753881456704e-10,
+      "loss": 3.3646,
+      "step": 8352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4373902818276552e-10,
+      "loss": 3.3935,
+      "step": 8353
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2900683656891232e-10,
+      "loss": 3.4774,
+      "step": 8354
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1507096632001891e-10,
+      "loss": 3.386,
+      "step": 8355
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0193141965486597e-10,
+      "loss": 3.4183,
+      "step": 8356
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.958819866622392e-11,
+      "loss": 3.4151,
+      "step": 8357
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.804130532085286e-11,
+      "loss": 3.2835,
+      "step": 8358
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.729074145672698e-11,
+      "loss": 3.3968,
+      "step": 8359
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.733650878692043e-11,
+      "loss": 3.4057,
+      "step": 8360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8178608896276566e-11,
+      "loss": 3.3603,
+      "step": 8361
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.981704324362845e-11,
+      "loss": 3.5238,
+      "step": 8362
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2251813160133486e-11,
+      "loss": 3.3204,
+      "step": 8363
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.548291985149387e-11,
+      "loss": 3.3809,
+      "step": 8364
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.951036439573617e-11,
+      "loss": 3.431,
+      "step": 8365
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.433414774321129e-11,
+      "loss": 3.3775,
+      "step": 8366
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.954270719370051e-12,
+      "loss": 3.5233,
+      "step": 8367
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.370734021432512e-12,
+      "loss": 3.4714,
+      "step": 8368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5835382200533065e-12,
+      "loss": 3.3192,
+      "step": 8369
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.592683759321645e-12,
+      "loss": 3.3737,
+      "step": 8370
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9817095565108933e-13,
+      "loss": 3.4083,
+      "step": 8371
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 3.3638,
+      "step": 8372
+    },
+    {
+      "epoch": 1.0,
+      "step": 8372,
+      "total_flos": 1.0794508479071519e+18,
+      "train_loss": 3.4384684080833803,
+      "train_runtime": 9209.2881,
+      "train_samples_per_second": 58.18,
+      "train_steps_per_second": 0.909
+    }
+  ],
+  "max_steps": 8372,
+  "num_train_epochs": 1,
+  "total_flos": 1.0794508479071519e+18,
+  "trial_name": null,
+  "trial_params": null
+}