diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,52344 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 8719,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 1.59,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.5555,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 1.191,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.4489,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.0551,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.3076,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 1.1791,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 1.5894,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-06,
+      "loss": 0.8208,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.2835,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4e-06,
+      "loss": 1.2915,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.8957,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2e-06,
+      "loss": 0.9646,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 1.4183,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 1.2379,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.3433,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 1.127,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2e-06,
+      "loss": 0.999,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.6e-06,
+      "loss": 1.0731,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.4052,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 1.9277,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.8e-06,
+      "loss": 1.3095,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.2e-06,
+      "loss": 0.7437,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 1.2866,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 0.9299,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.04e-05,
+      "loss": 1.646,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.08e-05,
+      "loss": 1.4822,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 1.0893,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.16e-05,
+      "loss": 1.3019,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 1.099,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.24e-05,
+      "loss": 1.4477,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 1.0746,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.32e-05,
+      "loss": 1.3542,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 1.1927,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 1.5071,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.44e-05,
+      "loss": 1.372,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.48e-05,
+      "loss": 1.4374,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.52e-05,
+      "loss": 1.1368,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.56e-05,
+      "loss": 0.8047,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.8055,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6400000000000002e-05,
+      "loss": 1.2335,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.9522,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7199999999999998e-05,
+      "loss": 1.2284,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.76e-05,
+      "loss": 1.1621,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-05,
+      "loss": 0.5993,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.84e-05,
+      "loss": 1.4192,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.88e-05,
+      "loss": 1.2076,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.9123,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 0.8213,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 1.2975,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.04e-05,
+      "loss": 1.276,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.08e-05,
+      "loss": 0.593,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.12e-05,
+      "loss": 1.2927,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.16e-05,
+      "loss": 1.3843,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 1.186,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 1.3208,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 0.4792,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.32e-05,
+      "loss": 0.7168,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.36e-05,
+      "loss": 1.0186,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-05,
+      "loss": 0.7344,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.44e-05,
+      "loss": 1.5661,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.48e-05,
+      "loss": 1.4551,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5200000000000003e-05,
+      "loss": 1.722,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 1.0609,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 1.276,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.64e-05,
+      "loss": 1.4054,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6800000000000004e-05,
+      "loss": 1.012,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 1.3638,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 1.5016,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 1.249,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.84e-05,
+      "loss": 1.3246,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.88e-05,
+      "loss": 1.3058,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9199999999999998e-05,
+      "loss": 1.1507,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.96e-05,
+      "loss": 0.8627,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 0.9712,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.04e-05,
+      "loss": 1.2334,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.08e-05,
+      "loss": 1.4454,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.12e-05,
+      "loss": 1.389,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.16e-05,
+      "loss": 1.3811,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.034,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.24e-05,
+      "loss": 0.9957,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2800000000000004e-05,
+      "loss": 0.8218,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.32e-05,
+      "loss": 0.9987,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 1.3163,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 1.2639,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 1.2931,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.48e-05,
+      "loss": 1.3636,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.52e-05,
+      "loss": 1.038,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.56e-05,
+      "loss": 1.2559,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-05,
+      "loss": 1.1861,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6400000000000004e-05,
+      "loss": 1.139,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.68e-05,
+      "loss": 0.7344,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.72e-05,
+      "loss": 1.2966,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.76e-05,
+      "loss": 1.4865,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8e-05,
+      "loss": 1.4908,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 1.1899,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.88e-05,
+      "loss": 1.1443,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9200000000000004e-05,
+      "loss": 1.1535,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.960000000000001e-05,
+      "loss": 1.3581,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-05,
+      "loss": 1.4928,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0400000000000006e-05,
+      "loss": 1.3599,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.08e-05,
+      "loss": 1.5214,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.12e-05,
+      "loss": 1.3349,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.16e-05,
+      "loss": 1.0022,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2e-05,
+      "loss": 1.3437,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.24e-05,
+      "loss": 1.3304,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2800000000000004e-05,
+      "loss": 1.3195,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.32e-05,
+      "loss": 1.077,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.36e-05,
+      "loss": 0.6842,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.9388,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.44e-05,
+      "loss": 0.5818,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4800000000000005e-05,
+      "loss": 1.2849,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.52e-05,
+      "loss": 1.0991,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 0.919,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 1.5382,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.64e-05,
+      "loss": 1.3063,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6800000000000006e-05,
+      "loss": 1.5382,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.72e-05,
+      "loss": 1.2625,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.76e-05,
+      "loss": 1.0472,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8e-05,
+      "loss": 0.8967,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8400000000000004e-05,
+      "loss": 1.1595,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.88e-05,
+      "loss": 1.006,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.92e-05,
+      "loss": 1.4412,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.96e-05,
+      "loss": 1.4794,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-05,
+      "loss": 0.8258,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.0400000000000005e-05,
+      "loss": 0.797,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.08e-05,
+      "loss": 1.6248,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.1200000000000004e-05,
+      "loss": 1.8287,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.16e-05,
+      "loss": 0.5549,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 1.1347,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2400000000000007e-05,
+      "loss": 0.9722,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.28e-05,
+      "loss": 1.0404,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3200000000000006e-05,
+      "loss": 1.0573,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.360000000000001e-05,
+      "loss": 0.91,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 1.0428,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.440000000000001e-05,
+      "loss": 1.2562,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4800000000000004e-05,
+      "loss": 1.3845,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.520000000000001e-05,
+      "loss": 1.1322,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.560000000000001e-05,
+      "loss": 1.3707,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 1.3682,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 1.4251,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.68e-05,
+      "loss": 1.2812,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.72e-05,
+      "loss": 1.0914,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.76e-05,
+      "loss": 1.1609,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8e-05,
+      "loss": 1.0848,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8399999999999997e-05,
+      "loss": 1.8199,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.88e-05,
+      "loss": 1.3316,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.92e-05,
+      "loss": 0.6912,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.96e-05,
+      "loss": 1.5523,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6e-05,
+      "loss": 1.069,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.04e-05,
+      "loss": 1.3569,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.08e-05,
+      "loss": 1.4175,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.12e-05,
+      "loss": 1.2685,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.16e-05,
+      "loss": 0.6854,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.2e-05,
+      "loss": 1.3181,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.24e-05,
+      "loss": 1.1677,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.280000000000001e-05,
+      "loss": 0.5815,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.32e-05,
+      "loss": 1.0517,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.36e-05,
+      "loss": 1.2837,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 1.4506,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.440000000000001e-05,
+      "loss": 1.6701,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.48e-05,
+      "loss": 0.9901,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.52e-05,
+      "loss": 1.5572,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.560000000000001e-05,
+      "loss": 1.3769,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.6e-05,
+      "loss": 0.6458,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.64e-05,
+      "loss": 0.9199,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.680000000000001e-05,
+      "loss": 1.6495,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.720000000000001e-05,
+      "loss": 0.696,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.76e-05,
+      "loss": 0.8833,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 0.7514,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.840000000000001e-05,
+      "loss": 1.1927,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.879999999999999e-05,
+      "loss": 1.3598,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.92e-05,
+      "loss": 0.9694,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.96e-05,
+      "loss": 0.7477,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7e-05,
+      "loss": 0.9523,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.04e-05,
+      "loss": 0.7164,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.08e-05,
+      "loss": 1.4172,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.12e-05,
+      "loss": 1.4544,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.16e-05,
+      "loss": 1.3846,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.2e-05,
+      "loss": 1.632,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.24e-05,
+      "loss": 1.2995,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.280000000000001e-05,
+      "loss": 1.0928,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.32e-05,
+      "loss": 0.7626,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.36e-05,
+      "loss": 1.4558,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.4e-05,
+      "loss": 1.4191,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.44e-05,
+      "loss": 1.9281,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.48e-05,
+      "loss": 1.1183,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.52e-05,
+      "loss": 1.2422,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.560000000000001e-05,
+      "loss": 1.1705,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.6e-05,
+      "loss": 1.3913,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.64e-05,
+      "loss": 1.3559,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.680000000000001e-05,
+      "loss": 1.1518,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.72e-05,
+      "loss": 1.3318,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.76e-05,
+      "loss": 1.4665,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 1.3557,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.840000000000001e-05,
+      "loss": 1.4249,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.88e-05,
+      "loss": 1.4546,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.920000000000001e-05,
+      "loss": 1.1145,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.960000000000001e-05,
+      "loss": 1.4066,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8e-05,
+      "loss": 0.6831,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.04e-05,
+      "loss": 0.8944,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.080000000000001e-05,
+      "loss": 1.2159,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.120000000000001e-05,
+      "loss": 1.3412,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.16e-05,
+      "loss": 0.6482,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.2e-05,
+      "loss": 1.3566,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.24e-05,
+      "loss": 1.5643,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.28e-05,
+      "loss": 0.6865,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.32e-05,
+      "loss": 0.6086,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.36e-05,
+      "loss": 1.3165,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.4e-05,
+      "loss": 1.2368,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.44e-05,
+      "loss": 1.2546,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.48e-05,
+      "loss": 1.4033,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.52e-05,
+      "loss": 0.9968,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.560000000000001e-05,
+      "loss": 1.1878,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.6e-05,
+      "loss": 0.9221,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.64e-05,
+      "loss": 0.9785,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.680000000000001e-05,
+      "loss": 1.0646,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.72e-05,
+      "loss": 1.6976,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.76e-05,
+      "loss": 0.5068,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 0.6858,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.840000000000001e-05,
+      "loss": 1.0887,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.88e-05,
+      "loss": 0.9805,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.92e-05,
+      "loss": 1.1805,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.960000000000001e-05,
+      "loss": 1.4833,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9e-05,
+      "loss": 1.5709,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.04e-05,
+      "loss": 0.7017,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.080000000000001e-05,
+      "loss": 0.7716,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.120000000000001e-05,
+      "loss": 0.6977,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.16e-05,
+      "loss": 1.4582,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 1.4483,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.240000000000001e-05,
+      "loss": 1.1269,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.28e-05,
+      "loss": 1.0563,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.320000000000002e-05,
+      "loss": 1.2786,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.360000000000001e-05,
+      "loss": 1.6226,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.4e-05,
+      "loss": 1.4762,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.44e-05,
+      "loss": 0.8827,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.48e-05,
+      "loss": 1.3293,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.52e-05,
+      "loss": 0.6789,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.56e-05,
+      "loss": 1.6894,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.6e-05,
+      "loss": 1.6629,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.64e-05,
+      "loss": 1.1783,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.680000000000001e-05,
+      "loss": 1.6919,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.72e-05,
+      "loss": 1.2123,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.76e-05,
+      "loss": 1.0399,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.8e-05,
+      "loss": 0.6128,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.84e-05,
+      "loss": 1.3906,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.88e-05,
+      "loss": 1.2509,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.92e-05,
+      "loss": 1.0034,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.960000000000001e-05,
+      "loss": 1.2799,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001,
+      "loss": 1.2298,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001004,
+      "loss": 1.1145,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010080000000000001,
+      "loss": 1.2134,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010120000000000001,
+      "loss": 1.2844,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001016,
+      "loss": 1.4757,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010200000000000001,
+      "loss": 1.164,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010240000000000001,
+      "loss": 1.2521,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001028,
+      "loss": 1.6354,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001032,
+      "loss": 0.8187,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010360000000000001,
+      "loss": 1.5141,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010400000000000001,
+      "loss": 0.9219,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001044,
+      "loss": 1.4559,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010480000000000001,
+      "loss": 1.0513,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010520000000000001,
+      "loss": 1.2838,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001056,
+      "loss": 1.4145,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010600000000000002,
+      "loss": 1.3919,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010640000000000001,
+      "loss": 1.2805,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010680000000000001,
+      "loss": 1.3145,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010720000000000002,
+      "loss": 1.4292,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010760000000000001,
+      "loss": 1.4066,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010800000000000001,
+      "loss": 1.2635,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010840000000000002,
+      "loss": 1.1356,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010880000000000002,
+      "loss": 1.1634,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010920000000000001,
+      "loss": 1.2752,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010960000000000001,
+      "loss": 1.0241,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011000000000000002,
+      "loss": 0.6824,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011040000000000001,
+      "loss": 1.0373,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011080000000000001,
+      "loss": 1.2445,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011120000000000002,
+      "loss": 1.0488,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011160000000000002,
+      "loss": 1.3936,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011200000000000001,
+      "loss": 0.9844,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011240000000000002,
+      "loss": 0.7345,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011279999999999999,
+      "loss": 0.8018,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001132,
+      "loss": 0.8089,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001136,
+      "loss": 0.9728,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 1.0466,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001144,
+      "loss": 1.8754,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001148,
+      "loss": 0.8496,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001152,
+      "loss": 1.1578,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011559999999999999,
+      "loss": 1.2558,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000116,
+      "loss": 0.7546,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001164,
+      "loss": 1.3444,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011679999999999999,
+      "loss": 1.3222,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001172,
+      "loss": 1.0735,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001176,
+      "loss": 0.9632,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000118,
+      "loss": 1.3831,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001184,
+      "loss": 0.8488,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001188,
+      "loss": 1.1903,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001192,
+      "loss": 1.007,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011960000000000001,
+      "loss": 1.2658,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00012,
+      "loss": 1.3398,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001204,
+      "loss": 1.4876,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001208,
+      "loss": 0.8235,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001212,
+      "loss": 1.0996,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001216,
+      "loss": 1.0975,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000122,
+      "loss": 1.3598,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001224,
+      "loss": 1.3973,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001228,
+      "loss": 1.216,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001232,
+      "loss": 1.0544,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001236,
+      "loss": 0.9963,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000124,
+      "loss": 1.1577,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00012440000000000002,
+      "loss": 1.5108,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001248,
+      "loss": 1.3321,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001252,
+      "loss": 1.3267,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00012560000000000002,
+      "loss": 1.3929,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000126,
+      "loss": 1.3267,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001264,
+      "loss": 1.2927,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00012680000000000002,
+      "loss": 0.9114,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001272,
+      "loss": 1.2153,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001276,
+      "loss": 1.2827,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00012800000000000002,
+      "loss": 1.5286,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001284,
+      "loss": 1.1947,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00012880000000000001,
+      "loss": 0.9286,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00012920000000000002,
+      "loss": 0.5431,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001296,
+      "loss": 1.1864,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 1.3632,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001304,
+      "loss": 1.5343,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001308,
+      "loss": 1.1085,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013120000000000002,
+      "loss": 1.042,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001316,
+      "loss": 0.7478,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000132,
+      "loss": 1.5384,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013240000000000002,
+      "loss": 1.3021,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001328,
+      "loss": 1.2257,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001332,
+      "loss": 1.2077,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013360000000000002,
+      "loss": 0.7453,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000134,
+      "loss": 1.1783,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013440000000000001,
+      "loss": 1.4051,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013480000000000002,
+      "loss": 1.2837,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001352,
+      "loss": 1.5519,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013560000000000002,
+      "loss": 0.9478,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013600000000000003,
+      "loss": 1.0151,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001364,
+      "loss": 1.2267,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013680000000000002,
+      "loss": 1.2461,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013720000000000003,
+      "loss": 1.4909,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013759999999999998,
+      "loss": 0.8472,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000138,
+      "loss": 1.9571,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001384,
+      "loss": 1.2572,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013879999999999999,
+      "loss": 0.9367,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001392,
+      "loss": 0.8206,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001396,
+      "loss": 0.8334,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00014,
+      "loss": 1.5481,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001404,
+      "loss": 0.8156,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001408,
+      "loss": 0.7433,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001412,
+      "loss": 1.0023,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001416,
+      "loss": 1.5298,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000142,
+      "loss": 0.5575,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001424,
+      "loss": 1.2796,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001428,
+      "loss": 1.2481,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001432,
+      "loss": 1.2266,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001436,
+      "loss": 1.4247,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000144,
+      "loss": 1.2168,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001444,
+      "loss": 0.9528,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001448,
+      "loss": 0.885,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001452,
+      "loss": 0.9612,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00014560000000000002,
+      "loss": 1.4621,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000146,
+      "loss": 1.2824,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001464,
+      "loss": 0.6158,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00014680000000000002,
+      "loss": 1.8492,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001472,
+      "loss": 1.1849,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001476,
+      "loss": 1.1406,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000148,
+      "loss": 1.3956,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001484,
+      "loss": 1.205,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001488,
+      "loss": 1.4385,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001492,
+      "loss": 0.7829,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001496,
+      "loss": 0.7504,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 1.2074,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001504,
+      "loss": 1.0478,
+      "step": 376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001508,
+      "loss": 1.2325,
+      "step": 377
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00015120000000000002,
+      "loss": 0.7488,
+      "step": 378
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001516,
+      "loss": 0.7718,
+      "step": 379
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000152,
+      "loss": 0.4098,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00015240000000000002,
+      "loss": 1.2784,
+      "step": 381
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001528,
+      "loss": 1.524,
+      "step": 382
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001532,
+      "loss": 0.6129,
+      "step": 383
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00015360000000000002,
+      "loss": 0.7016,
+      "step": 384
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000154,
+      "loss": 1.1314,
+      "step": 385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001544,
+      "loss": 1.2857,
+      "step": 386
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00015480000000000002,
+      "loss": 1.4161,
+      "step": 387
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001552,
+      "loss": 0.9287,
+      "step": 388
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00015560000000000001,
+      "loss": 1.0437,
+      "step": 389
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00015600000000000002,
+      "loss": 0.9181,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001564,
+      "loss": 1.5682,
+      "step": 391
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00015680000000000002,
+      "loss": 1.2175,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00015720000000000003,
+      "loss": 1.219,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001576,
+      "loss": 0.8228,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00015800000000000002,
+      "loss": 0.6061,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00015840000000000003,
+      "loss": 1.3276,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001588,
+      "loss": 0.8193,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00015920000000000002,
+      "loss": 1.3556,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001596,
+      "loss": 1.4371,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00016,
+      "loss": 0.9836,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00016040000000000002,
+      "loss": 1.594,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001608,
+      "loss": 1.3361,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00016120000000000002,
+      "loss": 1.2394,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00016160000000000002,
+      "loss": 1.3588,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000162,
+      "loss": 1.3142,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00016240000000000002,
+      "loss": 1.1509,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001628,
+      "loss": 1.3863,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001632,
+      "loss": 1.4315,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001636,
+      "loss": 1.4643,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000164,
+      "loss": 1.5086,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001644,
+      "loss": 1.362,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001648,
+      "loss": 1.7128,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001652,
+      "loss": 0.469,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001656,
+      "loss": 1.1058,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000166,
+      "loss": 1.3638,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001664,
+      "loss": 0.915,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001668,
+      "loss": 1.3962,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001672,
+      "loss": 1.0908,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001676,
+      "loss": 0.4318,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000168,
+      "loss": 1.3211,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001684,
+      "loss": 1.5239,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001688,
+      "loss": 0.6965,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001692,
+      "loss": 0.6567,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001696,
+      "loss": 1.0678,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017,
+      "loss": 0.82,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001704,
+      "loss": 0.9984,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001708,
+      "loss": 0.7419,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017120000000000001,
+      "loss": 1.3055,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001716,
+      "loss": 0.8651,
+      "step": 429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000172,
+      "loss": 1.153,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017240000000000002,
+      "loss": 1.1785,
+      "step": 431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001728,
+      "loss": 1.2746,
+      "step": 432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001732,
+      "loss": 1.4459,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017360000000000002,
+      "loss": 1.4755,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000174,
+      "loss": 0.8247,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001744,
+      "loss": 1.0984,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017480000000000002,
+      "loss": 1.0488,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001752,
+      "loss": 1.2797,
+      "step": 438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001756,
+      "loss": 1.8948,
+      "step": 439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017600000000000002,
+      "loss": 1.0739,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001764,
+      "loss": 1.1384,
+      "step": 441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017680000000000001,
+      "loss": 0.8979,
+      "step": 442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001772,
+      "loss": 1.1173,
+      "step": 443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001776,
+      "loss": 1.1107,
+      "step": 444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017800000000000002,
+      "loss": 1.343,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001784,
+      "loss": 1.2492,
+      "step": 446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001788,
+      "loss": 0.5339,
+      "step": 447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017920000000000002,
+      "loss": 1.1779,
+      "step": 448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001796,
+      "loss": 0.8903,
+      "step": 449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00018,
+      "loss": 1.044,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00018040000000000002,
+      "loss": 1.2839,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001808,
+      "loss": 1.319,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001812,
+      "loss": 1.0491,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00018160000000000002,
+      "loss": 0.5155,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000182,
+      "loss": 1.1035,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00018240000000000002,
+      "loss": 1.3109,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00018280000000000003,
+      "loss": 1.1736,
+      "step": 457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001832,
+      "loss": 1.3392,
+      "step": 458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00018360000000000002,
+      "loss": 1.0338,
+      "step": 459
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00018400000000000003,
+      "loss": 1.2713,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001844,
+      "loss": 1.4058,
+      "step": 461
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00018480000000000002,
+      "loss": 1.1222,
+      "step": 462
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00018520000000000003,
+      "loss": 1.2303,
+      "step": 463
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001856,
+      "loss": 1.406,
+      "step": 464
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00018600000000000002,
+      "loss": 1.1419,
+      "step": 465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00018640000000000003,
+      "loss": 0.5709,
+      "step": 466
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00018680000000000001,
+      "loss": 1.4781,
+      "step": 467
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00018720000000000002,
+      "loss": 0.7652,
+      "step": 468
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001876,
+      "loss": 1.5587,
+      "step": 469
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000188,
+      "loss": 1.4432,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001884,
+      "loss": 0.9523,
+      "step": 471
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001888,
+      "loss": 1.3929,
+      "step": 472
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001892,
+      "loss": 1.589,
+      "step": 473
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001896,
+      "loss": 1.1288,
+      "step": 474
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019,
+      "loss": 1.046,
+      "step": 475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001904,
+      "loss": 1.3956,
+      "step": 476
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001908,
+      "loss": 0.5756,
+      "step": 477
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001912,
+      "loss": 1.1035,
+      "step": 478
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001916,
+      "loss": 0.9761,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000192,
+      "loss": 1.3028,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019240000000000001,
+      "loss": 1.135,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001928,
+      "loss": 0.5956,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001932,
+      "loss": 1.7213,
+      "step": 483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019360000000000002,
+      "loss": 1.3531,
+      "step": 484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000194,
+      "loss": 1.333,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001944,
+      "loss": 1.017,
+      "step": 486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001948,
+      "loss": 1.035,
+      "step": 487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001952,
+      "loss": 0.635,
+      "step": 488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001956,
+      "loss": 1.4956,
+      "step": 489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000196,
+      "loss": 1.1555,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001964,
+      "loss": 1.3951,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001968,
+      "loss": 1.1295,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001972,
+      "loss": 1.2578,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001976,
+      "loss": 1.3786,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019800000000000002,
+      "loss": 0.8149,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001984,
+      "loss": 1.1619,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001988,
+      "loss": 1.1613,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019920000000000002,
+      "loss": 1.2329,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996,
+      "loss": 1.3406,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.8604,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999999269480553,
+      "loss": 1.3657,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999997077922306,
+      "loss": 1.1923,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999993425325591,
+      "loss": 1.1428,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999988311690935,
+      "loss": 1.151,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999981737019084,
+      "loss": 1.3871,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999973701311004,
+      "loss": 0.9818,
+      "step": 506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999964204567863,
+      "loss": 1.2752,
+      "step": 507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999953246791055,
+      "loss": 1.266,
+      "step": 508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999940827982176,
+      "loss": 1.1295,
+      "step": 509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999992694814304,
+      "loss": 0.8072,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999911607275681,
+      "loss": 1.1701,
+      "step": 511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999894805382336,
+      "loss": 1.2857,
+      "step": 512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999876542465457,
+      "loss": 0.5385,
+      "step": 513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999856818527718,
+      "loss": 0.5568,
+      "step": 514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999835633571994,
+      "loss": 0.3956,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999981298760139,
+      "loss": 1.4575,
+      "step": 516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999788880619205,
+      "loss": 0.9799,
+      "step": 517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999763312628964,
+      "loss": 0.846,
+      "step": 518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999736283634408,
+      "loss": 0.5316,
+      "step": 519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999970779363948,
+      "loss": 0.643,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999677842648343,
+      "loss": 1.8141,
+      "step": 521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999646430665378,
+      "loss": 1.1789,
+      "step": 522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999613557695166,
+      "loss": 0.9451,
+      "step": 523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999579223742514,
+      "loss": 1.2998,
+      "step": 524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999543428812444,
+      "loss": 0.6677,
+      "step": 525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999506172910177,
+      "loss": 1.1652,
+      "step": 526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999946745604116,
+      "loss": 1.3592,
+      "step": 527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999942727821105,
+      "loss": 1.0333,
+      "step": 528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999385639425717,
+      "loss": 0.6573,
+      "step": 529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999342539691245,
+      "loss": 0.7089,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999297979013929,
+      "loss": 1.4819,
+      "step": 531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999925195740028,
+      "loss": 1.4256,
+      "step": 532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999204474857024,
+      "loss": 1.0266,
+      "step": 533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999155531391096,
+      "loss": 1.2174,
+      "step": 534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999105127009646,
+      "loss": 1.1979,
+      "step": 535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999053261720044,
+      "loss": 1.3622,
+      "step": 536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019998999935529863,
+      "loss": 0.8187,
+      "step": 537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019998945148446895,
+      "loss": 0.6358,
+      "step": 538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019998888900479145,
+      "loss": 1.2633,
+      "step": 539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019998831191634828,
+      "loss": 1.2144,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999877202192238,
+      "loss": 1.0845,
+      "step": 541
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019998711391350446,
+      "loss": 0.6614,
+      "step": 542
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019998649299927878,
+      "loss": 0.7738,
+      "step": 543
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019998585747663756,
+      "loss": 1.1697,
+      "step": 544
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999852073456736,
+      "loss": 1.8047,
+      "step": 545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999845426064819,
+      "loss": 1.3518,
+      "step": 546
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019998386325915962,
+      "loss": 0.8803,
+      "step": 547
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019998316930380593,
+      "loss": 1.2598,
+      "step": 548
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019998246074052228,
+      "loss": 1.4039,
+      "step": 549
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999817375694122,
+      "loss": 1.0519,
+      "step": 550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019998099979058132,
+      "loss": 1.2682,
+      "step": 551
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019998024740413745,
+      "loss": 1.5572,
+      "step": 552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019997948041019047,
+      "loss": 1.0906,
+      "step": 553
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019997869880885252,
+      "loss": 1.2658,
+      "step": 554
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019997790260023774,
+      "loss": 1.1821,
+      "step": 555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019997709178446246,
+      "loss": 0.8092,
+      "step": 556
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019997626636164518,
+      "loss": 1.6625,
+      "step": 557
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019997542633190646,
+      "loss": 1.4741,
+      "step": 558
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019997457169536906,
+      "loss": 1.1457,
+      "step": 559
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019997370245215782,
+      "loss": 1.2023,
+      "step": 560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999728186023998,
+      "loss": 1.1092,
+      "step": 561
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000199971920146224,
+      "loss": 1.3356,
+      "step": 562
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019997100708376184,
+      "loss": 1.3148,
+      "step": 563
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019997007941514663,
+      "loss": 1.7354,
+      "step": 564
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999691371405139,
+      "loss": 1.3358,
+      "step": 565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999681802600014,
+      "loss": 1.5468,
+      "step": 566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001999672087737489,
+      "loss": 1.2894,
+      "step": 567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019996622268189827,
+      "loss": 1.2094,
+      "step": 568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019996522198459363,
+      "loss": 1.0498,
+      "step": 569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019996420668198122,
+      "loss": 0.5509,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019996317677420936,
+      "loss": 0.5036,
+      "step": 571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019996213226142852,
+      "loss": 0.7135,
+      "step": 572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019996107314379125,
+      "loss": 1.4116,
+      "step": 573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019995999942145238,
+      "loss": 0.9975,
+      "step": 574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019995891109456874,
+      "loss": 1.2534,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019995780816329934,
+      "loss": 0.4886,
+      "step": 576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019995669062780535,
+      "loss": 1.0747,
+      "step": 577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019995555848825004,
+      "loss": 0.8795,
+      "step": 578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019995441174479876,
+      "loss": 1.1114,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019995325039761912,
+      "loss": 0.8553,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019995207444688075,
+      "loss": 1.4004,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019995088389275552,
+      "loss": 1.1237,
+      "step": 582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019994967873541734,
+      "loss": 1.0266,
+      "step": 583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019994845897504228,
+      "loss": 1.2057,
+      "step": 584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019994722461180858,
+      "loss": 0.9819,
+      "step": 585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019994597564589655,
+      "loss": 1.1837,
+      "step": 586
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019994471207748868,
+      "loss": 1.2978,
+      "step": 587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001999434339067696,
+      "loss": 0.9465,
+      "step": 588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019994214113392604,
+      "loss": 0.622,
+      "step": 589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019994083375914686,
+      "loss": 1.5697,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019993951178262314,
+      "loss": 0.7436,
+      "step": 591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019993817520454793,
+      "loss": 1.4479,
+      "step": 592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019993682402511662,
+      "loss": 0.9147,
+      "step": 593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019993545824452654,
+      "loss": 0.7471,
+      "step": 594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019993407786297724,
+      "loss": 0.8088,
+      "step": 595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001999326828806704,
+      "loss": 1.0861,
+      "step": 596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001999312732978099,
+      "loss": 0.7142,
+      "step": 597
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019992984911460163,
+      "loss": 1.0844,
+      "step": 598
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019992841033125366,
+      "loss": 1.1889,
+      "step": 599
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019992695694797622,
+      "loss": 1.4044,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019992548896498162,
+      "loss": 0.8367,
+      "step": 601
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019992400638248442,
+      "loss": 1.4839,
+      "step": 602
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019992250920070114,
+      "loss": 1.4881,
+      "step": 603
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001999209974198506,
+      "loss": 1.2914,
+      "step": 604
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019991947104015364,
+      "loss": 0.9014,
+      "step": 605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019991793006183324,
+      "loss": 1.3261,
+      "step": 606
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001999163744851146,
+      "loss": 1.2216,
+      "step": 607
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019991480431022496,
+      "loss": 0.7003,
+      "step": 608
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019991321953739374,
+      "loss": 0.7713,
+      "step": 609
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001999116201668525,
+      "loss": 1.3797,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019991000619883488,
+      "loss": 1.4975,
+      "step": 611
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019990837763357667,
+      "loss": 1.5755,
+      "step": 612
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019990673447131587,
+      "loss": 0.9428,
+      "step": 613
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019990507671229253,
+      "loss": 0.5565,
+      "step": 614
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019990340435674882,
+      "loss": 0.654,
+      "step": 615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001999017174049292,
+      "loss": 0.8069,
+      "step": 616
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019990001585707992,
+      "loss": 0.6081,
+      "step": 617
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998982997134498,
+      "loss": 1.4694,
+      "step": 618
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019989656897428945,
+      "loss": 1.1468,
+      "step": 619
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019989482363985176,
+      "loss": 1.7123,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019989306371039176,
+      "loss": 0.7536,
+      "step": 621
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019989128918616658,
+      "loss": 1.0272,
+      "step": 622
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019988950006743546,
+      "loss": 0.7634,
+      "step": 623
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998876963544598,
+      "loss": 1.1632,
+      "step": 624
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019988587804750313,
+      "loss": 0.8252,
+      "step": 625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019988404514683112,
+      "loss": 1.2378,
+      "step": 626
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019988219765271158,
+      "loss": 0.975,
+      "step": 627
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998803355654144,
+      "loss": 1.0961,
+      "step": 628
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019987845888521166,
+      "loss": 1.7001,
+      "step": 629
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019987656761237755,
+      "loss": 0.9046,
+      "step": 630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998746617471884,
+      "loss": 1.4906,
+      "step": 631
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019987274128992265,
+      "loss": 0.7629,
+      "step": 632
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019987080624086088,
+      "loss": 0.8317,
+      "step": 633
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998688566002858,
+      "loss": 0.7619,
+      "step": 634
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019986689236848232,
+      "loss": 1.2733,
+      "step": 635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019986491354573739,
+      "loss": 1.4406,
+      "step": 636
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019986292013234004,
+      "loss": 0.9924,
+      "step": 637
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019986091212858168,
+      "loss": 0.7372,
+      "step": 638
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019985888953475553,
+      "loss": 1.4288,
+      "step": 639
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019985685235115718,
+      "loss": 1.2048,
+      "step": 640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019985480057808425,
+      "loss": 0.7091,
+      "step": 641
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019985273421583655,
+      "loss": 0.8814,
+      "step": 642
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019985065326471593,
+      "loss": 0.5736,
+      "step": 643
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019984855772502646,
+      "loss": 0.6794,
+      "step": 644
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019984644759707428,
+      "loss": 0.9241,
+      "step": 645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998443228811677,
+      "loss": 1.1766,
+      "step": 646
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019984218357761715,
+      "loss": 1.5283,
+      "step": 647
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019984002968673518,
+      "loss": 1.1622,
+      "step": 648
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019983786120883652,
+      "loss": 1.511,
+      "step": 649
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019983567814423796,
+      "loss": 1.2044,
+      "step": 650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019983348049325844,
+      "loss": 0.698,
+      "step": 651
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019983126825621908,
+      "loss": 0.5933,
+      "step": 652
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019982904143344308,
+      "loss": 1.3781,
+      "step": 653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998268000252558,
+      "loss": 0.9192,
+      "step": 654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019982454403198467,
+      "loss": 1.8633,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019982227345395937,
+      "loss": 0.9605,
+      "step": 656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998199882915116,
+      "loss": 0.6932,
+      "step": 657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019981768854497524,
+      "loss": 1.3127,
+      "step": 658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998153742146863,
+      "loss": 1.3698,
+      "step": 659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019981304530098287,
+      "loss": 1.3577,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998107018042053,
+      "loss": 1.2623,
+      "step": 661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998083437246959,
+      "loss": 1.2018,
+      "step": 662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998059710627992,
+      "loss": 1.4044,
+      "step": 663
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998035838188619,
+      "loss": 1.0575,
+      "step": 664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998011819932328,
+      "loss": 0.8833,
+      "step": 665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001997987655862627,
+      "loss": 1.5855,
+      "step": 666
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019979633459830482,
+      "loss": 1.1529,
+      "step": 667
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001997938890297142,
+      "loss": 1.433,
+      "step": 668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019979142888084818,
+      "loss": 1.6981,
+      "step": 669
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019978895415206623,
+      "loss": 1.2225,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001997864648437299,
+      "loss": 1.3031,
+      "step": 671
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019978396095620286,
+      "loss": 1.4388,
+      "step": 672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019978144248985097,
+      "loss": 1.0294,
+      "step": 673
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019977890944504218,
+      "loss": 1.4007,
+      "step": 674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001997763618221466,
+      "loss": 1.6759,
+      "step": 675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019977379962153642,
+      "loss": 0.8156,
+      "step": 676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000199771222843586,
+      "loss": 1.3731,
+      "step": 677
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019976863148867178,
+      "loss": 1.1592,
+      "step": 678
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001997660255571724,
+      "loss": 1.1101,
+      "step": 679
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001997634050494686,
+      "loss": 1.19,
+      "step": 680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019976076996594327,
+      "loss": 0.5617,
+      "step": 681
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019975812030698133,
+      "loss": 1.0919,
+      "step": 682
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019975545607296997,
+      "loss": 1.2496,
+      "step": 683
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019975277726429842,
+      "loss": 1.3485,
+      "step": 684
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001997500838813581,
+      "loss": 0.6853,
+      "step": 685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019974737592454245,
+      "loss": 1.1215,
+      "step": 686
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001997446533942472,
+      "loss": 0.7215,
+      "step": 687
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019974191629087008,
+      "loss": 1.7273,
+      "step": 688
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019973916461481095,
+      "loss": 1.2566,
+      "step": 689
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019973639836647192,
+      "loss": 1.4556,
+      "step": 690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019973361754625709,
+      "loss": 1.1767,
+      "step": 691
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019973082215457276,
+      "loss": 1.4416,
+      "step": 692
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019972801219182736,
+      "loss": 1.472,
+      "step": 693
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019972518765843145,
+      "loss": 1.3074,
+      "step": 694
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001997223485547977,
+      "loss": 1.4008,
+      "step": 695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001997194948813409,
+      "loss": 1.2495,
+      "step": 696
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019971662663847796,
+      "loss": 0.7441,
+      "step": 697
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019971374382662802,
+      "loss": 1.0709,
+      "step": 698
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019971084644621216,
+      "loss": 1.552,
+      "step": 699
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001997079344976538,
+      "loss": 0.6481,
+      "step": 700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001997050079813783,
+      "loss": 0.5896,
+      "step": 701
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001997020668978133,
+      "loss": 0.6904,
+      "step": 702
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019969911124738849,
+      "loss": 1.1872,
+      "step": 703
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001996961410305357,
+      "loss": 1.3406,
+      "step": 704
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019969315624768886,
+      "loss": 1.0831,
+      "step": 705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019969015689928413,
+      "loss": 1.023,
+      "step": 706
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019968714298575964,
+      "loss": 1.5102,
+      "step": 707
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019968411450755578,
+      "loss": 1.2799,
+      "step": 708
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019968107146511502,
+      "loss": 0.9416,
+      "step": 709
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019967801385888193,
+      "loss": 1.3287,
+      "step": 710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019967494168930333,
+      "loss": 0.8482,
+      "step": 711
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019967185495682797,
+      "loss": 1.2345,
+      "step": 712
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019966875366190688,
+      "loss": 0.8175,
+      "step": 713
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019966563780499317,
+      "loss": 1.2715,
+      "step": 714
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019966250738654204,
+      "loss": 1.5505,
+      "step": 715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019965936240701093,
+      "loss": 0.8543,
+      "step": 716
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019965620286685928,
+      "loss": 0.964,
+      "step": 717
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019965302876654873,
+      "loss": 1.0369,
+      "step": 718
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019964984010654302,
+      "loss": 1.0362,
+      "step": 719
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019964663688730805,
+      "loss": 1.6037,
+      "step": 720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001996434191093118,
+      "loss": 1.3312,
+      "step": 721
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019964018677302438,
+      "loss": 1.4138,
+      "step": 722
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001996369398789181,
+      "loss": 1.2144,
+      "step": 723
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019963367842746729,
+      "loss": 1.0383,
+      "step": 724
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019963040241914847,
+      "loss": 1.1909,
+      "step": 725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001996271118544403,
+      "loss": 1.35,
+      "step": 726
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019962380673382355,
+      "loss": 1.2645,
+      "step": 727
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019962048705778109,
+      "loss": 1.2193,
+      "step": 728
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019961715282679795,
+      "loss": 1.1232,
+      "step": 729
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019961380404136128,
+      "loss": 0.7168,
+      "step": 730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019961044070196033,
+      "loss": 1.0963,
+      "step": 731
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001996070628090865,
+      "loss": 1.4121,
+      "step": 732
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019960367036323331,
+      "loss": 1.1601,
+      "step": 733
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019960026336489643,
+      "loss": 1.1132,
+      "step": 734
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019959684181457363,
+      "loss": 1.4075,
+      "step": 735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019959340571276481,
+      "loss": 0.7064,
+      "step": 736
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000199589955059972,
+      "loss": 1.4693,
+      "step": 737
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001995864898566993,
+      "loss": 0.7409,
+      "step": 738
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019958301010345307,
+      "loss": 1.4049,
+      "step": 739
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001995795158007417,
+      "loss": 1.3573,
+      "step": 740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001995760069490757,
+      "loss": 0.5351,
+      "step": 741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001995724835489677,
+      "loss": 1.6805,
+      "step": 742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019956894560093255,
+      "loss": 1.3843,
+      "step": 743
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019956539310548714,
+      "loss": 1.3898,
+      "step": 744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019956182606315043,
+      "loss": 0.9044,
+      "step": 745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019955824447444366,
+      "loss": 1.4886,
+      "step": 746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001995546483398901,
+      "loss": 1.1764,
+      "step": 747
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019955103766001512,
+      "loss": 1.5632,
+      "step": 748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001995474124353463,
+      "loss": 1.2752,
+      "step": 749
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001995437726664133,
+      "loss": 1.1628,
+      "step": 750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019954011835374784,
+      "loss": 1.3888,
+      "step": 751
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001995364494978839,
+      "loss": 1.1106,
+      "step": 752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001995327660993575,
+      "loss": 1.3539,
+      "step": 753
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019952906815870675,
+      "loss": 1.8475,
+      "step": 754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000199525355676472,
+      "loss": 1.217,
+      "step": 755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019952162865319567,
+      "loss": 1.1385,
+      "step": 756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019951788708942218,
+      "loss": 1.3506,
+      "step": 757
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019951413098569828,
+      "loss": 1.4123,
+      "step": 758
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019951036034257275,
+      "loss": 1.0629,
+      "step": 759
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019950657516059646,
+      "loss": 1.5456,
+      "step": 760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019950277544032247,
+      "loss": 0.6774,
+      "step": 761
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019949896118230591,
+      "loss": 0.8204,
+      "step": 762
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019949513238710407,
+      "loss": 1.3718,
+      "step": 763
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019949128905527634,
+      "loss": 0.6613,
+      "step": 764
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019948743118738426,
+      "loss": 0.9659,
+      "step": 765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019948355878399146,
+      "loss": 0.4228,
+      "step": 766
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019947967184566376,
+      "loss": 1.4363,
+      "step": 767
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000199475770372969,
+      "loss": 0.5529,
+      "step": 768
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019947185436647725,
+      "loss": 1.0721,
+      "step": 769
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019946792382676063,
+      "loss": 0.7222,
+      "step": 770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001994639787543934,
+      "loss": 1.2474,
+      "step": 771
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001994600191499519,
+      "loss": 1.4328,
+      "step": 772
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019945604501401477,
+      "loss": 0.8377,
+      "step": 773
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019945205634716256,
+      "loss": 1.3165,
+      "step": 774
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019944805314997806,
+      "loss": 0.5745,
+      "step": 775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001994440354230461,
+      "loss": 1.0332,
+      "step": 776
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019944000316695376,
+      "loss": 1.0928,
+      "step": 777
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019943595638229013,
+      "loss": 0.7563,
+      "step": 778
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019943189506964643,
+      "loss": 1.1064,
+      "step": 779
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001994278192296161,
+      "loss": 1.1568,
+      "step": 780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001994237288627946,
+      "loss": 1.4752,
+      "step": 781
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019941962396977957,
+      "loss": 1.2423,
+      "step": 782
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001994155045511707,
+      "loss": 1.2534,
+      "step": 783
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001994113706075699,
+      "loss": 1.2641,
+      "step": 784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001994072221395811,
+      "loss": 0.747,
+      "step": 785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019940305914781053,
+      "loss": 0.752,
+      "step": 786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019939888163286628,
+      "loss": 1.2313,
+      "step": 787
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019939468959535877,
+      "loss": 1.1557,
+      "step": 788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019939048303590048,
+      "loss": 1.4521,
+      "step": 789
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019938626195510595,
+      "loss": 1.1293,
+      "step": 790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019938202635359198,
+      "loss": 0.7123,
+      "step": 791
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019937777623197733,
+      "loss": 1.1657,
+      "step": 792
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019937351159088297,
+      "loss": 0.9721,
+      "step": 793
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019936923243093205,
+      "loss": 1.0481,
+      "step": 794
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001993649387527497,
+      "loss": 1.242,
+      "step": 795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019936063055696327,
+      "loss": 0.9812,
+      "step": 796
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001993563078442022,
+      "loss": 1.2121,
+      "step": 797
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019935197061509804,
+      "loss": 1.029,
+      "step": 798
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019934761887028453,
+      "loss": 1.1497,
+      "step": 799
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001993432526103974,
+      "loss": 1.0973,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019933887183607466,
+      "loss": 1.6185,
+      "step": 801
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019933447654795627,
+      "loss": 0.5706,
+      "step": 802
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019933006674668446,
+      "loss": 1.1319,
+      "step": 803
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019932564243290349,
+      "loss": 1.2289,
+      "step": 804
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019932120360725982,
+      "loss": 1.4509,
+      "step": 805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019931675027040188,
+      "loss": 0.604,
+      "step": 806
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019931228242298042,
+      "loss": 0.9386,
+      "step": 807
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019930780006564817,
+      "loss": 1.3508,
+      "step": 808
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019930330319906,
+      "loss": 1.2068,
+      "step": 809
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019929879182387298,
+      "loss": 1.1245,
+      "step": 810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019929426594074613,
+      "loss": 1.3342,
+      "step": 811
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019928972555034085,
+      "loss": 1.1462,
+      "step": 812
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001992851706533204,
+      "loss": 1.0522,
+      "step": 813
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019928060125035026,
+      "loss": 1.2946,
+      "step": 814
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019927601734209808,
+      "loss": 1.2182,
+      "step": 815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001992714189292336,
+      "loss": 1.501,
+      "step": 816
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019926680601242867,
+      "loss": 1.1707,
+      "step": 817
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001992621785923572,
+      "loss": 1.2454,
+      "step": 818
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019925753666969533,
+      "loss": 1.0884,
+      "step": 819
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001992528802451212,
+      "loss": 1.1923,
+      "step": 820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019924820931931518,
+      "loss": 0.8147,
+      "step": 821
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019924352389295974,
+      "loss": 1.4205,
+      "step": 822
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019923882396673937,
+      "loss": 0.7202,
+      "step": 823
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019923410954134077,
+      "loss": 1.7716,
+      "step": 824
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019922938061745274,
+      "loss": 1.487,
+      "step": 825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019922463719576622,
+      "loss": 1.3827,
+      "step": 826
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019921987927697422,
+      "loss": 1.2737,
+      "step": 827
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019921510686177186,
+      "loss": 0.6801,
+      "step": 828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019921031995085646,
+      "loss": 1.6034,
+      "step": 829
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019920551854492736,
+      "loss": 1.406,
+      "step": 830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001992007026446861,
+      "loss": 1.5904,
+      "step": 831
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001991958722508363,
+      "loss": 0.4908,
+      "step": 832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001991910273640837,
+      "loss": 0.9411,
+      "step": 833
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019918616798513615,
+      "loss": 0.7162,
+      "step": 834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019918129411470358,
+      "loss": 1.2877,
+      "step": 835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019917640575349818,
+      "loss": 1.2593,
+      "step": 836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019917150290223404,
+      "loss": 1.1239,
+      "step": 837
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019916658556162758,
+      "loss": 0.9784,
+      "step": 838
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001991616537323972,
+      "loss": 1.09,
+      "step": 839
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019915670741526347,
+      "loss": 1.543,
+      "step": 840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019915174661094907,
+      "loss": 1.4526,
+      "step": 841
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019914677132017877,
+      "loss": 0.4827,
+      "step": 842
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019914178154367953,
+      "loss": 1.1366,
+      "step": 843
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001991367772821803,
+      "loss": 0.4818,
+      "step": 844
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001991317585364123,
+      "loss": 1.2174,
+      "step": 845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019912672530710874,
+      "loss": 1.1033,
+      "step": 846
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019912167759500503,
+      "loss": 0.9764,
+      "step": 847
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001991166154008386,
+      "loss": 1.0532,
+      "step": 848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019911153872534908,
+      "loss": 0.5462,
+      "step": 849
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019910644756927824,
+      "loss": 0.9702,
+      "step": 850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001991013419333699,
+      "loss": 1.4001,
+      "step": 851
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019909622181836997,
+      "loss": 0.566,
+      "step": 852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019909108722502655,
+      "loss": 1.0787,
+      "step": 853
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001990859381540898,
+      "loss": 1.4842,
+      "step": 854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019908077460631206,
+      "loss": 1.1463,
+      "step": 855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019907559658244775,
+      "loss": 1.0358,
+      "step": 856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019907040408325335,
+      "loss": 0.6627,
+      "step": 857
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001990651971094875,
+      "loss": 0.6402,
+      "step": 858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000199059975661911,
+      "loss": 0.7986,
+      "step": 859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019905473974128675,
+      "loss": 1.1531,
+      "step": 860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019904948934837963,
+      "loss": 1.1991,
+      "step": 861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019904422448395684,
+      "loss": 0.5682,
+      "step": 862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001990389451487876,
+      "loss": 1.1228,
+      "step": 863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019903365134364318,
+      "loss": 1.2793,
+      "step": 864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019902834306929706,
+      "loss": 1.345,
+      "step": 865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001990230203265248,
+      "loss": 0.9587,
+      "step": 866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019901768311610406,
+      "loss": 1.5756,
+      "step": 867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019901233143881464,
+      "loss": 1.4265,
+      "step": 868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019900696529543844,
+      "loss": 1.391,
+      "step": 869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019900158468675946,
+      "loss": 1.4773,
+      "step": 870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019899618961356387,
+      "loss": 1.1361,
+      "step": 871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019899078007663983,
+      "loss": 1.0601,
+      "step": 872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019898535607677778,
+      "loss": 1.3601,
+      "step": 873
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019897991761477015,
+      "loss": 1.5741,
+      "step": 874
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001989744646914115,
+      "loss": 1.0944,
+      "step": 875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019896899730749857,
+      "loss": 1.6258,
+      "step": 876
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019896351546383012,
+      "loss": 1.2056,
+      "step": 877
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001989580191612071,
+      "loss": 0.5746,
+      "step": 878
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019895250840043253,
+      "loss": 1.3877,
+      "step": 879
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019894698318231155,
+      "loss": 1.3506,
+      "step": 880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019894144350765145,
+      "loss": 1.0178,
+      "step": 881
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019893588937726153,
+      "loss": 0.9069,
+      "step": 882
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019893032079195333,
+      "loss": 1.1214,
+      "step": 883
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001989247377525404,
+      "loss": 0.6316,
+      "step": 884
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019891914025983848,
+      "loss": 0.7209,
+      "step": 885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001989135283146654,
+      "loss": 1.1491,
+      "step": 886
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019890790191784102,
+      "loss": 0.8493,
+      "step": 887
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001989022610701874,
+      "loss": 0.7945,
+      "step": 888
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019889660577252874,
+      "loss": 0.7981,
+      "step": 889
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019889093602569124,
+      "loss": 1.2184,
+      "step": 890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019888525183050333,
+      "loss": 1.2862,
+      "step": 891
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019887955318779543,
+      "loss": 0.8633,
+      "step": 892
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001988738400984002,
+      "loss": 1.4668,
+      "step": 893
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001988681125631523,
+      "loss": 0.8324,
+      "step": 894
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019886237058288852,
+      "loss": 1.3861,
+      "step": 895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019885661415844786,
+      "loss": 1.1208,
+      "step": 896
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001988508432906713,
+      "loss": 1.3656,
+      "step": 897
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000198845057980402,
+      "loss": 1.4574,
+      "step": 898
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019883925822848522,
+      "loss": 1.2789,
+      "step": 899
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019883344403576834,
+      "loss": 1.6562,
+      "step": 900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001988276154031008,
+      "loss": 0.6467,
+      "step": 901
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019882177233133423,
+      "loss": 1.336,
+      "step": 902
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001988159148213223,
+      "loss": 0.8387,
+      "step": 903
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001988100428739208,
+      "loss": 1.1773,
+      "step": 904
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019880415648998768,
+      "loss": 1.4102,
+      "step": 905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019879825567038295,
+      "loss": 1.2445,
+      "step": 906
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019879234041596873,
+      "loss": 1.5335,
+      "step": 907
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019878641072760926,
+      "loss": 1.29,
+      "step": 908
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001987804666061709,
+      "loss": 0.8062,
+      "step": 909
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019877450805252212,
+      "loss": 0.9254,
+      "step": 910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001987685350675335,
+      "loss": 1.398,
+      "step": 911
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019876254765207766,
+      "loss": 0.4382,
+      "step": 912
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019875654580702942,
+      "loss": 0.8849,
+      "step": 913
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001987505295332657,
+      "loss": 0.736,
+      "step": 914
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019874449883166545,
+      "loss": 0.5768,
+      "step": 915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001987384537031098,
+      "loss": 1.2655,
+      "step": 916
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000198732394148482,
+      "loss": 1.32,
+      "step": 917
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019872632016866733,
+      "loss": 1.4465,
+      "step": 918
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001987202317645532,
+      "loss": 1.1464,
+      "step": 919
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019871412893702922,
+      "loss": 1.4021,
+      "step": 920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000198708011686987,
+      "loss": 1.5264,
+      "step": 921
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019870188001532032,
+      "loss": 1.3587,
+      "step": 922
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019869573392292496,
+      "loss": 1.0626,
+      "step": 923
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000198689573410699,
+      "loss": 1.177,
+      "step": 924
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019868339847954242,
+      "loss": 1.1559,
+      "step": 925
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019867720913035745,
+      "loss": 0.6915,
+      "step": 926
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019867100536404838,
+      "loss": 1.0819,
+      "step": 927
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019866478718152158,
+      "loss": 0.9889,
+      "step": 928
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019865855458368556,
+      "loss": 1.278,
+      "step": 929
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019865230757145096,
+      "loss": 0.8351,
+      "step": 930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019864604614573043,
+      "loss": 0.8004,
+      "step": 931
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019863977030743884,
+      "loss": 1.181,
+      "step": 932
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019863348005749311,
+      "loss": 0.526,
+      "step": 933
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019862717539681225,
+      "loss": 0.9841,
+      "step": 934
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019862085632631738,
+      "loss": 1.2543,
+      "step": 935
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019861452284693177,
+      "loss": 1.0429,
+      "step": 936
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019860817495958078,
+      "loss": 0.6793,
+      "step": 937
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019860181266519183,
+      "loss": 0.7583,
+      "step": 938
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019859543596469447,
+      "loss": 1.215,
+      "step": 939
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001985890448590204,
+      "loss": 1.27,
+      "step": 940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019858263934910335,
+      "loss": 0.9275,
+      "step": 941
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001985762194358792,
+      "loss": 1.2305,
+      "step": 942
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019856978512028595,
+      "loss": 1.5363,
+      "step": 943
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001985633364032636,
+      "loss": 0.6898,
+      "step": 944
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019855687328575443,
+      "loss": 1.3799,
+      "step": 945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001985503957687027,
+      "loss": 0.6133,
+      "step": 946
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019854390385305477,
+      "loss": 1.3723,
+      "step": 947
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019853739753975913,
+      "loss": 0.7177,
+      "step": 948
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001985308768297664,
+      "loss": 1.4005,
+      "step": 949
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019852434172402928,
+      "loss": 1.4018,
+      "step": 950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001985177922235026,
+      "loss": 0.7915,
+      "step": 951
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019851122832914323,
+      "loss": 1.2336,
+      "step": 952
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019850465004191015,
+      "loss": 1.2157,
+      "step": 953
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019849805736276452,
+      "loss": 0.7794,
+      "step": 954
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019849145029266957,
+      "loss": 0.8346,
+      "step": 955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019848482883259063,
+      "loss": 1.1283,
+      "step": 956
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019847819298349505,
+      "loss": 1.3914,
+      "step": 957
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019847154274635237,
+      "loss": 0.7804,
+      "step": 958
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019846487812213428,
+      "loss": 1.2878,
+      "step": 959
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019845819911181446,
+      "loss": 1.5854,
+      "step": 960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019845150571636873,
+      "loss": 1.2694,
+      "step": 961
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019844479793677504,
+      "loss": 1.3892,
+      "step": 962
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019843807577401342,
+      "loss": 0.7534,
+      "step": 963
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019843133922906603,
+      "loss": 0.5834,
+      "step": 964
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019842458830291704,
+      "loss": 1.353,
+      "step": 965
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019841782299655285,
+      "loss": 1.1471,
+      "step": 966
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019841104331096184,
+      "loss": 1.5828,
+      "step": 967
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001984042492471346,
+      "loss": 0.7602,
+      "step": 968
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019839744080606377,
+      "loss": 1.2199,
+      "step": 969
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019839061798874403,
+      "loss": 1.0539,
+      "step": 970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001983837807961723,
+      "loss": 1.3977,
+      "step": 971
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019837692922934744,
+      "loss": 1.0113,
+      "step": 972
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019837006328927057,
+      "loss": 1.0516,
+      "step": 973
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019836318297694478,
+      "loss": 1.0194,
+      "step": 974
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001983562882933753,
+      "loss": 0.6957,
+      "step": 975
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001983493792395695,
+      "loss": 0.9776,
+      "step": 976
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019834245581653684,
+      "loss": 0.8377,
+      "step": 977
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001983355180252888,
+      "loss": 1.189,
+      "step": 978
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019832856586683907,
+      "loss": 1.3978,
+      "step": 979
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019832159934220336,
+      "loss": 1.3436,
+      "step": 980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019831461845239948,
+      "loss": 0.9802,
+      "step": 981
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019830762319844743,
+      "loss": 1.4291,
+      "step": 982
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001983006135813692,
+      "loss": 1.1611,
+      "step": 983
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019829358960218895,
+      "loss": 0.6537,
+      "step": 984
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019828655126193288,
+      "loss": 0.8799,
+      "step": 985
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019827949856162934,
+      "loss": 0.9494,
+      "step": 986
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019827243150230875,
+      "loss": 1.4072,
+      "step": 987
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019826535008500364,
+      "loss": 1.3972,
+      "step": 988
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001982582543107486,
+      "loss": 0.4759,
+      "step": 989
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001982511441805804,
+      "loss": 1.2822,
+      "step": 990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019824401969553785,
+      "loss": 1.1049,
+      "step": 991
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019823688085666183,
+      "loss": 1.6319,
+      "step": 992
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001982297276649954,
+      "loss": 0.6126,
+      "step": 993
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001982225601215836,
+      "loss": 1.0581,
+      "step": 994
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001982153782274737,
+      "loss": 1.5482,
+      "step": 995
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019820818198371497,
+      "loss": 0.8458,
+      "step": 996
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019820097139135884,
+      "loss": 0.8743,
+      "step": 997
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001981937464514588,
+      "loss": 0.9249,
+      "step": 998
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001981865071650704,
+      "loss": 1.0824,
+      "step": 999
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019817925353325135,
+      "loss": 1.482,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019817198555706144,
+      "loss": 1.4536,
+      "step": 1001
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019816470323756255,
+      "loss": 1.1578,
+      "step": 1002
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019815740657581866,
+      "loss": 0.6612,
+      "step": 1003
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019815009557289583,
+      "loss": 1.4154,
+      "step": 1004
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019814277022986226,
+      "loss": 1.3751,
+      "step": 1005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019813543054778815,
+      "loss": 0.8369,
+      "step": 1006
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001981280765277459,
+      "loss": 1.0492,
+      "step": 1007
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019812070817080995,
+      "loss": 0.9492,
+      "step": 1008
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001981133254780568,
+      "loss": 0.6357,
+      "step": 1009
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001981059284505652,
+      "loss": 1.1641,
+      "step": 1010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001980985170894158,
+      "loss": 1.4519,
+      "step": 1011
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019809109139569143,
+      "loss": 1.2072,
+      "step": 1012
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019808365137047705,
+      "loss": 1.1373,
+      "step": 1013
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019807619701485963,
+      "loss": 0.9016,
+      "step": 1014
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001980687283299283,
+      "loss": 1.5622,
+      "step": 1015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001980612453167743,
+      "loss": 0.8753,
+      "step": 1016
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001980537479764909,
+      "loss": 1.2502,
+      "step": 1017
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019804623631017348,
+      "loss": 0.7121,
+      "step": 1018
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019803871031891953,
+      "loss": 1.091,
+      "step": 1019
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019803117000382866,
+      "loss": 1.0731,
+      "step": 1020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019802361536600246,
+      "loss": 0.8807,
+      "step": 1021
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019801604640654475,
+      "loss": 1.591,
+      "step": 1022
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019800846312656138,
+      "loss": 0.929,
+      "step": 1023
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001980008655271603,
+      "loss": 1.2149,
+      "step": 1024
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019799325360945157,
+      "loss": 1.3427,
+      "step": 1025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019798562737454725,
+      "loss": 1.4774,
+      "step": 1026
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019797798682356162,
+      "loss": 0.5396,
+      "step": 1027
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019797033195761095,
+      "loss": 1.1385,
+      "step": 1028
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001979626627778137,
+      "loss": 1.4088,
+      "step": 1029
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019795497928529033,
+      "loss": 1.7236,
+      "step": 1030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019794728148116346,
+      "loss": 0.7714,
+      "step": 1031
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019793956936655773,
+      "loss": 1.3871,
+      "step": 1032
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019793184294259993,
+      "loss": 0.8727,
+      "step": 1033
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019792410221041893,
+      "loss": 1.4527,
+      "step": 1034
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001979163471711457,
+      "loss": 1.1509,
+      "step": 1035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019790857782591318,
+      "loss": 1.28,
+      "step": 1036
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001979007941758566,
+      "loss": 1.2164,
+      "step": 1037
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019789299622211318,
+      "loss": 0.2099,
+      "step": 1038
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001978851839658222,
+      "loss": 0.8003,
+      "step": 1039
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019787735740812504,
+      "loss": 1.4483,
+      "step": 1040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001978695165501652,
+      "loss": 0.6609,
+      "step": 1041
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019786166139308832,
+      "loss": 0.5822,
+      "step": 1042
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019785379193804203,
+      "loss": 0.8876,
+      "step": 1043
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019784590818617604,
+      "loss": 1.6651,
+      "step": 1044
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001978380101386423,
+      "loss": 0.5261,
+      "step": 1045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019783009779659466,
+      "loss": 1.4352,
+      "step": 1046
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019782217116118918,
+      "loss": 1.2435,
+      "step": 1047
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019781423023358392,
+      "loss": 1.0309,
+      "step": 1048
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019780627501493915,
+      "loss": 0.746,
+      "step": 1049
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019779830550641715,
+      "loss": 1.4159,
+      "step": 1050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019779032170918226,
+      "loss": 1.0304,
+      "step": 1051
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019778232362440096,
+      "loss": 0.9625,
+      "step": 1052
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001977743112532418,
+      "loss": 0.5888,
+      "step": 1053
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019776628459687543,
+      "loss": 1.1813,
+      "step": 1054
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019775824365647458,
+      "loss": 1.2893,
+      "step": 1055
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019775018843321403,
+      "loss": 0.5683,
+      "step": 1056
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001977421189282707,
+      "loss": 1.0242,
+      "step": 1057
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019773403514282355,
+      "loss": 0.8917,
+      "step": 1058
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001977259370780537,
+      "loss": 1.1228,
+      "step": 1059
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001977178247351443,
+      "loss": 1.1034,
+      "step": 1060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019770969811528058,
+      "loss": 1.8004,
+      "step": 1061
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019770155721964987,
+      "loss": 0.7164,
+      "step": 1062
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001976934020494416,
+      "loss": 1.112,
+      "step": 1063
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019768523260584722,
+      "loss": 0.5128,
+      "step": 1064
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019767704889006037,
+      "loss": 1.1074,
+      "step": 1065
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019766885090327673,
+      "loss": 0.8808,
+      "step": 1066
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019766063864669407,
+      "loss": 0.8266,
+      "step": 1067
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019765241212151217,
+      "loss": 1.2272,
+      "step": 1068
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019764417132893297,
+      "loss": 1.5483,
+      "step": 1069
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001976359162701605,
+      "loss": 1.4803,
+      "step": 1070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019762764694640088,
+      "loss": 1.3848,
+      "step": 1071
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019761936335886224,
+      "loss": 1.1559,
+      "step": 1072
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001976110655087549,
+      "loss": 0.8908,
+      "step": 1073
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019760275339729118,
+      "loss": 1.1587,
+      "step": 1074
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001975944270256855,
+      "loss": 1.5358,
+      "step": 1075
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019758608639515438,
+      "loss": 1.4201,
+      "step": 1076
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019757773150691642,
+      "loss": 1.1756,
+      "step": 1077
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019756936236219231,
+      "loss": 0.9915,
+      "step": 1078
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019756097896220484,
+      "loss": 1.3991,
+      "step": 1079
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001975525813081788,
+      "loss": 1.1826,
+      "step": 1080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019754416940134117,
+      "loss": 1.3684,
+      "step": 1081
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019753574324292092,
+      "loss": 1.4813,
+      "step": 1082
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001975273028341492,
+      "loss": 1.3165,
+      "step": 1083
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019751884817625911,
+      "loss": 1.2819,
+      "step": 1084
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000197510379270486,
+      "loss": 1.3424,
+      "step": 1085
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001975018961180671,
+      "loss": 0.7846,
+      "step": 1086
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019749339872024194,
+      "loss": 1.6303,
+      "step": 1087
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019748488707825194,
+      "loss": 0.9417,
+      "step": 1088
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019747636119334073,
+      "loss": 1.3763,
+      "step": 1089
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000197467821066754,
+      "loss": 1.0284,
+      "step": 1090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019745926669973942,
+      "loss": 1.2389,
+      "step": 1091
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019745069809354687,
+      "loss": 1.2941,
+      "step": 1092
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001974421152494282,
+      "loss": 1.2499,
+      "step": 1093
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001974335181686375,
+      "loss": 0.707,
+      "step": 1094
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019742490685243077,
+      "loss": 0.9711,
+      "step": 1095
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019741628130206612,
+      "loss": 1.9337,
+      "step": 1096
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019740764151880387,
+      "loss": 0.7866,
+      "step": 1097
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019739898750390625,
+      "loss": 1.4794,
+      "step": 1098
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019739031925863768,
+      "loss": 1.1223,
+      "step": 1099
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019738163678426461,
+      "loss": 0.7846,
+      "step": 1100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001973729400820556,
+      "loss": 0.6572,
+      "step": 1101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019736422915328126,
+      "loss": 0.5568,
+      "step": 1102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019735550399921427,
+      "loss": 0.8249,
+      "step": 1103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019734676462112945,
+      "loss": 1.4348,
+      "step": 1104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019733801102030362,
+      "loss": 1.4897,
+      "step": 1105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019732924319801576,
+      "loss": 0.6706,
+      "step": 1106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019732046115554684,
+      "loss": 0.7254,
+      "step": 1107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019731166489417995,
+      "loss": 1.6912,
+      "step": 1108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001973028544152003,
+      "loss": 1.281,
+      "step": 1109
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019729402971989507,
+      "loss": 0.7136,
+      "step": 1110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019728519080955368,
+      "loss": 0.7283,
+      "step": 1111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019727633768546742,
+      "loss": 1.2406,
+      "step": 1112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019726747034892986,
+      "loss": 1.6447,
+      "step": 1113
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001972585888012365,
+      "loss": 0.9839,
+      "step": 1114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019724969304368495,
+      "loss": 1.5637,
+      "step": 1115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019724078307757495,
+      "loss": 1.2544,
+      "step": 1116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019723185890420828,
+      "loss": 0.8163,
+      "step": 1117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001972229205248888,
+      "loss": 1.1191,
+      "step": 1118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001972139679409224,
+      "loss": 1.1149,
+      "step": 1119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019720500115361715,
+      "loss": 1.1553,
+      "step": 1120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019719602016428313,
+      "loss": 0.5889,
+      "step": 1121
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001971870249742324,
+      "loss": 1.2329,
+      "step": 1122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001971780155847793,
+      "loss": 1.5173,
+      "step": 1123
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001971689919972401,
+      "loss": 1.2405,
+      "step": 1124
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019715995421293317,
+      "loss": 1.2669,
+      "step": 1125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019715090223317895,
+      "loss": 1.5222,
+      "step": 1126
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019714183605930004,
+      "loss": 1.6265,
+      "step": 1127
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019713275569262098,
+      "loss": 1.3809,
+      "step": 1128
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019712366113446846,
+      "loss": 1.2418,
+      "step": 1129
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019711455238617122,
+      "loss": 0.1797,
+      "step": 1130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019710542944906015,
+      "loss": 0.5275,
+      "step": 1131
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019709629232446806,
+      "loss": 1.3054,
+      "step": 1132
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019708714101372994,
+      "loss": 1.1219,
+      "step": 1133
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019707797551818285,
+      "loss": 0.7903,
+      "step": 1134
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019706879583916592,
+      "loss": 1.1857,
+      "step": 1135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019705960197802033,
+      "loss": 0.4305,
+      "step": 1136
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019705039393608928,
+      "loss": 1.1755,
+      "step": 1137
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019704117171471817,
+      "loss": 0.8619,
+      "step": 1138
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019703193531525438,
+      "loss": 1.4339,
+      "step": 1139
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001970226847390474,
+      "loss": 1.0046,
+      "step": 1140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019701341998744874,
+      "loss": 0.9353,
+      "step": 1141
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019700414106181205,
+      "loss": 0.6948,
+      "step": 1142
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000196994847963493,
+      "loss": 1.3245,
+      "step": 1143
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019698554069384934,
+      "loss": 1.2269,
+      "step": 1144
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001969762192542409,
+      "loss": 1.3095,
+      "step": 1145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001969668836460296,
+      "loss": 0.9519,
+      "step": 1146
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001969575338705794,
+      "loss": 0.5266,
+      "step": 1147
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019694816992925636,
+      "loss": 1.0988,
+      "step": 1148
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001969387918234285,
+      "loss": 1.4987,
+      "step": 1149
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001969293995544661,
+      "loss": 0.6357,
+      "step": 1150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019691999312374136,
+      "loss": 1.7033,
+      "step": 1151
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001969105725326286,
+      "loss": 1.1974,
+      "step": 1152
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019690113778250421,
+      "loss": 0.926,
+      "step": 1153
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019689168887474664,
+      "loss": 1.3542,
+      "step": 1154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019688222581073643,
+      "loss": 1.3587,
+      "step": 1155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019687274859185613,
+      "loss": 0.7466,
+      "step": 1156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019686325721949045,
+      "loss": 0.9141,
+      "step": 1157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019685375169502608,
+      "loss": 0.6188,
+      "step": 1158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001968442320198518,
+      "loss": 1.1144,
+      "step": 1159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019683469819535852,
+      "loss": 0.6362,
+      "step": 1160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019682515022293915,
+      "loss": 1.4258,
+      "step": 1161
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001968155881039887,
+      "loss": 1.2953,
+      "step": 1162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019680601183990414,
+      "loss": 0.816,
+      "step": 1163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001967964214320847,
+      "loss": 0.6627,
+      "step": 1164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001967868168819316,
+      "loss": 1.6128,
+      "step": 1165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000196777198190848,
+      "loss": 1.3119,
+      "step": 1166
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001967675653602393,
+      "loss": 0.6373,
+      "step": 1167
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019675791839151286,
+      "loss": 0.9154,
+      "step": 1168
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019674825728607817,
+      "loss": 1.3072,
+      "step": 1169
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019673858204534672,
+      "loss": 1.2848,
+      "step": 1170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001967288926707321,
+      "loss": 1.3656,
+      "step": 1171
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019671918916365,
+      "loss": 1.3751,
+      "step": 1172
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019670947152551816,
+      "loss": 1.2732,
+      "step": 1173
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001966997397577563,
+      "loss": 0.8023,
+      "step": 1174
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019668999386178627,
+      "loss": 1.2947,
+      "step": 1175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019668023383903203,
+      "loss": 1.127,
+      "step": 1176
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019667045969091955,
+      "loss": 1.3838,
+      "step": 1177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019666067141887685,
+      "loss": 1.2381,
+      "step": 1178
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019665086902433406,
+      "loss": 1.0671,
+      "step": 1179
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001966410525087233,
+      "loss": 1.2467,
+      "step": 1180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019663122187347885,
+      "loss": 0.6468,
+      "step": 1181
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019662137712003695,
+      "loss": 1.3458,
+      "step": 1182
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019661151824983605,
+      "loss": 1.1794,
+      "step": 1183
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001966016452643165,
+      "loss": 1.2872,
+      "step": 1184
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019659175816492077,
+      "loss": 1.1091,
+      "step": 1185
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019658185695309345,
+      "loss": 1.1532,
+      "step": 1186
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019657194163028108,
+      "loss": 0.5389,
+      "step": 1187
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019656201219793245,
+      "loss": 1.1177,
+      "step": 1188
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019655206865749816,
+      "loss": 0.8917,
+      "step": 1189
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019654211101043103,
+      "loss": 1.7171,
+      "step": 1190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019653213925818595,
+      "loss": 1.4443,
+      "step": 1191
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019652215340221983,
+      "loss": 1.5236,
+      "step": 1192
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001965121534439916,
+      "loss": 1.203,
+      "step": 1193
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019650213938496232,
+      "loss": 1.2802,
+      "step": 1194
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001964921112265951,
+      "loss": 1.373,
+      "step": 1195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019648206897035503,
+      "loss": 1.402,
+      "step": 1196
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019647201261770942,
+      "loss": 1.3511,
+      "step": 1197
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019646194217012747,
+      "loss": 1.0212,
+      "step": 1198
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001964518576290805,
+      "loss": 0.7947,
+      "step": 1199
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019644175899604199,
+      "loss": 0.9526,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019643164627248727,
+      "loss": 0.3822,
+      "step": 1201
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019642151945989395,
+      "loss": 1.1188,
+      "step": 1202
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001964113785597415,
+      "loss": 1.3211,
+      "step": 1203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019640122357351164,
+      "loss": 0.9283,
+      "step": 1204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019639105450268803,
+      "loss": 0.7478,
+      "step": 1205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019638087134875635,
+      "loss": 1.2735,
+      "step": 1206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019637067411320449,
+      "loss": 1.1879,
+      "step": 1207
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019636046279752223,
+      "loss": 1.3358,
+      "step": 1208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019635023740320148,
+      "loss": 0.5663,
+      "step": 1209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019633999793173626,
+      "loss": 1.0211,
+      "step": 1210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019632974438462262,
+      "loss": 0.6365,
+      "step": 1211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019631947676335858,
+      "loss": 0.8579,
+      "step": 1212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019630919506944427,
+      "loss": 1.2658,
+      "step": 1213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019629889930438194,
+      "loss": 1.2975,
+      "step": 1214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019628858946967582,
+      "loss": 0.8279,
+      "step": 1215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001962782655668322,
+      "loss": 0.6709,
+      "step": 1216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019626792759735946,
+      "loss": 0.6422,
+      "step": 1217
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019625757556276802,
+      "loss": 1.087,
+      "step": 1218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019624720946457034,
+      "loss": 0.7848,
+      "step": 1219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019623682930428092,
+      "loss": 1.1132,
+      "step": 1220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019622643508341643,
+      "loss": 1.175,
+      "step": 1221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001962160268034954,
+      "loss": 0.744,
+      "step": 1222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001962056044660386,
+      "loss": 1.0082,
+      "step": 1223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001961951680725688,
+      "loss": 1.4729,
+      "step": 1224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019618471762461069,
+      "loss": 1.0986,
+      "step": 1225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019617425312369117,
+      "loss": 1.0134,
+      "step": 1226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019616377457133913,
+      "loss": 1.3048,
+      "step": 1227
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019615328196908558,
+      "loss": 1.1995,
+      "step": 1228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001961427753184635,
+      "loss": 1.7458,
+      "step": 1229
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001961322546210079,
+      "loss": 0.7805,
+      "step": 1230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000196121719878256,
+      "loss": 1.1358,
+      "step": 1231
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019611117109174684,
+      "loss": 1.256,
+      "step": 1232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019610060826302176,
+      "loss": 0.6706,
+      "step": 1233
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019609003139362398,
+      "loss": 0.809,
+      "step": 1234
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001960794404850988,
+      "loss": 0.9478,
+      "step": 1235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019606883553899363,
+      "loss": 1.0258,
+      "step": 1236
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019605821655685788,
+      "loss": 1.2648,
+      "step": 1237
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000196047583540243,
+      "loss": 1.729,
+      "step": 1238
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019603693649070255,
+      "loss": 1.0613,
+      "step": 1239
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001960262754097921,
+      "loss": 1.0157,
+      "step": 1240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019601560029906926,
+      "loss": 1.2506,
+      "step": 1241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019600491116009372,
+      "loss": 0.8876,
+      "step": 1242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001959942079944272,
+      "loss": 1.3644,
+      "step": 1243
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019598349080363346,
+      "loss": 1.203,
+      "step": 1244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019597275958927834,
+      "loss": 0.8058,
+      "step": 1245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019596201435292973,
+      "loss": 1.1298,
+      "step": 1246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019595125509615748,
+      "loss": 1.6914,
+      "step": 1247
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019594048182053366,
+      "loss": 1.0927,
+      "step": 1248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001959296945276322,
+      "loss": 1.1566,
+      "step": 1249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001959188932190292,
+      "loss": 1.3983,
+      "step": 1250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001959080778963028,
+      "loss": 0.9287,
+      "step": 1251
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019589724856103314,
+      "loss": 1.1454,
+      "step": 1252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001958864052148024,
+      "loss": 1.1332,
+      "step": 1253
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019587554785919484,
+      "loss": 0.9199,
+      "step": 1254
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001958646764957968,
+      "loss": 0.9997,
+      "step": 1255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019585379112619663,
+      "loss": 0.8649,
+      "step": 1256
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019584289175198465,
+      "loss": 1.1783,
+      "step": 1257
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001958319783747534,
+      "loss": 1.4246,
+      "step": 1258
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019582105099609729,
+      "loss": 1.2224,
+      "step": 1259
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001958101096176129,
+      "loss": 0.9996,
+      "step": 1260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019579915424089878,
+      "loss": 0.8875,
+      "step": 1261
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019578818486755558,
+      "loss": 0.7989,
+      "step": 1262
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019577720149918594,
+      "loss": 1.1209,
+      "step": 1263
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001957662041373946,
+      "loss": 0.6029,
+      "step": 1264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001957551927837883,
+      "loss": 1.3323,
+      "step": 1265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001957441674399758,
+      "loss": 0.7727,
+      "step": 1266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019573312810756806,
+      "loss": 1.2078,
+      "step": 1267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019572207478817786,
+      "loss": 1.343,
+      "step": 1268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019571100748342018,
+      "loss": 1.396,
+      "step": 1269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019569992619491197,
+      "loss": 1.306,
+      "step": 1270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019568883092427227,
+      "loss": 1.4422,
+      "step": 1271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019567772167312217,
+      "loss": 1.1,
+      "step": 1272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001956665984430847,
+      "loss": 1.1034,
+      "step": 1273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019565546123578514,
+      "loss": 1.5699,
+      "step": 1274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001956443100528505,
+      "loss": 1.1817,
+      "step": 1275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019563314489591015,
+      "loss": 1.5065,
+      "step": 1276
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001956219657665953,
+      "loss": 0.6792,
+      "step": 1277
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019561077266653926,
+      "loss": 0.4608,
+      "step": 1278
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019559956559737744,
+      "loss": 1.6218,
+      "step": 1279
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001955883445607472,
+      "loss": 1.3092,
+      "step": 1280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019557710955828795,
+      "loss": 1.5239,
+      "step": 1281
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019556586059164122,
+      "loss": 1.2577,
+      "step": 1282
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019555459766245047,
+      "loss": 0.7771,
+      "step": 1283
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019554332077236132,
+      "loss": 1.0628,
+      "step": 1284
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019553202992302136,
+      "loss": 1.4483,
+      "step": 1285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019552072511608016,
+      "loss": 0.7611,
+      "step": 1286
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019550940635318947,
+      "loss": 0.6107,
+      "step": 1287
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019549807363600298,
+      "loss": 1.3293,
+      "step": 1288
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019548672696617644,
+      "loss": 0.6138,
+      "step": 1289
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019547536634536764,
+      "loss": 1.0605,
+      "step": 1290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019546399177523642,
+      "loss": 1.339,
+      "step": 1291
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019545260325744463,
+      "loss": 1.3262,
+      "step": 1292
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019544120079365623,
+      "loss": 1.4937,
+      "step": 1293
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019542978438553707,
+      "loss": 1.247,
+      "step": 1294
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019541835403475523,
+      "loss": 0.6945,
+      "step": 1295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019540690974298065,
+      "loss": 1.1769,
+      "step": 1296
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019539545151188542,
+      "loss": 0.9602,
+      "step": 1297
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019538397934314367,
+      "loss": 0.7321,
+      "step": 1298
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001953724932384315,
+      "loss": 1.3143,
+      "step": 1299
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019536099319942703,
+      "loss": 1.3611,
+      "step": 1300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001953494792278105,
+      "loss": 0.9514,
+      "step": 1301
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019533795132526416,
+      "loss": 1.4122,
+      "step": 1302
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019532640949347226,
+      "loss": 1.1751,
+      "step": 1303
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019531485373412108,
+      "loss": 1.104,
+      "step": 1304
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019530328404889904,
+      "loss": 1.324,
+      "step": 1305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019529170043949644,
+      "loss": 1.118,
+      "step": 1306
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001952801029076057,
+      "loss": 1.5083,
+      "step": 1307
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019526849145492134,
+      "loss": 1.3014,
+      "step": 1308
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019525686608313976,
+      "loss": 0.714,
+      "step": 1309
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019524522679395946,
+      "loss": 1.1027,
+      "step": 1310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019523357358908105,
+      "loss": 0.7682,
+      "step": 1311
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001952219064702071,
+      "loss": 1.1951,
+      "step": 1312
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019521022543904216,
+      "loss": 0.6936,
+      "step": 1313
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001951985304972929,
+      "loss": 1.0584,
+      "step": 1314
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019518682164666805,
+      "loss": 0.4939,
+      "step": 1315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019517509888887825,
+      "loss": 1.4001,
+      "step": 1316
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019516336222563627,
+      "loss": 0.6662,
+      "step": 1317
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019515161165865693,
+      "loss": 0.6048,
+      "step": 1318
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019513984718965693,
+      "loss": 0.6576,
+      "step": 1319
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019512806882035518,
+      "loss": 1.4596,
+      "step": 1320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001951162765524725,
+      "loss": 0.6728,
+      "step": 1321
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019510447038773184,
+      "loss": 1.1774,
+      "step": 1322
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001950926503278581,
+      "loss": 1.1466,
+      "step": 1323
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001950808163745782,
+      "loss": 0.9851,
+      "step": 1324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001950689685296212,
+      "loss": 1.4683,
+      "step": 1325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019505710679471806,
+      "loss": 0.556,
+      "step": 1326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019504523117160183,
+      "loss": 1.4975,
+      "step": 1327
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001950333416620076,
+      "loss": 1.1622,
+      "step": 1328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001950214382676725,
+      "loss": 1.419,
+      "step": 1329
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001950095209903356,
+      "loss": 0.9361,
+      "step": 1330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019499758983173812,
+      "loss": 0.6171,
+      "step": 1331
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019498564479362323,
+      "loss": 0.4977,
+      "step": 1332
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001949736858777361,
+      "loss": 1.3414,
+      "step": 1333
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019496171308582404,
+      "loss": 0.3962,
+      "step": 1334
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019494972641963628,
+      "loss": 1.0656,
+      "step": 1335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019493772588092414,
+      "loss": 1.2016,
+      "step": 1336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019492571147144093,
+      "loss": 1.1137,
+      "step": 1337
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019491368319294203,
+      "loss": 1.1063,
+      "step": 1338
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001949016410471848,
+      "loss": 0.838,
+      "step": 1339
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019488958503592865,
+      "loss": 1.8067,
+      "step": 1340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019487751516093495,
+      "loss": 0.8546,
+      "step": 1341
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019486543142396726,
+      "loss": 1.3033,
+      "step": 1342
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000194853333826791,
+      "loss": 1.5622,
+      "step": 1343
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019484122237117368,
+      "loss": 1.1018,
+      "step": 1344
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019482909705888484,
+      "loss": 1.0279,
+      "step": 1345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019481695789169603,
+      "loss": 0.8284,
+      "step": 1346
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001948048048713808,
+      "loss": 0.8259,
+      "step": 1347
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019479263799971483,
+      "loss": 0.6214,
+      "step": 1348
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019478045727847572,
+      "loss": 1.4846,
+      "step": 1349
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019476826270944307,
+      "loss": 1.1527,
+      "step": 1350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019475605429439857,
+      "loss": 1.0823,
+      "step": 1351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019474383203512598,
+      "loss": 0.5776,
+      "step": 1352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019473159593341092,
+      "loss": 1.0343,
+      "step": 1353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019471934599104125,
+      "loss": 1.3473,
+      "step": 1354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019470708220980663,
+      "loss": 1.0632,
+      "step": 1355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019469480459149887,
+      "loss": 1.1563,
+      "step": 1356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019468251313791183,
+      "loss": 1.1895,
+      "step": 1357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019467020785084128,
+      "loss": 1.1999,
+      "step": 1358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001946578887320851,
+      "loss": 1.0293,
+      "step": 1359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019464555578344316,
+      "loss": 1.2608,
+      "step": 1360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019463320900671733,
+      "loss": 1.1272,
+      "step": 1361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019462084840371157,
+      "loss": 1.0277,
+      "step": 1362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019460847397623178,
+      "loss": 1.1385,
+      "step": 1363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019459608572608588,
+      "loss": 1.1584,
+      "step": 1364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019458368365508387,
+      "loss": 1.1868,
+      "step": 1365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019457126776503775,
+      "loss": 0.6894,
+      "step": 1366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019455883805776156,
+      "loss": 1.2752,
+      "step": 1367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019454639453507126,
+      "loss": 1.1653,
+      "step": 1368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019453393719878496,
+      "loss": 1.4092,
+      "step": 1369
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001945214660507227,
+      "loss": 0.969,
+      "step": 1370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001945089810927065,
+      "loss": 1.279,
+      "step": 1371
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019449648232656056,
+      "loss": 0.8416,
+      "step": 1372
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019448396975411096,
+      "loss": 1.2533,
+      "step": 1373
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019447144337718584,
+      "loss": 0.9753,
+      "step": 1374
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019445890319761536,
+      "loss": 0.8039,
+      "step": 1375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001944463492172316,
+      "loss": 1.3822,
+      "step": 1376
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001944337814378689,
+      "loss": 1.3882,
+      "step": 1377
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019442119986136337,
+      "loss": 1.1812,
+      "step": 1378
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001944086044895532,
+      "loss": 1.0366,
+      "step": 1379
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001943959953242787,
+      "loss": 1.4088,
+      "step": 1380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019438337236738206,
+      "loss": 1.1152,
+      "step": 1381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019437073562070757,
+      "loss": 1.2085,
+      "step": 1382
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001943580850861015,
+      "loss": 0.9068,
+      "step": 1383
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019434542076541216,
+      "loss": 1.1089,
+      "step": 1384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019433274266048982,
+      "loss": 1.2035,
+      "step": 1385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001943200507731868,
+      "loss": 0.9288,
+      "step": 1386
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019430734510535752,
+      "loss": 0.7041,
+      "step": 1387
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001942946256588582,
+      "loss": 1.277,
+      "step": 1388
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001942818924355473,
+      "loss": 1.1653,
+      "step": 1389
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019426914543728513,
+      "loss": 1.1439,
+      "step": 1390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019425638466593413,
+      "loss": 1.2075,
+      "step": 1391
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019424361012335863,
+      "loss": 1.5675,
+      "step": 1392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001942308218114251,
+      "loss": 1.4552,
+      "step": 1393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019421801973200196,
+      "loss": 0.9401,
+      "step": 1394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019420520388695962,
+      "loss": 1.4874,
+      "step": 1395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001941923742781705,
+      "loss": 1.1256,
+      "step": 1396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019417953090750914,
+      "loss": 0.9126,
+      "step": 1397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019416667377685195,
+      "loss": 1.5535,
+      "step": 1398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019415380288807736,
+      "loss": 1.0443,
+      "step": 1399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019414091824306593,
+      "loss": 1.1838,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019412801984370015,
+      "loss": 0.6923,
+      "step": 1401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001941151076918645,
+      "loss": 1.3061,
+      "step": 1402
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001941021817894455,
+      "loss": 1.1457,
+      "step": 1403
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001940892421383317,
+      "loss": 1.2544,
+      "step": 1404
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019407628874041358,
+      "loss": 1.1219,
+      "step": 1405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019406332159758373,
+      "loss": 1.1846,
+      "step": 1406
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001940503407117367,
+      "loss": 0.914,
+      "step": 1407
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019403734608476904,
+      "loss": 1.3771,
+      "step": 1408
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001940243377185793,
+      "loss": 1.7017,
+      "step": 1409
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019401131561506808,
+      "loss": 0.6964,
+      "step": 1410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019399827977613788,
+      "loss": 1.5495,
+      "step": 1411
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019398523020369338,
+      "loss": 0.6345,
+      "step": 1412
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019397216689964118,
+      "loss": 0.5556,
+      "step": 1413
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001939590898658898,
+      "loss": 1.26,
+      "step": 1414
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019394599910434993,
+      "loss": 0.7784,
+      "step": 1415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001939328946169341,
+      "loss": 0.6654,
+      "step": 1416
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019391977640555695,
+      "loss": 1.0566,
+      "step": 1417
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019390664447213515,
+      "loss": 1.0851,
+      "step": 1418
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001938934988185873,
+      "loss": 1.0723,
+      "step": 1419
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019388033944683398,
+      "loss": 0.6653,
+      "step": 1420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001938671663587979,
+      "loss": 1.2531,
+      "step": 1421
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001938539795564037,
+      "loss": 0.5674,
+      "step": 1422
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019384077904157797,
+      "loss": 1.3222,
+      "step": 1423
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019382756481624934,
+      "loss": 1.3934,
+      "step": 1424
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019381433688234854,
+      "loss": 0.9894,
+      "step": 1425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019380109524180818,
+      "loss": 0.3855,
+      "step": 1426
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001937878398965629,
+      "loss": 1.4618,
+      "step": 1427
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019377457084854938,
+      "loss": 1.142,
+      "step": 1428
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019376128809970629,
+      "loss": 1.3836,
+      "step": 1429
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019374799165197427,
+      "loss": 1.3592,
+      "step": 1430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019373468150729598,
+      "loss": 0.7069,
+      "step": 1431
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019372135766761606,
+      "loss": 0.6349,
+      "step": 1432
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019370802013488124,
+      "loss": 1.482,
+      "step": 1433
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019369466891104014,
+      "loss": 1.2632,
+      "step": 1434
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019368130399804344,
+      "loss": 1.2384,
+      "step": 1435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019366792539784382,
+      "loss": 1.3628,
+      "step": 1436
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019365453311239592,
+      "loss": 1.8586,
+      "step": 1437
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001936411271436564,
+      "loss": 1.719,
+      "step": 1438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019362770749358394,
+      "loss": 1.1557,
+      "step": 1439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019361427416413922,
+      "loss": 0.4752,
+      "step": 1440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001936008271572849,
+      "loss": 0.8127,
+      "step": 1441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019358736647498556,
+      "loss": 0.9974,
+      "step": 1442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019357389211920799,
+      "loss": 1.3706,
+      "step": 1443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019356040409192074,
+      "loss": 0.558,
+      "step": 1444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019354690239509447,
+      "loss": 1.3805,
+      "step": 1445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019353338703070189,
+      "loss": 1.5071,
+      "step": 1446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019351985800071764,
+      "loss": 1.3533,
+      "step": 1447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019350631530711833,
+      "loss": 0.6864,
+      "step": 1448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001934927589518826,
+      "loss": 0.6064,
+      "step": 1449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001934791889369911,
+      "loss": 1.3029,
+      "step": 1450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019346560526442645,
+      "loss": 1.1011,
+      "step": 1451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001934520079361733,
+      "loss": 1.2862,
+      "step": 1452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019343839695421826,
+      "loss": 0.9326,
+      "step": 1453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019342477232054997,
+      "loss": 1.2615,
+      "step": 1454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019341113403715897,
+      "loss": 0.8611,
+      "step": 1455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019339748210603794,
+      "loss": 0.7262,
+      "step": 1456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019338381652918147,
+      "loss": 1.2394,
+      "step": 1457
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001933701373085861,
+      "loss": 1.2003,
+      "step": 1458
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001933564444462505,
+      "loss": 1.3764,
+      "step": 1459
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001933427379441752,
+      "loss": 1.5172,
+      "step": 1460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019332901780436275,
+      "loss": 1.4841,
+      "step": 1461
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019331528402881777,
+      "loss": 1.3864,
+      "step": 1462
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001933015366195468,
+      "loss": 0.7483,
+      "step": 1463
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019328777557855837,
+      "loss": 1.0522,
+      "step": 1464
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019327400090786304,
+      "loss": 1.2312,
+      "step": 1465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019326021260947335,
+      "loss": 1.1168,
+      "step": 1466
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019324641068540382,
+      "loss": 0.6288,
+      "step": 1467
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019323259513767094,
+      "loss": 0.8905,
+      "step": 1468
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019321876596829323,
+      "loss": 1.2105,
+      "step": 1469
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019320492317929117,
+      "loss": 1.1865,
+      "step": 1470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019319106677268732,
+      "loss": 1.2974,
+      "step": 1471
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019317719675050605,
+      "loss": 1.3388,
+      "step": 1472
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001931633131147739,
+      "loss": 1.2985,
+      "step": 1473
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019314941586751928,
+      "loss": 0.6592,
+      "step": 1474
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019313550501077266,
+      "loss": 0.4594,
+      "step": 1475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019312158054656645,
+      "loss": 1.4944,
+      "step": 1476
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001931076424769351,
+      "loss": 0.8976,
+      "step": 1477
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019309369080391497,
+      "loss": 1.4449,
+      "step": 1478
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019307972552954448,
+      "loss": 1.5024,
+      "step": 1479
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019306574665586398,
+      "loss": 1.3376,
+      "step": 1480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019305175418491593,
+      "loss": 1.2726,
+      "step": 1481
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019303774811874457,
+      "loss": 0.6459,
+      "step": 1482
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001930237284593963,
+      "loss": 1.3614,
+      "step": 1483
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019300969520891944,
+      "loss": 0.669,
+      "step": 1484
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019299564836936433,
+      "loss": 1.6789,
+      "step": 1485
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001929815879427832,
+      "loss": 1.506,
+      "step": 1486
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019296751393123036,
+      "loss": 1.2748,
+      "step": 1487
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001929534263367621,
+      "loss": 0.6641,
+      "step": 1488
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001929393251614367,
+      "loss": 1.2649,
+      "step": 1489
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001929252104073143,
+      "loss": 1.1939,
+      "step": 1490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019291108207645724,
+      "loss": 0.7778,
+      "step": 1491
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001928969401709296,
+      "loss": 1.4775,
+      "step": 1492
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019288278469279768,
+      "loss": 1.4279,
+      "step": 1493
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019286861564412957,
+      "loss": 1.2368,
+      "step": 1494
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019285443302699545,
+      "loss": 1.1612,
+      "step": 1495
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019284023684346748,
+      "loss": 0.5491,
+      "step": 1496
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019282602709561975,
+      "loss": 1.527,
+      "step": 1497
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019281180378552835,
+      "loss": 1.0182,
+      "step": 1498
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001927975669152714,
+      "loss": 0.675,
+      "step": 1499
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019278331648692892,
+      "loss": 0.4963,
+      "step": 1500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019276905250258298,
+      "loss": 0.6346,
+      "step": 1501
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001927547749643176,
+      "loss": 0.9669,
+      "step": 1502
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019274048387421877,
+      "loss": 1.3118,
+      "step": 1503
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001927261792343745,
+      "loss": 0.9494,
+      "step": 1504
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019271186104687473,
+      "loss": 0.8994,
+      "step": 1505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001926975293138114,
+      "loss": 0.6192,
+      "step": 1506
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019268318403727844,
+      "loss": 1.0485,
+      "step": 1507
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019266882521937176,
+      "loss": 1.0221,
+      "step": 1508
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001926544528621892,
+      "loss": 1.0048,
+      "step": 1509
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001926400669678307,
+      "loss": 0.691,
+      "step": 1510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019262566753839802,
+      "loss": 0.6507,
+      "step": 1511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000192611254575995,
+      "loss": 0.9147,
+      "step": 1512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001925968280827274,
+      "loss": 1.2525,
+      "step": 1513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019258238806070304,
+      "loss": 1.4399,
+      "step": 1514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019256793451203166,
+      "loss": 1.0585,
+      "step": 1515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001925534674388249,
+      "loss": 0.9492,
+      "step": 1516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019253898684319656,
+      "loss": 1.4068,
+      "step": 1517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019252449272726225,
+      "loss": 0.9782,
+      "step": 1518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019250998509313962,
+      "loss": 1.03,
+      "step": 1519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019249546394294831,
+      "loss": 0.6373,
+      "step": 1520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001924809292788099,
+      "loss": 1.4966,
+      "step": 1521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019246638110284798,
+      "loss": 0.7342,
+      "step": 1522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019245181941718808,
+      "loss": 1.19,
+      "step": 1523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001924372442239577,
+      "loss": 0.6012,
+      "step": 1524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019242265552528638,
+      "loss": 1.2366,
+      "step": 1525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019240805332330555,
+      "loss": 1.2322,
+      "step": 1526
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001923934376201487,
+      "loss": 0.8986,
+      "step": 1527
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019237880841795115,
+      "loss": 1.1672,
+      "step": 1528
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019236416571885036,
+      "loss": 1.3703,
+      "step": 1529
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019234950952498565,
+      "loss": 1.1628,
+      "step": 1530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019233483983849837,
+      "loss": 0.6019,
+      "step": 1531
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001923201566615318,
+      "loss": 0.7839,
+      "step": 1532
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001923054599962312,
+      "loss": 1.3657,
+      "step": 1533
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019229074984474385,
+      "loss": 1.0999,
+      "step": 1534
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001922760262092189,
+      "loss": 1.3778,
+      "step": 1535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001922612890918076,
+      "loss": 1.398,
+      "step": 1536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019224653849466303,
+      "loss": 0.667,
+      "step": 1537
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019223177441994035,
+      "loss": 0.7959,
+      "step": 1538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019221699686979666,
+      "loss": 0.6358,
+      "step": 1539
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000192202205846391,
+      "loss": 1.1305,
+      "step": 1540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019218740135188437,
+      "loss": 1.1221,
+      "step": 1541
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019217258338843985,
+      "loss": 1.4066,
+      "step": 1542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001921577519582223,
+      "loss": 0.8724,
+      "step": 1543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001921429070633987,
+      "loss": 1.4473,
+      "step": 1544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019212804870613793,
+      "loss": 0.9919,
+      "step": 1545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019211317688861085,
+      "loss": 0.8309,
+      "step": 1546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019209829161299035,
+      "loss": 1.0785,
+      "step": 1547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019208339288145115,
+      "loss": 0.4634,
+      "step": 1548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019206848069617006,
+      "loss": 1.5931,
+      "step": 1549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019205355505932579,
+      "loss": 1.3627,
+      "step": 1550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000192038615973099,
+      "loss": 1.6168,
+      "step": 1551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001920236634396724,
+      "loss": 1.518,
+      "step": 1552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019200869746123064,
+      "loss": 1.0623,
+      "step": 1553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019199371803996026,
+      "loss": 1.2879,
+      "step": 1554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019197872517804978,
+      "loss": 1.5717,
+      "step": 1555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019196371887768976,
+      "loss": 1.245,
+      "step": 1556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019194869914107264,
+      "loss": 1.1743,
+      "step": 1557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019193366597039293,
+      "loss": 1.0682,
+      "step": 1558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000191918619367847,
+      "loss": 0.6836,
+      "step": 1559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019190355933563323,
+      "loss": 1.3702,
+      "step": 1560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001918884858759519,
+      "loss": 1.2716,
+      "step": 1561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019187339899100534,
+      "loss": 0.9638,
+      "step": 1562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019185829868299781,
+      "loss": 1.0841,
+      "step": 1563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001918431849541355,
+      "loss": 1.0008,
+      "step": 1564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001918280578066266,
+      "loss": 1.3336,
+      "step": 1565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019181291724268125,
+      "loss": 0.7572,
+      "step": 1566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001917977632645115,
+      "loss": 0.7927,
+      "step": 1567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001917825958743315,
+      "loss": 1.1677,
+      "step": 1568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019176741507435718,
+      "loss": 1.2137,
+      "step": 1569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019175222086680654,
+      "loss": 1.0914,
+      "step": 1570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001917370132538995,
+      "loss": 1.5207,
+      "step": 1571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019172179223785797,
+      "loss": 0.6292,
+      "step": 1572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001917065578209058,
+      "loss": 1.1499,
+      "step": 1573
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001916913100052688,
+      "loss": 0.7429,
+      "step": 1574
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001916760487931747,
+      "loss": 0.8924,
+      "step": 1575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019166077418685326,
+      "loss": 1.1075,
+      "step": 1576
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019164548618853618,
+      "loss": 1.003,
+      "step": 1577
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000191630184800457,
+      "loss": 1.1718,
+      "step": 1578
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019161487002485142,
+      "loss": 0.8551,
+      "step": 1579
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019159954186395695,
+      "loss": 1.2327,
+      "step": 1580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001915842003200131,
+      "loss": 1.1929,
+      "step": 1581
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001915688453952613,
+      "loss": 0.783,
+      "step": 1582
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019155347709194498,
+      "loss": 1.1514,
+      "step": 1583
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001915380954123095,
+      "loss": 1.26,
+      "step": 1584
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001915227003586022,
+      "loss": 1.1275,
+      "step": 1585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019150729193307236,
+      "loss": 0.8997,
+      "step": 1586
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019149187013797118,
+      "loss": 1.2101,
+      "step": 1587
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001914764349755519,
+      "loss": 0.6509,
+      "step": 1588
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019146098644806962,
+      "loss": 1.2085,
+      "step": 1589
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019144552455778143,
+      "loss": 1.1225,
+      "step": 1590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019143004930694639,
+      "loss": 1.0763,
+      "step": 1591
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019141456069782543,
+      "loss": 1.298,
+      "step": 1592
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001913990587326816,
+      "loss": 1.5268,
+      "step": 1593
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019138354341377973,
+      "loss": 0.8912,
+      "step": 1594
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001913680147433867,
+      "loss": 1.2322,
+      "step": 1595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019135247272377125,
+      "loss": 0.9863,
+      "step": 1596
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001913369173572042,
+      "loss": 1.1684,
+      "step": 1597
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001913213486459582,
+      "loss": 1.4914,
+      "step": 1598
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019130576659230796,
+      "loss": 1.247,
+      "step": 1599
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019129017119852998,
+      "loss": 1.341,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019127456246690293,
+      "loss": 0.8985,
+      "step": 1601
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001912589403997072,
+      "loss": 1.0407,
+      "step": 1602
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019124330499922532,
+      "loss": 0.7289,
+      "step": 1603
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001912276562677416,
+      "loss": 1.4188,
+      "step": 1604
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019121199420754243,
+      "loss": 0.7022,
+      "step": 1605
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001911963188209161,
+      "loss": 1.3944,
+      "step": 1606
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019118063011015283,
+      "loss": 0.5996,
+      "step": 1607
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019116492807754482,
+      "loss": 0.9828,
+      "step": 1608
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019114921272538616,
+      "loss": 1.2034,
+      "step": 1609
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019113348405597295,
+      "loss": 0.5457,
+      "step": 1610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019111774207160323,
+      "loss": 1.3503,
+      "step": 1611
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019110198677457692,
+      "loss": 0.8069,
+      "step": 1612
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019108621816719596,
+      "loss": 1.1342,
+      "step": 1613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019107043625176418,
+      "loss": 0.8007,
+      "step": 1614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019105464103058743,
+      "loss": 1.2806,
+      "step": 1615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001910388325059734,
+      "loss": 1.2269,
+      "step": 1616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001910230106802318,
+      "loss": 1.0044,
+      "step": 1617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019100717555567426,
+      "loss": 0.9006,
+      "step": 1618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019099132713461438,
+      "loss": 1.3468,
+      "step": 1619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019097546541936762,
+      "loss": 0.8293,
+      "step": 1620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019095959041225146,
+      "loss": 1.0575,
+      "step": 1621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001909437021155853,
+      "loss": 0.6973,
+      "step": 1622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001909278005316905,
+      "loss": 1.0805,
+      "step": 1623
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019091188566289035,
+      "loss": 1.2008,
+      "step": 1624
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019089595751151,
+      "loss": 1.299,
+      "step": 1625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001908800160798767,
+      "loss": 1.0086,
+      "step": 1626
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019086406137031956,
+      "loss": 1.2545,
+      "step": 1627
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019084809338516955,
+      "loss": 0.8789,
+      "step": 1628
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019083211212675972,
+      "loss": 1.0566,
+      "step": 1629
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019081611759742495,
+      "loss": 1.4455,
+      "step": 1630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019080010979950212,
+      "loss": 0.6264,
+      "step": 1631
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019078408873533006,
+      "loss": 1.3296,
+      "step": 1632
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019076805440724942,
+      "loss": 0.9934,
+      "step": 1633
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019075200681760302,
+      "loss": 1.2281,
+      "step": 1634
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019073594596873537,
+      "loss": 1.0071,
+      "step": 1635
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019071987186299302,
+      "loss": 1.0097,
+      "step": 1636
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001907037845027245,
+      "loss": 1.3781,
+      "step": 1637
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019068768389028025,
+      "loss": 1.1793,
+      "step": 1638
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019067157002801256,
+      "loss": 1.919,
+      "step": 1639
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001906554429182758,
+      "loss": 1.2534,
+      "step": 1640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019063930256342618,
+      "loss": 1.2658,
+      "step": 1641
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001906231489658219,
+      "loss": 1.3807,
+      "step": 1642
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000190606982127823,
+      "loss": 1.4433,
+      "step": 1643
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019059080205179155,
+      "loss": 0.5447,
+      "step": 1644
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001905746087400915,
+      "loss": 0.9555,
+      "step": 1645
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001905584021950888,
+      "loss": 0.6214,
+      "step": 1646
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019054218241915124,
+      "loss": 0.8898,
+      "step": 1647
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019052594941464866,
+      "loss": 0.7997,
+      "step": 1648
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019050970318395266,
+      "loss": 0.4687,
+      "step": 1649
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019049344372943698,
+      "loss": 0.5032,
+      "step": 1650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019047717105347714,
+      "loss": 1.1274,
+      "step": 1651
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019046088515845068,
+      "loss": 0.9972,
+      "step": 1652
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019044458604673697,
+      "loss": 1.2664,
+      "step": 1653
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001904282737207174,
+      "loss": 0.4992,
+      "step": 1654
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001904119481827753,
+      "loss": 1.2053,
+      "step": 1655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019039560943529584,
+      "loss": 1.2526,
+      "step": 1656
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001903792574806662,
+      "loss": 1.1008,
+      "step": 1657
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019036289232127548,
+      "loss": 0.5247,
+      "step": 1658
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001903465139595147,
+      "loss": 0.9362,
+      "step": 1659
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019033012239777676,
+      "loss": 1.2595,
+      "step": 1660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019031371763845653,
+      "loss": 1.6105,
+      "step": 1661
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019029729968395086,
+      "loss": 0.8167,
+      "step": 1662
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019028086853665842,
+      "loss": 0.9322,
+      "step": 1663
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001902644241989799,
+      "loss": 0.9207,
+      "step": 1664
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019024796667331789,
+      "loss": 0.7912,
+      "step": 1665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019023149596207684,
+      "loss": 0.3584,
+      "step": 1666
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019021501206766328,
+      "loss": 1.3047,
+      "step": 1667
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001901985149924855,
+      "loss": 1.4519,
+      "step": 1668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001901820047389538,
+      "loss": 1.0475,
+      "step": 1669
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019016548130948037,
+      "loss": 0.7928,
+      "step": 1670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019014894470647945,
+      "loss": 0.5374,
+      "step": 1671
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019013239493236698,
+      "loss": 1.2519,
+      "step": 1672
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000190115831989561,
+      "loss": 1.5327,
+      "step": 1673
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019009925588048135,
+      "loss": 1.0035,
+      "step": 1674
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019008266660755,
+      "loss": 1.4381,
+      "step": 1675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019006606417319063,
+      "loss": 1.0539,
+      "step": 1676
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001900494485798289,
+      "loss": 1.3728,
+      "step": 1677
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019003281982989248,
+      "loss": 0.5328,
+      "step": 1678
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019001617792581082,
+      "loss": 1.401,
+      "step": 1679
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001899995228700154,
+      "loss": 1.1401,
+      "step": 1680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001899828546649396,
+      "loss": 1.3307,
+      "step": 1681
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018996617331301868,
+      "loss": 1.1334,
+      "step": 1682
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018994947881668988,
+      "loss": 0.6001,
+      "step": 1683
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018993277117839233,
+      "loss": 0.9152,
+      "step": 1684
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018991605040056704,
+      "loss": 1.0532,
+      "step": 1685
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000189899316485657,
+      "loss": 1.1965,
+      "step": 1686
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018988256943610718,
+      "loss": 0.9825,
+      "step": 1687
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018986580925436428,
+      "loss": 0.788,
+      "step": 1688
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018984903594287706,
+      "loss": 0.9492,
+      "step": 1689
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001898322495040962,
+      "loss": 0.895,
+      "step": 1690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001898154499404742,
+      "loss": 1.1173,
+      "step": 1691
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018979863725446561,
+      "loss": 0.7344,
+      "step": 1692
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018978181144852678,
+      "loss": 0.8988,
+      "step": 1693
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018976497252511606,
+      "loss": 1.5158,
+      "step": 1694
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018974812048669365,
+      "loss": 1.4929,
+      "step": 1695
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018973125533572173,
+      "loss": 0.4431,
+      "step": 1696
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018971437707466433,
+      "loss": 1.3457,
+      "step": 1697
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018969748570598743,
+      "loss": 1.3985,
+      "step": 1698
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018968058123215901,
+      "loss": 1.2073,
+      "step": 1699
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018966366365564876,
+      "loss": 1.1702,
+      "step": 1700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018964673297892847,
+      "loss": 1.3237,
+      "step": 1701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018962978920447176,
+      "loss": 0.9055,
+      "step": 1702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001896128323347542,
+      "loss": 0.9964,
+      "step": 1703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018959586237225323,
+      "loss": 0.8281,
+      "step": 1704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018957887931944823,
+      "loss": 1.3417,
+      "step": 1705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018956188317882048,
+      "loss": 1.1283,
+      "step": 1706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018954487395285324,
+      "loss": 1.1676,
+      "step": 1707
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018952785164403157,
+      "loss": 1.1875,
+      "step": 1708
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001895108162548425,
+      "loss": 1.1393,
+      "step": 1709
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018949376778777498,
+      "loss": 1.3687,
+      "step": 1710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018947670624531985,
+      "loss": 0.9969,
+      "step": 1711
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001894596316299699,
+      "loss": 1.2113,
+      "step": 1712
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018944254394421975,
+      "loss": 1.5447,
+      "step": 1713
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018942544319056596,
+      "loss": 0.5612,
+      "step": 1714
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018940832937150708,
+      "loss": 0.4582,
+      "step": 1715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001893912024895435,
+      "loss": 1.3141,
+      "step": 1716
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018937406254717747,
+      "loss": 1.3752,
+      "step": 1717
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018935690954691324,
+      "loss": 1.3855,
+      "step": 1718
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018933974349125693,
+      "loss": 1.0884,
+      "step": 1719
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018932256438271657,
+      "loss": 0.5859,
+      "step": 1720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001893053722238021,
+      "loss": 0.8756,
+      "step": 1721
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001892881670170253,
+      "loss": 1.1989,
+      "step": 1722
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018927094876489999,
+      "loss": 1.5218,
+      "step": 1723
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018925371746994181,
+      "loss": 1.2973,
+      "step": 1724
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018923647313466831,
+      "loss": 1.1966,
+      "step": 1725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018921921576159893,
+      "loss": 0.6172,
+      "step": 1726
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001892019453532551,
+      "loss": 0.5866,
+      "step": 1727
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018918466191216002,
+      "loss": 1.2506,
+      "step": 1728
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018916736544083893,
+      "loss": 0.7726,
+      "step": 1729
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018915005594181885,
+      "loss": 0.8277,
+      "step": 1730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018913273341762885,
+      "loss": 1.1766,
+      "step": 1731
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018911539787079974,
+      "loss": 0.9208,
+      "step": 1732
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018909804930386433,
+      "loss": 0.9029,
+      "step": 1733
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018908068771935735,
+      "loss": 0.9532,
+      "step": 1734
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018906331311981534,
+      "loss": 1.3784,
+      "step": 1735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018904592550777685,
+      "loss": 0.5575,
+      "step": 1736
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018902852488578224,
+      "loss": 1.4729,
+      "step": 1737
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018901111125637382,
+      "loss": 1.5714,
+      "step": 1738
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001889936846220958,
+      "loss": 1.314,
+      "step": 1739
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018897624498549425,
+      "loss": 1.0325,
+      "step": 1740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001889587923491172,
+      "loss": 1.0721,
+      "step": 1741
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018894132671551454,
+      "loss": 0.9561,
+      "step": 1742
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018892384808723804,
+      "loss": 1.147,
+      "step": 1743
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018890635646684145,
+      "loss": 1.2906,
+      "step": 1744
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018888885185688032,
+      "loss": 1.0645,
+      "step": 1745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018887133425991214,
+      "loss": 1.1957,
+      "step": 1746
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018885380367849631,
+      "loss": 1.053,
+      "step": 1747
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018883626011519415,
+      "loss": 1.2378,
+      "step": 1748
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018881870357256882,
+      "loss": 1.4905,
+      "step": 1749
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018880113405318535,
+      "loss": 1.5111,
+      "step": 1750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018878355155961077,
+      "loss": 1.0388,
+      "step": 1751
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018876595609441395,
+      "loss": 1.373,
+      "step": 1752
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018874834766016563,
+      "loss": 0.8267,
+      "step": 1753
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018873072625943848,
+      "loss": 1.1855,
+      "step": 1754
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018871309189480705,
+      "loss": 1.1184,
+      "step": 1755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018869544456884784,
+      "loss": 1.0421,
+      "step": 1756
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018867778428413912,
+      "loss": 0.7704,
+      "step": 1757
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018866011104326115,
+      "loss": 1.3111,
+      "step": 1758
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018864242484879608,
+      "loss": 1.1051,
+      "step": 1759
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018862472570332792,
+      "loss": 0.9951,
+      "step": 1760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018860701360944257,
+      "loss": 0.791,
+      "step": 1761
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018858928856972787,
+      "loss": 0.8719,
+      "step": 1762
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018857155058677347,
+      "loss": 1.2565,
+      "step": 1763
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000188553799663171,
+      "loss": 0.442,
+      "step": 1764
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018853603580151393,
+      "loss": 0.8197,
+      "step": 1765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001885182590043976,
+      "loss": 1.4139,
+      "step": 1766
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018850046927441932,
+      "loss": 1.14,
+      "step": 1767
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018848266661417818,
+      "loss": 1.2995,
+      "step": 1768
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018846485102627525,
+      "loss": 0.4951,
+      "step": 1769
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018844702251331349,
+      "loss": 0.6626,
+      "step": 1770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018842918107789763,
+      "loss": 1.2134,
+      "step": 1771
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018841132672263444,
+      "loss": 1.2734,
+      "step": 1772
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001883934594501325,
+      "loss": 1.3194,
+      "step": 1773
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001883755792630023,
+      "loss": 0.9032,
+      "step": 1774
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018835768616385617,
+      "loss": 1.0871,
+      "step": 1775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018833978015530834,
+      "loss": 1.7721,
+      "step": 1776
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018832186123997504,
+      "loss": 1.3112,
+      "step": 1777
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018830392942047425,
+      "loss": 1.2099,
+      "step": 1778
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018828598469942582,
+      "loss": 1.4477,
+      "step": 1779
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001882680270794516,
+      "loss": 0.6659,
+      "step": 1780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018825005656317528,
+      "loss": 1.508,
+      "step": 1781
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018823207315322237,
+      "loss": 1.225,
+      "step": 1782
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018821407685222039,
+      "loss": 0.8617,
+      "step": 1783
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001881960676627986,
+      "loss": 1.2731,
+      "step": 1784
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018817804558758826,
+      "loss": 0.8411,
+      "step": 1785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001881600106292224,
+      "loss": 0.6154,
+      "step": 1786
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018814196279033608,
+      "loss": 1.1072,
+      "step": 1787
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001881239020735661,
+      "loss": 1.393,
+      "step": 1788
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018810582848155123,
+      "loss": 1.4636,
+      "step": 1789
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001880877420169321,
+      "loss": 0.5152,
+      "step": 1790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018806964268235118,
+      "loss": 1.4918,
+      "step": 1791
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018805153048045287,
+      "loss": 0.6001,
+      "step": 1792
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018803340541388345,
+      "loss": 1.3925,
+      "step": 1793
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018801526748529103,
+      "loss": 1.1937,
+      "step": 1794
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018799711669732565,
+      "loss": 1.1351,
+      "step": 1795
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018797895305263924,
+      "loss": 1.502,
+      "step": 1796
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018796077655388552,
+      "loss": 1.0678,
+      "step": 1797
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001879425872037202,
+      "loss": 0.6948,
+      "step": 1798
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018792438500480073,
+      "loss": 1.529,
+      "step": 1799
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018790616995978664,
+      "loss": 0.5604,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018788794207133915,
+      "loss": 0.8635,
+      "step": 1801
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018786970134212144,
+      "loss": 1.1501,
+      "step": 1802
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018785144777479852,
+      "loss": 0.5758,
+      "step": 1803
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018783318137203736,
+      "loss": 0.6585,
+      "step": 1804
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018781490213650673,
+      "loss": 0.9043,
+      "step": 1805
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001877966100708773,
+      "loss": 1.3309,
+      "step": 1806
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001877783051778216,
+      "loss": 0.6915,
+      "step": 1807
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018775998746001406,
+      "loss": 0.9543,
+      "step": 1808
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018774165692013095,
+      "loss": 0.9274,
+      "step": 1809
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018772331356085045,
+      "loss": 0.6291,
+      "step": 1810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018770495738485262,
+      "loss": 1.4054,
+      "step": 1811
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001876865883948193,
+      "loss": 1.077,
+      "step": 1812
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001876682065934343,
+      "loss": 1.3186,
+      "step": 1813
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018764981198338333,
+      "loss": 1.1572,
+      "step": 1814
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018763140456735385,
+      "loss": 0.3967,
+      "step": 1815
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001876129843480353,
+      "loss": 0.6314,
+      "step": 1816
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001875945513281189,
+      "loss": 0.6285,
+      "step": 1817
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001875761055102978,
+      "loss": 0.5189,
+      "step": 1818
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018755764689726702,
+      "loss": 1.4995,
+      "step": 1819
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018753917549172342,
+      "loss": 1.4002,
+      "step": 1820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018752069129636577,
+      "loss": 0.7863,
+      "step": 1821
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018750219431389465,
+      "loss": 1.117,
+      "step": 1822
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018748368454701253,
+      "loss": 0.8706,
+      "step": 1823
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018746516199842384,
+      "loss": 0.6413,
+      "step": 1824
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018744662667083468,
+      "loss": 1.2692,
+      "step": 1825
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018742807856695324,
+      "loss": 1.4746,
+      "step": 1826
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001874095176894894,
+      "loss": 1.5397,
+      "step": 1827
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018739094404115502,
+      "loss": 1.1364,
+      "step": 1828
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001873723576246637,
+      "loss": 0.6628,
+      "step": 1829
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018735375844273114,
+      "loss": 1.6258,
+      "step": 1830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001873351464980746,
+      "loss": 1.0757,
+      "step": 1831
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018731652179341347,
+      "loss": 1.4091,
+      "step": 1832
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001872978843314688,
+      "loss": 1.4504,
+      "step": 1833
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018727923411496366,
+      "loss": 1.5143,
+      "step": 1834
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018726057114662288,
+      "loss": 1.0175,
+      "step": 1835
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001872418954291732,
+      "loss": 1.1219,
+      "step": 1836
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018722320696534328,
+      "loss": 1.2372,
+      "step": 1837
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001872045057578635,
+      "loss": 1.4364,
+      "step": 1838
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018718579180946617,
+      "loss": 1.263,
+      "step": 1839
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001871670651228855,
+      "loss": 1.3216,
+      "step": 1840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018714832570085755,
+      "loss": 0.8492,
+      "step": 1841
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018712957354612023,
+      "loss": 1.2707,
+      "step": 1842
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018711080866141324,
+      "loss": 0.6856,
+      "step": 1843
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018709203104947829,
+      "loss": 1.3746,
+      "step": 1844
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018707324071305878,
+      "loss": 1.228,
+      "step": 1845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018705443765490006,
+      "loss": 1.5457,
+      "step": 1846
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001870356218777494,
+      "loss": 0.6249,
+      "step": 1847
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001870167933843558,
+      "loss": 1.3917,
+      "step": 1848
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018699795217747017,
+      "loss": 1.3241,
+      "step": 1849
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018697909825984532,
+      "loss": 1.0535,
+      "step": 1850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018696023163423587,
+      "loss": 0.7228,
+      "step": 1851
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001869413523033983,
+      "loss": 1.0684,
+      "step": 1852
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018692246027009093,
+      "loss": 0.9085,
+      "step": 1853
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018690355553707398,
+      "loss": 1.6068,
+      "step": 1854
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018688463810710956,
+      "loss": 0.8108,
+      "step": 1855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001868657079829615,
+      "loss": 0.5358,
+      "step": 1856
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018684676516739557,
+      "loss": 1.0851,
+      "step": 1857
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018682780966317944,
+      "loss": 1.1935,
+      "step": 1858
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018680884147308254,
+      "loss": 0.5764,
+      "step": 1859
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018678986059987625,
+      "loss": 0.8293,
+      "step": 1860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018677086704633369,
+      "loss": 1.3834,
+      "step": 1861
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001867518608152299,
+      "loss": 0.7954,
+      "step": 1862
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001867328419093418,
+      "loss": 1.4171,
+      "step": 1863
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018671381033144807,
+      "loss": 1.0027,
+      "step": 1864
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018669476608432935,
+      "loss": 1.2748,
+      "step": 1865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018667570917076806,
+      "loss": 1.4103,
+      "step": 1866
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001866566395935485,
+      "loss": 1.1988,
+      "step": 1867
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018663755735545679,
+      "loss": 1.135,
+      "step": 1868
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001866184624592809,
+      "loss": 1.2275,
+      "step": 1869
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018659935490781076,
+      "loss": 1.1433,
+      "step": 1870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018658023470383795,
+      "loss": 1.1722,
+      "step": 1871
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018656110185015604,
+      "loss": 1.1096,
+      "step": 1872
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018654195634956047,
+      "loss": 0.985,
+      "step": 1873
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018652279820484838,
+      "loss": 1.1818,
+      "step": 1874
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018650362741881894,
+      "loss": 1.0175,
+      "step": 1875
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018648444399427301,
+      "loss": 1.2312,
+      "step": 1876
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018646524793401338,
+      "loss": 1.6352,
+      "step": 1877
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001864460392408447,
+      "loss": 1.0016,
+      "step": 1878
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018642681791757336,
+      "loss": 1.4272,
+      "step": 1879
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018640758396700776,
+      "loss": 0.7507,
+      "step": 1880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018638833739195802,
+      "loss": 1.38,
+      "step": 1881
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001863690781952361,
+      "loss": 0.6997,
+      "step": 1882
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001863498063796559,
+      "loss": 1.2741,
+      "step": 1883
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001863305219480331,
+      "loss": 1.1386,
+      "step": 1884
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018631122490318517,
+      "loss": 1.2025,
+      "step": 1885
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018629191524793156,
+      "loss": 1.1562,
+      "step": 1886
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018627259298509348,
+      "loss": 0.9511,
+      "step": 1887
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018625325811749393,
+      "loss": 0.7982,
+      "step": 1888
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018623391064795784,
+      "loss": 1.0254,
+      "step": 1889
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018621455057931194,
+      "loss": 0.9443,
+      "step": 1890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018619517791438486,
+      "loss": 1.3402,
+      "step": 1891
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018617579265600695,
+      "loss": 0.7218,
+      "step": 1892
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018615639480701055,
+      "loss": 1.1644,
+      "step": 1893
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001861369843702297,
+      "loss": 0.846,
+      "step": 1894
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001861175613485003,
+      "loss": 1.3926,
+      "step": 1895
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018609812574466026,
+      "loss": 1.3618,
+      "step": 1896
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001860786775615491,
+      "loss": 0.7966,
+      "step": 1897
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018605921680200834,
+      "loss": 1.2672,
+      "step": 1898
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018603974346888114,
+      "loss": 1.3093,
+      "step": 1899
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001860202575650128,
+      "loss": 1.304,
+      "step": 1900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018600075909325022,
+      "loss": 0.764,
+      "step": 1901
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018598124805644213,
+      "loss": 1.4238,
+      "step": 1902
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018596172445743925,
+      "loss": 0.9438,
+      "step": 1903
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018594218829909404,
+      "loss": 0.8069,
+      "step": 1904
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001859226395842608,
+      "loss": 0.9729,
+      "step": 1905
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001859030783157957,
+      "loss": 0.9921,
+      "step": 1906
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018588350449655664,
+      "loss": 1.4681,
+      "step": 1907
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018586391812940354,
+      "loss": 1.0921,
+      "step": 1908
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018584431921719795,
+      "loss": 0.7899,
+      "step": 1909
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018582470776280343,
+      "loss": 0.9694,
+      "step": 1910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018580508376908522,
+      "loss": 1.3768,
+      "step": 1911
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018578544723891048,
+      "loss": 0.9933,
+      "step": 1912
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018576579817514816,
+      "loss": 0.9029,
+      "step": 1913
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018574613658066914,
+      "loss": 0.6615,
+      "step": 1914
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000185726462458346,
+      "loss": 1.1417,
+      "step": 1915
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018570677581105323,
+      "loss": 0.8339,
+      "step": 1916
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018568707664166708,
+      "loss": 1.4287,
+      "step": 1917
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018566736495306577,
+      "loss": 1.2047,
+      "step": 1918
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018564764074812914,
+      "loss": 1.6605,
+      "step": 1919
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000185627904029739,
+      "loss": 1.329,
+      "step": 1920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018560815480077905,
+      "loss": 1.163,
+      "step": 1921
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001855883930641346,
+      "loss": 1.3724,
+      "step": 1922
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018556861882269306,
+      "loss": 0.8119,
+      "step": 1923
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001855488320793434,
+      "loss": 0.8586,
+      "step": 1924
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018552903283697658,
+      "loss": 0.4742,
+      "step": 1925
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018550922109848537,
+      "loss": 1.1616,
+      "step": 1926
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018548939686676433,
+      "loss": 1.1011,
+      "step": 1927
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018546956014470984,
+      "loss": 0.6752,
+      "step": 1928
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001854497109352201,
+      "loss": 1.447,
+      "step": 1929
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018542984924119527,
+      "loss": 1.1793,
+      "step": 1930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001854099750655371,
+      "loss": 0.578,
+      "step": 1931
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001853900884111493,
+      "loss": 0.7817,
+      "step": 1932
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018537018928093746,
+      "loss": 0.8956,
+      "step": 1933
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018535027767780884,
+      "loss": 1.5453,
+      "step": 1934
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018533035360467263,
+      "loss": 1.4871,
+      "step": 1935
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018531041706443984,
+      "loss": 1.2886,
+      "step": 1936
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018529046806002326,
+      "loss": 1.1936,
+      "step": 1937
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018527050659433752,
+      "loss": 1.5492,
+      "step": 1938
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018525053267029904,
+      "loss": 1.4062,
+      "step": 1939
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018523054629082612,
+      "loss": 0.8568,
+      "step": 1940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018521054745883887,
+      "loss": 1.2556,
+      "step": 1941
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001851905361772591,
+      "loss": 0.711,
+      "step": 1942
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018517051244901065,
+      "loss": 1.0887,
+      "step": 1943
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000185150476277019,
+      "loss": 1.149,
+      "step": 1944
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018513042766421154,
+      "loss": 1.2167,
+      "step": 1945
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018511036661351742,
+      "loss": 1.2098,
+      "step": 1946
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018509029312786766,
+      "loss": 0.7091,
+      "step": 1947
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001850702072101951,
+      "loss": 1.2757,
+      "step": 1948
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001850501088634343,
+      "loss": 1.3464,
+      "step": 1949
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018502999809052176,
+      "loss": 0.5479,
+      "step": 1950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018500987489439573,
+      "loss": 1.4049,
+      "step": 1951
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001849897392779963,
+      "loss": 1.2561,
+      "step": 1952
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018496959124426535,
+      "loss": 1.0915,
+      "step": 1953
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018494943079614657,
+      "loss": 0.8812,
+      "step": 1954
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018492925793658552,
+      "loss": 1.1284,
+      "step": 1955
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018490907266852948,
+      "loss": 1.1402,
+      "step": 1956
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001848888749949276,
+      "loss": 1.6625,
+      "step": 1957
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018486866491873092,
+      "loss": 1.087,
+      "step": 1958
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018484844244289215,
+      "loss": 1.3124,
+      "step": 1959
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018482820757036587,
+      "loss": 1.3606,
+      "step": 1960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018480796030410844,
+      "loss": 1.0535,
+      "step": 1961
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018478770064707815,
+      "loss": 0.6045,
+      "step": 1962
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018476742860223494,
+      "loss": 1.6295,
+      "step": 1963
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001847471441725407,
+      "loss": 1.1171,
+      "step": 1964
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000184726847360959,
+      "loss": 1.3431,
+      "step": 1965
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001847065381704553,
+      "loss": 0.5112,
+      "step": 1966
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001846862166039969,
+      "loss": 1.5048,
+      "step": 1967
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018466588266455278,
+      "loss": 0.5931,
+      "step": 1968
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001846455363550939,
+      "loss": 1.3409,
+      "step": 1969
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018462517767859287,
+      "loss": 1.6754,
+      "step": 1970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018460480663802414,
+      "loss": 0.4722,
+      "step": 1971
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001845844232363641,
+      "loss": 1.3028,
+      "step": 1972
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018456402747659076,
+      "loss": 0.8299,
+      "step": 1973
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018454361936168408,
+      "loss": 0.8051,
+      "step": 1974
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001845231988946257,
+      "loss": 1.3768,
+      "step": 1975
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001845027660783992,
+      "loss": 1.4241,
+      "step": 1976
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001844823209159898,
+      "loss": 0.854,
+      "step": 1977
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018446186341038474,
+      "loss": 1.2637,
+      "step": 1978
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018444139356457284,
+      "loss": 1.5696,
+      "step": 1979
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018442091138154486,
+      "loss": 1.0974,
+      "step": 1980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018440041686429334,
+      "loss": 1.6269,
+      "step": 1981
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018437991001581257,
+      "loss": 1.3399,
+      "step": 1982
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018435939083909876,
+      "loss": 0.8259,
+      "step": 1983
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018433885933714973,
+      "loss": 1.0109,
+      "step": 1984
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018431831551296533,
+      "loss": 1.3727,
+      "step": 1985
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000184297759369547,
+      "loss": 0.7124,
+      "step": 1986
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001842771909098981,
+      "loss": 0.7836,
+      "step": 1987
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018425661013702377,
+      "loss": 1.0649,
+      "step": 1988
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018423601705393094,
+      "loss": 1.2865,
+      "step": 1989
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018421541166362837,
+      "loss": 1.2127,
+      "step": 1990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018419479396912652,
+      "loss": 0.9755,
+      "step": 1991
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018417416397343775,
+      "loss": 1.5081,
+      "step": 1992
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001841535216795762,
+      "loss": 1.069,
+      "step": 1993
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018413286709055777,
+      "loss": 0.957,
+      "step": 1994
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001841122002094002,
+      "loss": 0.7449,
+      "step": 1995
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018409152103912292,
+      "loss": 0.6598,
+      "step": 1996
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018407082958274735,
+      "loss": 1.2796,
+      "step": 1997
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018405012584329653,
+      "loss": 0.4186,
+      "step": 1998
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018402940982379534,
+      "loss": 0.6103,
+      "step": 1999
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018400868152727054,
+      "loss": 1.1533,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018398794095675057,
+      "loss": 0.9718,
+      "step": 2001
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018396718811526568,
+      "loss": 1.4949,
+      "step": 2002
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018394642300584798,
+      "loss": 1.216,
+      "step": 2003
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018392564563153135,
+      "loss": 0.6764,
+      "step": 2004
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018390485599535138,
+      "loss": 1.2823,
+      "step": 2005
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018388405410034556,
+      "loss": 0.6827,
+      "step": 2006
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018386323994955316,
+      "loss": 1.1028,
+      "step": 2007
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018384241354601517,
+      "loss": 1.3968,
+      "step": 2008
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001838215748927744,
+      "loss": 0.8564,
+      "step": 2009
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018380072399287547,
+      "loss": 0.8507,
+      "step": 2010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001837798608493648,
+      "loss": 1.0363,
+      "step": 2011
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001837589854652905,
+      "loss": 0.7487,
+      "step": 2012
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018373809784370262,
+      "loss": 1.2174,
+      "step": 2013
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018371719798765293,
+      "loss": 1.001,
+      "step": 2014
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018369628590019494,
+      "loss": 1.5109,
+      "step": 2015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000183675361584384,
+      "loss": 1.2479,
+      "step": 2016
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001836544250432772,
+      "loss": 1.4278,
+      "step": 2017
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018363347627993354,
+      "loss": 1.0708,
+      "step": 2018
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018361251529741362,
+      "loss": 0.5869,
+      "step": 2019
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018359154209877997,
+      "loss": 0.8701,
+      "step": 2020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018357055668709684,
+      "loss": 1.1453,
+      "step": 2021
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001835495590654303,
+      "loss": 1.0292,
+      "step": 2022
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018352854923684814,
+      "loss": 0.7087,
+      "step": 2023
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018350752720442003,
+      "loss": 0.9727,
+      "step": 2024
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018348649297121735,
+      "loss": 0.6948,
+      "step": 2025
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001834654465403133,
+      "loss": 0.5315,
+      "step": 2026
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001834443879147828,
+      "loss": 1.1326,
+      "step": 2027
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018342331709770264,
+      "loss": 1.1122,
+      "step": 2028
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018340223409215133,
+      "loss": 0.6508,
+      "step": 2029
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018338113890120918,
+      "loss": 0.9626,
+      "step": 2030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018336003152795826,
+      "loss": 0.9299,
+      "step": 2031
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001833389119754825,
+      "loss": 0.8185,
+      "step": 2032
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001833177802468675,
+      "loss": 1.4671,
+      "step": 2033
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018329663634520068,
+      "loss": 1.0893,
+      "step": 2034
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001832754802735713,
+      "loss": 1.201,
+      "step": 2035
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018325431203507027,
+      "loss": 0.6927,
+      "step": 2036
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001832331316327904,
+      "loss": 1.1761,
+      "step": 2037
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018321193906982623,
+      "loss": 1.1868,
+      "step": 2038
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018319073434927406,
+      "loss": 1.31,
+      "step": 2039
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018316951747423196,
+      "loss": 1.2368,
+      "step": 2040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018314828844779988,
+      "loss": 0.7739,
+      "step": 2041
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018312704727307936,
+      "loss": 1.1614,
+      "step": 2042
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001831057939531739,
+      "loss": 1.3172,
+      "step": 2043
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018308452849118863,
+      "loss": 0.8406,
+      "step": 2044
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018306325089023057,
+      "loss": 1.3096,
+      "step": 2045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018304196115340843,
+      "loss": 1.355,
+      "step": 2046
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018302065928383273,
+      "loss": 1.3652,
+      "step": 2047
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018299934528461577,
+      "loss": 1.5479,
+      "step": 2048
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018297801915887157,
+      "loss": 0.9439,
+      "step": 2049
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000182956680909716,
+      "loss": 1.1537,
+      "step": 2050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018293533054026666,
+      "loss": 1.3186,
+      "step": 2051
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018291396805364288,
+      "loss": 0.5416,
+      "step": 2052
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018289259345296584,
+      "loss": 1.0408,
+      "step": 2053
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018287120674135844,
+      "loss": 1.4571,
+      "step": 2054
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018284980792194534,
+      "loss": 1.4152,
+      "step": 2055
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018282839699785308,
+      "loss": 1.6428,
+      "step": 2056
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018280697397220975,
+      "loss": 0.8232,
+      "step": 2057
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018278553884814542,
+      "loss": 1.7124,
+      "step": 2058
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018276409162879185,
+      "loss": 1.356,
+      "step": 2059
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018274263231728252,
+      "loss": 0.9134,
+      "step": 2060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018272116091675274,
+      "loss": 1.4296,
+      "step": 2061
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018269967743033956,
+      "loss": 1.3082,
+      "step": 2062
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001826781818611818,
+      "loss": 0.6385,
+      "step": 2063
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018265667421242006,
+      "loss": 1.314,
+      "step": 2064
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018263515448719666,
+      "loss": 1.0358,
+      "step": 2065
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018261362268865577,
+      "loss": 1.0874,
+      "step": 2066
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001825920788199432,
+      "loss": 1.0422,
+      "step": 2067
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018257052288420664,
+      "loss": 1.0159,
+      "step": 2068
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001825489548845955,
+      "loss": 0.9313,
+      "step": 2069
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018252737482426094,
+      "loss": 1.3998,
+      "step": 2070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018250578270635588,
+      "loss": 1.0319,
+      "step": 2071
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018248417853403498,
+      "loss": 0.6623,
+      "step": 2072
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018246256231045478,
+      "loss": 1.3378,
+      "step": 2073
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018244093403877343,
+      "loss": 1.284,
+      "step": 2074
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001824192937221509,
+      "loss": 1.3622,
+      "step": 2075
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018239764136374894,
+      "loss": 1.1764,
+      "step": 2076
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018237597696673107,
+      "loss": 1.4568,
+      "step": 2077
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001823543005342625,
+      "loss": 1.1787,
+      "step": 2078
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001823326120695103,
+      "loss": 1.4561,
+      "step": 2079
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018231091157564318,
+      "loss": 1.1491,
+      "step": 2080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018228919905583172,
+      "loss": 0.5577,
+      "step": 2081
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001822674745132481,
+      "loss": 1.0039,
+      "step": 2082
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018224573795106648,
+      "loss": 1.2275,
+      "step": 2083
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018222398937246258,
+      "loss": 1.1524,
+      "step": 2084
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000182202228780614,
+      "loss": 1.0352,
+      "step": 2085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018218045617870003,
+      "loss": 1.6691,
+      "step": 2086
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018215867156990169,
+      "loss": 1.2616,
+      "step": 2087
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001821368749574019,
+      "loss": 0.8479,
+      "step": 2088
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018211506634438512,
+      "loss": 1.1309,
+      "step": 2089
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018209324573403773,
+      "loss": 1.1899,
+      "step": 2090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018207141312954778,
+      "loss": 1.3639,
+      "step": 2091
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018204956853410512,
+      "loss": 1.191,
+      "step": 2092
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001820277119509013,
+      "loss": 1.0463,
+      "step": 2093
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018200584338312972,
+      "loss": 1.1175,
+      "step": 2094
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018198396283398537,
+      "loss": 0.6149,
+      "step": 2095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018196207030666517,
+      "loss": 1.2848,
+      "step": 2096
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018194016580436767,
+      "loss": 1.3307,
+      "step": 2097
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018191824933029316,
+      "loss": 1.3842,
+      "step": 2098
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018189632088764376,
+      "loss": 0.5482,
+      "step": 2099
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001818743804796233,
+      "loss": 0.6925,
+      "step": 2100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001818524281094374,
+      "loss": 1.2252,
+      "step": 2101
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018183046378029332,
+      "loss": 0.8609,
+      "step": 2102
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018180848749540013,
+      "loss": 1.2029,
+      "step": 2103
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001817864992579687,
+      "loss": 1.3959,
+      "step": 2104
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001817644990712116,
+      "loss": 0.9362,
+      "step": 2105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001817424869383431,
+      "loss": 1.2018,
+      "step": 2106
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018172046286257925,
+      "loss": 1.2489,
+      "step": 2107
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018169842684713792,
+      "loss": 1.1906,
+      "step": 2108
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018167637889523864,
+      "loss": 0.8843,
+      "step": 2109
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018165431901010263,
+      "loss": 0.9307,
+      "step": 2110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000181632247194953,
+      "loss": 1.3506,
+      "step": 2111
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018161016345301451,
+      "loss": 0.8119,
+      "step": 2112
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018158806778751364,
+      "loss": 1.0757,
+      "step": 2113
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018156596020167873,
+      "loss": 1.1894,
+      "step": 2114
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001815438406987397,
+      "loss": 1.1689,
+      "step": 2115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018152170928192836,
+      "loss": 1.0112,
+      "step": 2116
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018149956595447817,
+      "loss": 1.2375,
+      "step": 2117
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018147741071962436,
+      "loss": 1.5293,
+      "step": 2118
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001814552435806039,
+      "loss": 0.6345,
+      "step": 2119
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018143306454065548,
+      "loss": 1.2109,
+      "step": 2120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018141087360301955,
+      "loss": 1.2669,
+      "step": 2121
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001813886707709383,
+      "loss": 0.6705,
+      "step": 2122
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018136645604765566,
+      "loss": 1.3587,
+      "step": 2123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018134422943641724,
+      "loss": 1.5021,
+      "step": 2124
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018132199094047052,
+      "loss": 1.2191,
+      "step": 2125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018129974056306454,
+      "loss": 1.374,
+      "step": 2126
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001812774783074502,
+      "loss": 1.1714,
+      "step": 2127
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018125520417688014,
+      "loss": 0.6602,
+      "step": 2128
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018123291817460864,
+      "loss": 1.2139,
+      "step": 2129
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001812106203038918,
+      "loss": 1.1225,
+      "step": 2130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001811883105679874,
+      "loss": 1.1913,
+      "step": 2131
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018116598897015504,
+      "loss": 0.8283,
+      "step": 2132
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018114365551365595,
+      "loss": 1.3756,
+      "step": 2133
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018112131020175314,
+      "loss": 1.3253,
+      "step": 2134
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018109895303771134,
+      "loss": 0.926,
+      "step": 2135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018107658402479704,
+      "loss": 0.8862,
+      "step": 2136
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001810542031662784,
+      "loss": 0.9636,
+      "step": 2137
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018103181046542536,
+      "loss": 1.3865,
+      "step": 2138
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018100940592550964,
+      "loss": 1.4222,
+      "step": 2139
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018098698954980454,
+      "loss": 1.2533,
+      "step": 2140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018096456134158526,
+      "loss": 0.8333,
+      "step": 2141
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001809421213041286,
+      "loss": 0.8803,
+      "step": 2142
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018091966944071316,
+      "loss": 1.2993,
+      "step": 2143
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018089720575461922,
+      "loss": 0.6674,
+      "step": 2144
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001808747302491288,
+      "loss": 1.4678,
+      "step": 2145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001808522429275257,
+      "loss": 0.8569,
+      "step": 2146
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018082974379309542,
+      "loss": 0.429,
+      "step": 2147
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018080723284912516,
+      "loss": 0.9285,
+      "step": 2148
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001807847100989038,
+      "loss": 1.5662,
+      "step": 2149
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018076217554572206,
+      "loss": 0.878,
+      "step": 2150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018073962919287227,
+      "loss": 1.5532,
+      "step": 2151
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001807170710436486,
+      "loss": 1.2266,
+      "step": 2152
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018069450110134687,
+      "loss": 1.2282,
+      "step": 2153
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018067191936926462,
+      "loss": 0.9255,
+      "step": 2154
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018064932585070114,
+      "loss": 1.1934,
+      "step": 2155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001806267205489574,
+      "loss": 1.4562,
+      "step": 2156
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018060410346733616,
+      "loss": 1.2433,
+      "step": 2157
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001805814746091419,
+      "loss": 1.4965,
+      "step": 2158
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018055883397768066,
+      "loss": 1.2329,
+      "step": 2159
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018053618157626042,
+      "loss": 0.9546,
+      "step": 2160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018051351740819077,
+      "loss": 1.3735,
+      "step": 2161
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018049084147678303,
+      "loss": 1.5871,
+      "step": 2162
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018046815378535027,
+      "loss": 0.6235,
+      "step": 2163
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018044545433720718,
+      "loss": 0.7997,
+      "step": 2164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018042274313567028,
+      "loss": 0.6758,
+      "step": 2165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018040002018405776,
+      "loss": 1.1441,
+      "step": 2166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018037728548568954,
+      "loss": 1.4504,
+      "step": 2167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018035453904388727,
+      "loss": 1.4327,
+      "step": 2168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018033178086197422,
+      "loss": 0.611,
+      "step": 2169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001803090109432755,
+      "loss": 0.6701,
+      "step": 2170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018028622929111787,
+      "loss": 0.7301,
+      "step": 2171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018026343590882987,
+      "loss": 0.945,
+      "step": 2172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001802406307997416,
+      "loss": 1.1977,
+      "step": 2173
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018021781396718506,
+      "loss": 0.9938,
+      "step": 2174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018019498541449385,
+      "loss": 1.3612,
+      "step": 2175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001801721451450033,
+      "loss": 1.4993,
+      "step": 2176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018014929316205052,
+      "loss": 1.1875,
+      "step": 2177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018012642946897418,
+      "loss": 1.2837,
+      "step": 2178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018010355406911481,
+      "loss": 1.0023,
+      "step": 2179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018008066696581458,
+      "loss": 0.9773,
+      "step": 2180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001800577681624174,
+      "loss": 0.8403,
+      "step": 2181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018003485766226887,
+      "loss": 1.3165,
+      "step": 2182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001800119354687163,
+      "loss": 1.4511,
+      "step": 2183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017998900158510867,
+      "loss": 1.2699,
+      "step": 2184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001799660560147968,
+      "loss": 0.9618,
+      "step": 2185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017994309876113301,
+      "loss": 1.117,
+      "step": 2186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017992012982747158,
+      "loss": 1.2059,
+      "step": 2187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017989714921716826,
+      "loss": 1.4366,
+      "step": 2188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017987415693358064,
+      "loss": 1.1327,
+      "step": 2189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017985115298006798,
+      "loss": 1.3186,
+      "step": 2190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017982813735999126,
+      "loss": 0.6831,
+      "step": 2191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017980511007671313,
+      "loss": 1.3865,
+      "step": 2192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017978207113359798,
+      "loss": 1.3074,
+      "step": 2193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017975902053401187,
+      "loss": 1.3544,
+      "step": 2194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017973595828132258,
+      "loss": 0.6003,
+      "step": 2195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017971288437889962,
+      "loss": 1.0126,
+      "step": 2196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001796897988301142,
+      "loss": 1.3038,
+      "step": 2197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017966670163833913,
+      "loss": 1.3954,
+      "step": 2198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001796435928069491,
+      "loss": 1.247,
+      "step": 2199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017962047233932032,
+      "loss": 0.9823,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001795973402388308,
+      "loss": 0.6131,
+      "step": 2201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017957419650886024,
+      "loss": 1.2659,
+      "step": 2202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017955104115279,
+      "loss": 1.4526,
+      "step": 2203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001795278741740032,
+      "loss": 1.4498,
+      "step": 2204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017950469557588463,
+      "loss": 1.6175,
+      "step": 2205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017948150536182075,
+      "loss": 0.8248,
+      "step": 2206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017945830353519976,
+      "loss": 0.8663,
+      "step": 2207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017943509009941155,
+      "loss": 1.0632,
+      "step": 2208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017941186505784765,
+      "loss": 0.8179,
+      "step": 2209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017938862841390134,
+      "loss": 0.6404,
+      "step": 2210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001793653801709676,
+      "loss": 1.3874,
+      "step": 2211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001793421203324431,
+      "loss": 1.2475,
+      "step": 2212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017931884890172616,
+      "loss": 0.7675,
+      "step": 2213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017929556588221684,
+      "loss": 1.2151,
+      "step": 2214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017927227127731692,
+      "loss": 0.912,
+      "step": 2215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017924896509042976,
+      "loss": 1.1846,
+      "step": 2216
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017922564732496054,
+      "loss": 1.3076,
+      "step": 2217
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017920231798431604,
+      "loss": 0.5289,
+      "step": 2218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017917897707190477,
+      "loss": 0.8601,
+      "step": 2219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000179155624591137,
+      "loss": 0.4549,
+      "step": 2220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017913226054542451,
+      "loss": 1.1541,
+      "step": 2221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017910888493818093,
+      "loss": 1.5332,
+      "step": 2222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017908549777282152,
+      "loss": 1.3124,
+      "step": 2223
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017906209905276327,
+      "loss": 0.6459,
+      "step": 2224
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001790386887814248,
+      "loss": 1.1506,
+      "step": 2225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001790152669622264,
+      "loss": 1.6595,
+      "step": 2226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017899183359859012,
+      "loss": 1.4351,
+      "step": 2227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001789683886939397,
+      "loss": 1.2866,
+      "step": 2228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001789449322517005,
+      "loss": 0.7585,
+      "step": 2229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001789214642752996,
+      "loss": 1.0789,
+      "step": 2230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017889798476816574,
+      "loss": 1.2626,
+      "step": 2231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017887449373372942,
+      "loss": 1.1936,
+      "step": 2232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017885099117542273,
+      "loss": 1.3966,
+      "step": 2233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001788274770966795,
+      "loss": 1.6503,
+      "step": 2234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001788039515009352,
+      "loss": 1.5013,
+      "step": 2235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017878041439162707,
+      "loss": 1.3284,
+      "step": 2236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017875686577219392,
+      "loss": 0.5729,
+      "step": 2237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017873330564607633,
+      "loss": 1.3362,
+      "step": 2238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017870973401671648,
+      "loss": 1.3408,
+      "step": 2239
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017868615088755836,
+      "loss": 1.2113,
+      "step": 2240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017866255626204747,
+      "loss": 0.9598,
+      "step": 2241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001786389501436311,
+      "loss": 0.6196,
+      "step": 2242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017861533253575823,
+      "loss": 1.0343,
+      "step": 2243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017859170344187945,
+      "loss": 1.2363,
+      "step": 2244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001785680628654471,
+      "loss": 1.2263,
+      "step": 2245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001785444108099151,
+      "loss": 1.0974,
+      "step": 2246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017852074727873912,
+      "loss": 1.1405,
+      "step": 2247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017849707227537658,
+      "loss": 1.3817,
+      "step": 2248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017847338580328636,
+      "loss": 1.2601,
+      "step": 2249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017844968786592927,
+      "loss": 1.0475,
+      "step": 2250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017842597846676758,
+      "loss": 1.2851,
+      "step": 2251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017840225760926533,
+      "loss": 0.6748,
+      "step": 2252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017837852529688832,
+      "loss": 0.7667,
+      "step": 2253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017835478153310384,
+      "loss": 0.9711,
+      "step": 2254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017833102632138097,
+      "loss": 1.0246,
+      "step": 2255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017830725966519042,
+      "loss": 1.0694,
+      "step": 2256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017828348156800466,
+      "loss": 1.2062,
+      "step": 2257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001782596920332977,
+      "loss": 0.94,
+      "step": 2258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017823589106454531,
+      "loss": 1.2201,
+      "step": 2259
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017821207866522488,
+      "loss": 1.381,
+      "step": 2260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017818825483881552,
+      "loss": 1.0013,
+      "step": 2261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017816441958879798,
+      "loss": 0.9568,
+      "step": 2262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017814057291865468,
+      "loss": 1.3184,
+      "step": 2263
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017811671483186967,
+      "loss": 0.6286,
+      "step": 2264
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017809284533192878,
+      "loss": 1.073,
+      "step": 2265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001780689644223194,
+      "loss": 0.7364,
+      "step": 2266
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017804507210653062,
+      "loss": 1.3419,
+      "step": 2267
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001780211683880532,
+      "loss": 0.9776,
+      "step": 2268
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001779972532703796,
+      "loss": 1.0114,
+      "step": 2269
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017797332675700386,
+      "loss": 0.5646,
+      "step": 2270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001779493888514218,
+      "loss": 1.6079,
+      "step": 2271
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017792543955713077,
+      "loss": 1.247,
+      "step": 2272
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017790147887762988,
+      "loss": 1.1862,
+      "step": 2273
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017787750681641992,
+      "loss": 1.1937,
+      "step": 2274
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017785352337700326,
+      "loss": 1.4404,
+      "step": 2275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017782952856288398,
+      "loss": 1.172,
+      "step": 2276
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017780552237756783,
+      "loss": 1.0232,
+      "step": 2277
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017778150482456216,
+      "loss": 0.8903,
+      "step": 2278
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017775747590737607,
+      "loss": 0.8368,
+      "step": 2279
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001777334356295203,
+      "loss": 0.7123,
+      "step": 2280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017770938399450716,
+      "loss": 1.0384,
+      "step": 2281
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017768532100585077,
+      "loss": 0.8419,
+      "step": 2282
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017766124666706676,
+      "loss": 1.2338,
+      "step": 2283
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001776371609816725,
+      "loss": 0.7717,
+      "step": 2284
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017761306395318703,
+      "loss": 1.3322,
+      "step": 2285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017758895558513097,
+      "loss": 0.8572,
+      "step": 2286
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017756483588102673,
+      "loss": 0.4466,
+      "step": 2287
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001775407048443982,
+      "loss": 1.0642,
+      "step": 2288
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017751656247877107,
+      "loss": 0.5428,
+      "step": 2289
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001774924087876726,
+      "loss": 0.618,
+      "step": 2290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001774682437746318,
+      "loss": 1.0692,
+      "step": 2291
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017744406744317915,
+      "loss": 1.5751,
+      "step": 2292
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017741987979684706,
+      "loss": 1.3631,
+      "step": 2293
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017739568083916936,
+      "loss": 1.0381,
+      "step": 2294
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017737147057368162,
+      "loss": 1.1742,
+      "step": 2295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000177347249003921,
+      "loss": 1.1727,
+      "step": 2296
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017732301613342647,
+      "loss": 0.7824,
+      "step": 2297
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001772987719657385,
+      "loss": 1.2294,
+      "step": 2298
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017727451650439922,
+      "loss": 1.1671,
+      "step": 2299
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001772502497529525,
+      "loss": 0.9291,
+      "step": 2300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017722597171494376,
+      "loss": 0.9069,
+      "step": 2301
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017720168239392017,
+      "loss": 0.7598,
+      "step": 2302
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017717738179343044,
+      "loss": 1.1432,
+      "step": 2303
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017715306991702504,
+      "loss": 0.8034,
+      "step": 2304
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017712874676825596,
+      "loss": 0.6502,
+      "step": 2305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017710441235067696,
+      "loss": 0.8164,
+      "step": 2306
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017708006666784337,
+      "loss": 1.365,
+      "step": 2307
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017705570972331223,
+      "loss": 0.9938,
+      "step": 2308
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001770313415206421,
+      "loss": 1.29,
+      "step": 2309
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017700696206339337,
+      "loss": 1.3607,
+      "step": 2310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001769825713551279,
+      "loss": 1.3274,
+      "step": 2311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001769581693994093,
+      "loss": 0.5242,
+      "step": 2312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017693375619980278,
+      "loss": 1.4715,
+      "step": 2313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017690933175987518,
+      "loss": 1.4322,
+      "step": 2314
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017688489608319504,
+      "loss": 1.2622,
+      "step": 2315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017686044917333255,
+      "loss": 0.5721,
+      "step": 2316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017683599103385938,
+      "loss": 1.2182,
+      "step": 2317
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017681152166834903,
+      "loss": 0.9097,
+      "step": 2318
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001767870410803766,
+      "loss": 1.3293,
+      "step": 2319
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017676254927351877,
+      "loss": 0.7328,
+      "step": 2320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017673804625135384,
+      "loss": 0.4695,
+      "step": 2321
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017671353201746185,
+      "loss": 1.1404,
+      "step": 2322
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017668900657542445,
+      "loss": 0.931,
+      "step": 2323
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017666446992882484,
+      "loss": 0.9624,
+      "step": 2324
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017663992208124795,
+      "loss": 1.271,
+      "step": 2325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017661536303628032,
+      "loss": 1.1324,
+      "step": 2326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001765907927975101,
+      "loss": 1.1419,
+      "step": 2327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017656621136852714,
+      "loss": 0.7828,
+      "step": 2328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017654161875292284,
+      "loss": 1.0294,
+      "step": 2329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001765170149542903,
+      "loss": 1.0315,
+      "step": 2330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001764923999762242,
+      "loss": 1.2863,
+      "step": 2331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017646777382232092,
+      "loss": 0.98,
+      "step": 2332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001764431364961784,
+      "loss": 1.7245,
+      "step": 2333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017641848800139628,
+      "loss": 1.3501,
+      "step": 2334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001763938283415758,
+      "loss": 0.5255,
+      "step": 2335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017636915752031985,
+      "loss": 0.792,
+      "step": 2336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017634447554123287,
+      "loss": 1.2282,
+      "step": 2337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017631978240792106,
+      "loss": 1.2603,
+      "step": 2338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017629507812399212,
+      "loss": 1.2259,
+      "step": 2339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001762703626930555,
+      "loss": 1.3599,
+      "step": 2340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017624563611872224,
+      "loss": 1.2615,
+      "step": 2341
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001762208984046049,
+      "loss": 1.0033,
+      "step": 2342
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001761961495543178,
+      "loss": 0.9348,
+      "step": 2343
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017617138957147683,
+      "loss": 1.616,
+      "step": 2344
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017614661845969954,
+      "loss": 1.3801,
+      "step": 2345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001761218362226051,
+      "loss": 1.5282,
+      "step": 2346
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017609704286381428,
+      "loss": 1.7657,
+      "step": 2347
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017607223838694947,
+      "loss": 1.3318,
+      "step": 2348
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017604742279563471,
+      "loss": 1.1087,
+      "step": 2349
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017602259609349563,
+      "loss": 1.3707,
+      "step": 2350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017599775828415956,
+      "loss": 1.1532,
+      "step": 2351
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017597290937125534,
+      "loss": 1.1572,
+      "step": 2352
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017594804935841356,
+      "loss": 1.2846,
+      "step": 2353
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001759231782492663,
+      "loss": 0.6209,
+      "step": 2354
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001758982960474474,
+      "loss": 1.2745,
+      "step": 2355
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017587340275659213,
+      "loss": 0.4669,
+      "step": 2356
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017584849838033764,
+      "loss": 1.1476,
+      "step": 2357
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017582358292232245,
+      "loss": 1.2945,
+      "step": 2358
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017579865638618682,
+      "loss": 0.6533,
+      "step": 2359
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017577371877557264,
+      "loss": 0.9923,
+      "step": 2360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017574877009412344,
+      "loss": 0.668,
+      "step": 2361
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001757238103454842,
+      "loss": 0.6126,
+      "step": 2362
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017569883953330172,
+      "loss": 0.5534,
+      "step": 2363
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017567385766122432,
+      "loss": 1.2194,
+      "step": 2364
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017564886473290194,
+      "loss": 1.3125,
+      "step": 2365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017562386075198615,
+      "loss": 1.2786,
+      "step": 2366
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001755988457221301,
+      "loss": 1.4277,
+      "step": 2367
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017557381964698865,
+      "loss": 1.4419,
+      "step": 2368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017554878253021815,
+      "loss": 1.3955,
+      "step": 2369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017552373437547663,
+      "loss": 1.3037,
+      "step": 2370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017549867518642373,
+      "loss": 1.4962,
+      "step": 2371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001754736049667207,
+      "loss": 0.738,
+      "step": 2372
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017544852372003037,
+      "loss": 1.2311,
+      "step": 2373
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017542343145001725,
+      "loss": 0.9691,
+      "step": 2374
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017539832816034737,
+      "loss": 1.2855,
+      "step": 2375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017537321385468848,
+      "loss": 0.7008,
+      "step": 2376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017534808853670984,
+      "loss": 1.8156,
+      "step": 2377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017532295221008232,
+      "loss": 0.8122,
+      "step": 2378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017529780487847852,
+      "loss": 0.8804,
+      "step": 2379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017527264654557248,
+      "loss": 0.5178,
+      "step": 2380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017524747721503998,
+      "loss": 1.2602,
+      "step": 2381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017522229689055836,
+      "loss": 1.1719,
+      "step": 2382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017519710557580654,
+      "loss": 1.0537,
+      "step": 2383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017517190327446505,
+      "loss": 0.9046,
+      "step": 2384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001751466899902161,
+      "loss": 0.986,
+      "step": 2385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001751214657267434,
+      "loss": 1.2687,
+      "step": 2386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017509623048773233,
+      "loss": 0.9957,
+      "step": 2387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001750709842768699,
+      "loss": 1.214,
+      "step": 2388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017504572709784458,
+      "loss": 1.074,
+      "step": 2389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017502045895434666,
+      "loss": 0.8303,
+      "step": 2390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017499517985006784,
+      "loss": 1.4204,
+      "step": 2391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001749698897887015,
+      "loss": 0.7027,
+      "step": 2392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017494458877394262,
+      "loss": 0.9905,
+      "step": 2393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001749192768094878,
+      "loss": 0.565,
+      "step": 2394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001748939538990352,
+      "loss": 1.0568,
+      "step": 2395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017486862004628456,
+      "loss": 1.2598,
+      "step": 2396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017484327525493733,
+      "loss": 0.5609,
+      "step": 2397
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017481791952869645,
+      "loss": 1.5286,
+      "step": 2398
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017479255287126643,
+      "loss": 0.976,
+      "step": 2399
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017476717528635354,
+      "loss": 1.1095,
+      "step": 2400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017474178677766546,
+      "loss": 0.804,
+      "step": 2401
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017471638734891163,
+      "loss": 1.4177,
+      "step": 2402
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017469097700380292,
+      "loss": 1.506,
+      "step": 2403
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017466555574605192,
+      "loss": 0.4967,
+      "step": 2404
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001746401235793728,
+      "loss": 1.2176,
+      "step": 2405
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017461468050748128,
+      "loss": 1.5432,
+      "step": 2406
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017458922653409466,
+      "loss": 1.1638,
+      "step": 2407
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001745637616629319,
+      "loss": 0.9754,
+      "step": 2408
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001745382858977135,
+      "loss": 1.3511,
+      "step": 2409
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017451279924216156,
+      "loss": 0.5964,
+      "step": 2410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001744873016999998,
+      "loss": 0.9091,
+      "step": 2411
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017446179327495353,
+      "loss": 1.0446,
+      "step": 2412
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001744362739707496,
+      "loss": 1.1593,
+      "step": 2413
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017441074379111646,
+      "loss": 1.113,
+      "step": 2414
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001743852027397842,
+      "loss": 0.8844,
+      "step": 2415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017435965082048446,
+      "loss": 0.6475,
+      "step": 2416
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017433408803695048,
+      "loss": 1.0691,
+      "step": 2417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001743085143929171,
+      "loss": 0.6544,
+      "step": 2418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017428292989212066,
+      "loss": 0.9643,
+      "step": 2419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017425733453829923,
+      "loss": 1.3988,
+      "step": 2420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017423172833519235,
+      "loss": 0.5918,
+      "step": 2421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017420611128654117,
+      "loss": 1.1023,
+      "step": 2422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017418048339608853,
+      "loss": 0.8635,
+      "step": 2423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017415484466757865,
+      "loss": 1.0596,
+      "step": 2424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017412919510475754,
+      "loss": 0.7798,
+      "step": 2425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017410353471137264,
+      "loss": 0.4879,
+      "step": 2426
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017407786349117306,
+      "loss": 0.5942,
+      "step": 2427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001740521814479094,
+      "loss": 1.3576,
+      "step": 2428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017402648858533403,
+      "loss": 1.0367,
+      "step": 2429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001740007849072007,
+      "loss": 1.6041,
+      "step": 2430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017397507041726483,
+      "loss": 0.6402,
+      "step": 2431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017394934511928342,
+      "loss": 1.3659,
+      "step": 2432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017392360901701502,
+      "loss": 1.2295,
+      "step": 2433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017389786211421975,
+      "loss": 1.9333,
+      "step": 2434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001738721044146594,
+      "loss": 1.3472,
+      "step": 2435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001738463359220972,
+      "loss": 0.4539,
+      "step": 2436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017382055664029808,
+      "loss": 1.444,
+      "step": 2437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017379476657302845,
+      "loss": 1.6233,
+      "step": 2438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017376896572405637,
+      "loss": 0.7911,
+      "step": 2439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017374315409715143,
+      "loss": 1.5013,
+      "step": 2440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017371733169608481,
+      "loss": 1.3147,
+      "step": 2441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017369149852462928,
+      "loss": 1.3261,
+      "step": 2442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017366565458655915,
+      "loss": 0.8035,
+      "step": 2443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017363979988565032,
+      "loss": 1.1072,
+      "step": 2444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017361393442568028,
+      "loss": 1.2357,
+      "step": 2445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017358805821042803,
+      "loss": 1.0814,
+      "step": 2446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017356217124367423,
+      "loss": 1.3497,
+      "step": 2447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017353627352920107,
+      "loss": 1.3294,
+      "step": 2448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001735103650707923,
+      "loss": 1.34,
+      "step": 2449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017348444587223319,
+      "loss": 1.0098,
+      "step": 2450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017345851593731067,
+      "loss": 1.0574,
+      "step": 2451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017343257526981326,
+      "loss": 1.382,
+      "step": 2452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017340662387353096,
+      "loss": 1.0065,
+      "step": 2453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001733806617522553,
+      "loss": 0.5167,
+      "step": 2454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017335468890977959,
+      "loss": 0.8673,
+      "step": 2455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001733287053498984,
+      "loss": 1.3703,
+      "step": 2456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017330271107640814,
+      "loss": 1.2526,
+      "step": 2457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017327670609310663,
+      "loss": 1.3403,
+      "step": 2458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017325069040379334,
+      "loss": 1.3611,
+      "step": 2459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017322466401226922,
+      "loss": 1.2989,
+      "step": 2460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001731986269223368,
+      "loss": 0.9793,
+      "step": 2461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017317257913780033,
+      "loss": 1.2617,
+      "step": 2462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017314652066246534,
+      "loss": 0.6901,
+      "step": 2463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017312045150013917,
+      "loss": 0.6977,
+      "step": 2464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001730943716546306,
+      "loss": 1.2804,
+      "step": 2465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017306828112974992,
+      "loss": 0.3892,
+      "step": 2466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001730421799293092,
+      "loss": 0.6593,
+      "step": 2467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017301606805712185,
+      "loss": 0.7475,
+      "step": 2468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017298994551700293,
+      "loss": 0.6941,
+      "step": 2469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000172963812312769,
+      "loss": 1.1425,
+      "step": 2470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017293766844823826,
+      "loss": 1.0232,
+      "step": 2471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017291151392723046,
+      "loss": 1.3042,
+      "step": 2472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017288534875356683,
+      "loss": 0.8919,
+      "step": 2473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017285917293107024,
+      "loss": 1.5746,
+      "step": 2474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017283298646356504,
+      "loss": 0.523,
+      "step": 2475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001728067893548772,
+      "loss": 2.0262,
+      "step": 2476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017278058160883424,
+      "loss": 0.46,
+      "step": 2477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017275436322926514,
+      "loss": 1.3499,
+      "step": 2478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001727281342200006,
+      "loss": 1.1831,
+      "step": 2479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017270189458487274,
+      "loss": 1.0826,
+      "step": 2480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017267564432771524,
+      "loss": 1.4018,
+      "step": 2481
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001726493834523634,
+      "loss": 1.0105,
+      "step": 2482
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017262311196265402,
+      "loss": 0.7153,
+      "step": 2483
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001725968298624255,
+      "loss": 1.3014,
+      "step": 2484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017257053715551772,
+      "loss": 0.6923,
+      "step": 2485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017254423384577215,
+      "loss": 0.6969,
+      "step": 2486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017251791993703184,
+      "loss": 1.2688,
+      "step": 2487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017249159543314134,
+      "loss": 1.5634,
+      "step": 2488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001724652603379467,
+      "loss": 1.3213,
+      "step": 2489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001724389146552957,
+      "loss": 0.9253,
+      "step": 2490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017241255838903742,
+      "loss": 1.3721,
+      "step": 2491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017238619154302272,
+      "loss": 1.3204,
+      "step": 2492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017235981412110383,
+      "loss": 0.7019,
+      "step": 2493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001723334261271346,
+      "loss": 1.145,
+      "step": 2494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017230702756497045,
+      "loss": 1.2502,
+      "step": 2495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001722806184384683,
+      "loss": 1.1397,
+      "step": 2496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017225419875148662,
+      "loss": 1.3711,
+      "step": 2497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017222776850788542,
+      "loss": 1.3839,
+      "step": 2498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001722013277115263,
+      "loss": 1.2675,
+      "step": 2499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001721748763662723,
+      "loss": 1.1359,
+      "step": 2500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017214841447598812,
+      "loss": 1.5795,
+      "step": 2501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017212194204453996,
+      "loss": 1.4188,
+      "step": 2502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017209545907579546,
+      "loss": 1.1122,
+      "step": 2503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017206896557362395,
+      "loss": 0.6645,
+      "step": 2504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017204246154189622,
+      "loss": 1.4012,
+      "step": 2505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017201594698448462,
+      "loss": 1.4487,
+      "step": 2506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017198942190526303,
+      "loss": 0.9856,
+      "step": 2507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017196288630810685,
+      "loss": 1.2836,
+      "step": 2508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000171936340196893,
+      "loss": 0.7893,
+      "step": 2509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017190978357550007,
+      "loss": 1.4053,
+      "step": 2510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000171883216447808,
+      "loss": 1.2591,
+      "step": 2511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001718566388176984,
+      "loss": 1.2971,
+      "step": 2512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017183005068905437,
+      "loss": 0.7482,
+      "step": 2513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017180345206576044,
+      "loss": 1.5682,
+      "step": 2514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017177684295170288,
+      "loss": 1.0789,
+      "step": 2515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017175022335076938,
+      "loss": 1.4559,
+      "step": 2516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001717235932668491,
+      "loss": 0.8443,
+      "step": 2517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017169695270383286,
+      "loss": 1.3074,
+      "step": 2518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017167030166561291,
+      "loss": 0.7301,
+      "step": 2519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001716436401560831,
+      "loss": 0.7288,
+      "step": 2520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017161696817913878,
+      "loss": 1.3033,
+      "step": 2521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017159028573867678,
+      "loss": 0.9623,
+      "step": 2522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017156359283859558,
+      "loss": 1.1123,
+      "step": 2523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017153688948279506,
+      "loss": 1.4461,
+      "step": 2524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001715101756751767,
+      "loss": 1.0468,
+      "step": 2525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001714834514196435,
+      "loss": 1.1708,
+      "step": 2526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001714567167201,
+      "loss": 0.9887,
+      "step": 2527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001714299715804522,
+      "loss": 1.2259,
+      "step": 2528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017140321600460769,
+      "loss": 1.3094,
+      "step": 2529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017137644999647557,
+      "loss": 1.2847,
+      "step": 2530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017134967355996646,
+      "loss": 1.4404,
+      "step": 2531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017132288669899248,
+      "loss": 1.3528,
+      "step": 2532
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017129608941746728,
+      "loss": 1.2731,
+      "step": 2533
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001712692817193061,
+      "loss": 1.3602,
+      "step": 2534
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017124246360842563,
+      "loss": 1.6788,
+      "step": 2535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017121563508874405,
+      "loss": 1.3061,
+      "step": 2536
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001711887961641812,
+      "loss": 1.3015,
+      "step": 2537
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017116194683865828,
+      "loss": 0.594,
+      "step": 2538
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001711350871160981,
+      "loss": 1.1689,
+      "step": 2539
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017110821700042498,
+      "loss": 1.4714,
+      "step": 2540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017108133649556474,
+      "loss": 1.2921,
+      "step": 2541
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017105444560544473,
+      "loss": 1.1112,
+      "step": 2542
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017102754433399384,
+      "loss": 1.1277,
+      "step": 2543
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001710006326851424,
+      "loss": 1.0527,
+      "step": 2544
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017097371066282234,
+      "loss": 1.2309,
+      "step": 2545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017094677827096707,
+      "loss": 1.0073,
+      "step": 2546
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017091983551351147,
+      "loss": 1.4706,
+      "step": 2547
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017089288239439205,
+      "loss": 0.7948,
+      "step": 2548
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017086591891754674,
+      "loss": 0.9228,
+      "step": 2549
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000170838945086915,
+      "loss": 1.3324,
+      "step": 2550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017081196090643782,
+      "loss": 1.2059,
+      "step": 2551
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001707849663800577,
+      "loss": 0.7072,
+      "step": 2552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001707579615117186,
+      "loss": 1.4247,
+      "step": 2553
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001707309463053661,
+      "loss": 1.0414,
+      "step": 2554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017070392076494725,
+      "loss": 0.8037,
+      "step": 2555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017067688489441046,
+      "loss": 0.6961,
+      "step": 2556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001706498386977059,
+      "loss": 0.6884,
+      "step": 2557
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017062278217878503,
+      "loss": 0.4684,
+      "step": 2558
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017059571534160098,
+      "loss": 1.4313,
+      "step": 2559
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001705686381901083,
+      "loss": 1.2401,
+      "step": 2560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017054155072826306,
+      "loss": 0.8977,
+      "step": 2561
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001705144529600229,
+      "loss": 0.9089,
+      "step": 2562
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001704873448893468,
+      "loss": 1.1616,
+      "step": 2563
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017046022652019541,
+      "loss": 0.638,
+      "step": 2564
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017043309785653087,
+      "loss": 1.1589,
+      "step": 2565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017040595890231675,
+      "loss": 1.3224,
+      "step": 2566
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017037880966151812,
+      "loss": 0.8922,
+      "step": 2567
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017035165013810165,
+      "loss": 1.3926,
+      "step": 2568
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001703244803360354,
+      "loss": 1.213,
+      "step": 2569
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017029730025928906,
+      "loss": 1.2441,
+      "step": 2570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017027010991183363,
+      "loss": 1.366,
+      "step": 2571
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017024290929764182,
+      "loss": 1.5841,
+      "step": 2572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017021569842068772,
+      "loss": 1.023,
+      "step": 2573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001701884772849469,
+      "loss": 1.5082,
+      "step": 2574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017016124589439659,
+      "loss": 1.4148,
+      "step": 2575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017013400425301527,
+      "loss": 1.1727,
+      "step": 2576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001701067523647831,
+      "loss": 1.2282,
+      "step": 2577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017007949023368173,
+      "loss": 1.1228,
+      "step": 2578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001700522178636942,
+      "loss": 1.3193,
+      "step": 2579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017002493525880515,
+      "loss": 1.3528,
+      "step": 2580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016999764242300064,
+      "loss": 1.3924,
+      "step": 2581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001699703393602683,
+      "loss": 0.5589,
+      "step": 2582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016994302607459716,
+      "loss": 1.342,
+      "step": 2583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001699157025699779,
+      "loss": 1.0454,
+      "step": 2584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016988836885040243,
+      "loss": 0.5356,
+      "step": 2585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016986102491986445,
+      "loss": 1.0996,
+      "step": 2586
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016983367078235898,
+      "loss": 1.1642,
+      "step": 2587
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016980630644188252,
+      "loss": 1.5725,
+      "step": 2588
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016977893190243316,
+      "loss": 1.1015,
+      "step": 2589
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016975154716801044,
+      "loss": 0.8484,
+      "step": 2590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001697241522426153,
+      "loss": 1.3508,
+      "step": 2591
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001696967471302503,
+      "loss": 1.2285,
+      "step": 2592
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001696693318349194,
+      "loss": 0.6495,
+      "step": 2593
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016964190636062812,
+      "loss": 0.8397,
+      "step": 2594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016961447071138345,
+      "loss": 0.5537,
+      "step": 2595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016958702489119375,
+      "loss": 0.5498,
+      "step": 2596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016955956890406904,
+      "loss": 1.0779,
+      "step": 2597
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016953210275402075,
+      "loss": 1.0862,
+      "step": 2598
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016950462644506174,
+      "loss": 0.8698,
+      "step": 2599
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001694771399812064,
+      "loss": 1.1743,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001694496433664707,
+      "loss": 0.5861,
+      "step": 2601
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001694221366048719,
+      "loss": 1.2359,
+      "step": 2602
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016939461970042892,
+      "loss": 1.4512,
+      "step": 2603
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016936709265716207,
+      "loss": 1.2774,
+      "step": 2604
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001693395554790931,
+      "loss": 1.128,
+      "step": 2605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016931200817024537,
+      "loss": 1.3847,
+      "step": 2606
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001692844507346436,
+      "loss": 1.0816,
+      "step": 2607
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016925688317631407,
+      "loss": 0.681,
+      "step": 2608
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001692293054992845,
+      "loss": 1.103,
+      "step": 2609
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016920171770758407,
+      "loss": 1.0333,
+      "step": 2610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016917411980524352,
+      "loss": 1.1158,
+      "step": 2611
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016914651179629497,
+      "loss": 1.3022,
+      "step": 2612
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016911889368477202,
+      "loss": 1.3259,
+      "step": 2613
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016909126547470988,
+      "loss": 0.7217,
+      "step": 2614
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016906362717014509,
+      "loss": 1.051,
+      "step": 2615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016903597877511567,
+      "loss": 0.6622,
+      "step": 2616
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016900832029366125,
+      "loss": 1.0223,
+      "step": 2617
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016898065172982275,
+      "loss": 0.7872,
+      "step": 2618
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001689529730876427,
+      "loss": 0.591,
+      "step": 2619
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016892528437116507,
+      "loss": 1.2716,
+      "step": 2620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016889758558443526,
+      "loss": 1.4028,
+      "step": 2621
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016886987673150018,
+      "loss": 1.1283,
+      "step": 2622
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001688421578164082,
+      "loss": 0.7595,
+      "step": 2623
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001688144288432092,
+      "loss": 1.5434,
+      "step": 2624
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016878668981595442,
+      "loss": 1.7384,
+      "step": 2625
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016875894073869669,
+      "loss": 1.4912,
+      "step": 2626
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016873118161549026,
+      "loss": 1.0707,
+      "step": 2627
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016870341245039082,
+      "loss": 1.1327,
+      "step": 2628
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016867563324745555,
+      "loss": 1.1789,
+      "step": 2629
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016864784401074312,
+      "loss": 1.0285,
+      "step": 2630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016862004474431364,
+      "loss": 0.5507,
+      "step": 2631
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001685922354522287,
+      "loss": 1.245,
+      "step": 2632
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001685644161385513,
+      "loss": 1.2674,
+      "step": 2633
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000168536586807346,
+      "loss": 0.6165,
+      "step": 2634
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016850874746267874,
+      "loss": 0.7604,
+      "step": 2635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000168480898108617,
+      "loss": 1.3585,
+      "step": 2636
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016845303874922968,
+      "loss": 0.9113,
+      "step": 2637
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016842516938858707,
+      "loss": 1.0454,
+      "step": 2638
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016839729003076108,
+      "loss": 0.6466,
+      "step": 2639
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016836940067982493,
+      "loss": 1.3239,
+      "step": 2640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016834150133985336,
+      "loss": 1.343,
+      "step": 2641
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001683135920149226,
+      "loss": 0.8157,
+      "step": 2642
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016828567270911032,
+      "loss": 1.7149,
+      "step": 2643
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016825774342649563,
+      "loss": 1.3485,
+      "step": 2644
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016822980417115907,
+      "loss": 0.8508,
+      "step": 2645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016820185494718275,
+      "loss": 0.9008,
+      "step": 2646
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016817389575865007,
+      "loss": 1.2977,
+      "step": 2647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016814592660964602,
+      "loss": 1.2585,
+      "step": 2648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016811794750425704,
+      "loss": 0.5599,
+      "step": 2649
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016808995844657095,
+      "loss": 1.6437,
+      "step": 2650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016806195944067703,
+      "loss": 1.2296,
+      "step": 2651
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001680339504906661,
+      "loss": 1.0741,
+      "step": 2652
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016800593160063034,
+      "loss": 1.2947,
+      "step": 2653
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001679779027746634,
+      "loss": 1.3679,
+      "step": 2654
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016794986401686045,
+      "loss": 0.8104,
+      "step": 2655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016792181533131805,
+      "loss": 1.2234,
+      "step": 2656
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016789375672213421,
+      "loss": 1.2581,
+      "step": 2657
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001678656881934084,
+      "loss": 1.4108,
+      "step": 2658
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016783760974924158,
+      "loss": 1.0044,
+      "step": 2659
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000167809521393736,
+      "loss": 1.4495,
+      "step": 2660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016778142313099564,
+      "loss": 1.4106,
+      "step": 2661
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016775331496512568,
+      "loss": 0.5837,
+      "step": 2662
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001677251969002328,
+      "loss": 1.3092,
+      "step": 2663
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016769706894042522,
+      "loss": 1.6529,
+      "step": 2664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016766893108981253,
+      "loss": 1.1667,
+      "step": 2665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001676407833525058,
+      "loss": 0.7094,
+      "step": 2666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016761262573261748,
+      "loss": 1.3692,
+      "step": 2667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001675844582342615,
+      "loss": 1.0512,
+      "step": 2668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016755628086155327,
+      "loss": 1.5051,
+      "step": 2669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016752809361860965,
+      "loss": 1.3288,
+      "step": 2670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016749989650954883,
+      "loss": 0.5705,
+      "step": 2671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016747168953849055,
+      "loss": 1.2204,
+      "step": 2672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016744347270955602,
+      "loss": 1.0034,
+      "step": 2673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016741524602686768,
+      "loss": 0.8705,
+      "step": 2674
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016738700949454972,
+      "loss": 0.8302,
+      "step": 2675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001673587631167275,
+      "loss": 1.2003,
+      "step": 2676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016733050689752798,
+      "loss": 1.5917,
+      "step": 2677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016730224084107948,
+      "loss": 1.4096,
+      "step": 2678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016727396495151177,
+      "loss": 1.4592,
+      "step": 2679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001672456792329561,
+      "loss": 1.3641,
+      "step": 2680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001672173836895451,
+      "loss": 1.0681,
+      "step": 2681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016718907832541288,
+      "loss": 1.3293,
+      "step": 2682
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016716076314469494,
+      "loss": 1.3539,
+      "step": 2683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016713243815152823,
+      "loss": 0.6551,
+      "step": 2684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001671041033500512,
+      "loss": 1.0662,
+      "step": 2685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001670757587444036,
+      "loss": 1.0992,
+      "step": 2686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016704740433872673,
+      "loss": 0.8741,
+      "step": 2687
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016701904013716328,
+      "loss": 1.1767,
+      "step": 2688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016699066614385733,
+      "loss": 1.2228,
+      "step": 2689
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016696228236295452,
+      "loss": 1.1887,
+      "step": 2690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001669338887986017,
+      "loss": 1.2585,
+      "step": 2691
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001669054854549474,
+      "loss": 0.7438,
+      "step": 2692
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016687707233614137,
+      "loss": 0.6102,
+      "step": 2693
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016684864944633497,
+      "loss": 0.6568,
+      "step": 2694
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001668202167896808,
+      "loss": 1.3427,
+      "step": 2695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016679177437033306,
+      "loss": 0.5418,
+      "step": 2696
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016676332219244723,
+      "loss": 1.2185,
+      "step": 2697
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016673486026018032,
+      "loss": 1.3278,
+      "step": 2698
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016670638857769078,
+      "loss": 0.6209,
+      "step": 2699
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001666779071491383,
+      "loss": 1.0237,
+      "step": 2700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016664941597868427,
+      "loss": 0.9321,
+      "step": 2701
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016662091507049125,
+      "loss": 0.5172,
+      "step": 2702
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016659240442872341,
+      "loss": 1.6986,
+      "step": 2703
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016656388405754621,
+      "loss": 1.1896,
+      "step": 2704
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016653535396112665,
+      "loss": 1.374,
+      "step": 2705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016650681414363304,
+      "loss": 1.1249,
+      "step": 2706
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016647826460923517,
+      "loss": 1.3356,
+      "step": 2707
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016644970536210422,
+      "loss": 1.6772,
+      "step": 2708
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016642113640641287,
+      "loss": 1.307,
+      "step": 2709
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016639255774633505,
+      "loss": 1.646,
+      "step": 2710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001663639693860463,
+      "loss": 1.4382,
+      "step": 2711
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001663353713297235,
+      "loss": 1.1366,
+      "step": 2712
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016630676358154488,
+      "loss": 1.4491,
+      "step": 2713
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016627814614569015,
+      "loss": 1.273,
+      "step": 2714
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016624951902634045,
+      "loss": 1.4943,
+      "step": 2715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001662208822276783,
+      "loss": 1.0469,
+      "step": 2716
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016619223575388767,
+      "loss": 1.2965,
+      "step": 2717
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001661635796091539,
+      "loss": 1.0316,
+      "step": 2718
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016613491379766378,
+      "loss": 0.6418,
+      "step": 2719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001661062383236055,
+      "loss": 1.1424,
+      "step": 2720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001660775531911686,
+      "loss": 1.1674,
+      "step": 2721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016604885840454416,
+      "loss": 1.2546,
+      "step": 2722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016602015396792458,
+      "loss": 1.062,
+      "step": 2723
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016599143988550366,
+      "loss": 0.9288,
+      "step": 2724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016596271616147674,
+      "loss": 0.7662,
+      "step": 2725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016593398280004032,
+      "loss": 1.2901,
+      "step": 2726
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016590523980539255,
+      "loss": 0.8343,
+      "step": 2727
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016587648718173289,
+      "loss": 1.1941,
+      "step": 2728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016584772493326218,
+      "loss": 1.21,
+      "step": 2729
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016581895306418267,
+      "loss": 1.2224,
+      "step": 2730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001657901715786981,
+      "loss": 1.1472,
+      "step": 2731
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016576138048101356,
+      "loss": 1.2503,
+      "step": 2732
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001657325797753355,
+      "loss": 0.6391,
+      "step": 2733
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016570376946587186,
+      "loss": 0.775,
+      "step": 2734
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016567494955683186,
+      "loss": 1.0932,
+      "step": 2735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016564612005242629,
+      "loss": 1.1927,
+      "step": 2736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001656172809568672,
+      "loss": 1.1955,
+      "step": 2737
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016558843227436814,
+      "loss": 1.5514,
+      "step": 2738
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016555957400914397,
+      "loss": 1.7536,
+      "step": 2739
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000165530706165411,
+      "loss": 1.077,
+      "step": 2740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016550182874738693,
+      "loss": 0.8795,
+      "step": 2741
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001654729417592909,
+      "loss": 1.1036,
+      "step": 2742
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016544404520534334,
+      "loss": 0.8601,
+      "step": 2743
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016541513908976624,
+      "loss": 1.4329,
+      "step": 2744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016538622341678288,
+      "loss": 0.5446,
+      "step": 2745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016535729819061788,
+      "loss": 0.6225,
+      "step": 2746
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001653283634154974,
+      "loss": 1.1096,
+      "step": 2747
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016529941909564885,
+      "loss": 1.2323,
+      "step": 2748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001652704652353012,
+      "loss": 0.9988,
+      "step": 2749
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016524150183868465,
+      "loss": 1.1629,
+      "step": 2750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001652125289100309,
+      "loss": 1.0892,
+      "step": 2751
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000165183546453573,
+      "loss": 1.5294,
+      "step": 2752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001651545544735454,
+      "loss": 0.8825,
+      "step": 2753
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016512555297418393,
+      "loss": 1.5373,
+      "step": 2754
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016509654195972581,
+      "loss": 0.5373,
+      "step": 2755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016506752143440973,
+      "loss": 1.2617,
+      "step": 2756
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016503849140247562,
+      "loss": 1.4852,
+      "step": 2757
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016500945186816493,
+      "loss": 0.6156,
+      "step": 2758
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016498040283572044,
+      "loss": 0.5307,
+      "step": 2759
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001649513443093863,
+      "loss": 1.1819,
+      "step": 2760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016492227629340812,
+      "loss": 1.5644,
+      "step": 2761
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001648931987920328,
+      "loss": 1.0383,
+      "step": 2762
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001648641118095087,
+      "loss": 1.5136,
+      "step": 2763
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016483501535008555,
+      "loss": 0.6576,
+      "step": 2764
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016480590941801446,
+      "loss": 1.4193,
+      "step": 2765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001647767940175479,
+      "loss": 1.249,
+      "step": 2766
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016474766915293974,
+      "loss": 0.9292,
+      "step": 2767
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016471853482844523,
+      "loss": 1.1117,
+      "step": 2768
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016468939104832107,
+      "loss": 0.9803,
+      "step": 2769
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001646602378168252,
+      "loss": 1.06,
+      "step": 2770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016463107513821706,
+      "loss": 0.9876,
+      "step": 2771
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016460190301675743,
+      "loss": 1.2858,
+      "step": 2772
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016457272145670848,
+      "loss": 1.1562,
+      "step": 2773
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016454353046233374,
+      "loss": 1.4151,
+      "step": 2774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001645143300378981,
+      "loss": 0.9276,
+      "step": 2775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016448512018766786,
+      "loss": 0.7836,
+      "step": 2776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016445590091591073,
+      "loss": 1.3794,
+      "step": 2777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016442667222689577,
+      "loss": 1.5921,
+      "step": 2778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016439743412489336,
+      "loss": 1.0819,
+      "step": 2779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001643681866141753,
+      "loss": 0.9342,
+      "step": 2780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001643389296990148,
+      "loss": 0.9389,
+      "step": 2781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016430966338368638,
+      "loss": 0.5523,
+      "step": 2782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016428038767246598,
+      "loss": 0.6381,
+      "step": 2783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001642511025696309,
+      "loss": 0.8681,
+      "step": 2784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016422180807945978,
+      "loss": 1.0162,
+      "step": 2785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016419250420623266,
+      "loss": 1.3857,
+      "step": 2786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000164163190954231,
+      "loss": 1.3568,
+      "step": 2787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016413386832773752,
+      "loss": 1.4676,
+      "step": 2788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016410453633103643,
+      "loss": 1.0622,
+      "step": 2789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016407519496841318,
+      "loss": 1.1474,
+      "step": 2790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016404584424415468,
+      "loss": 1.466,
+      "step": 2791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016401648416254923,
+      "loss": 1.3983,
+      "step": 2792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016398711472788638,
+      "loss": 1.0035,
+      "step": 2793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001639577359444572,
+      "loss": 1.2167,
+      "step": 2794
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016392834781655397,
+      "loss": 1.2692,
+      "step": 2795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016389895034847046,
+      "loss": 1.3289,
+      "step": 2796
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016386954354450172,
+      "loss": 0.8975,
+      "step": 2797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016384012740894424,
+      "loss": 1.2202,
+      "step": 2798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016381070194609575,
+      "loss": 1.5919,
+      "step": 2799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016378126716025553,
+      "loss": 1.4456,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016375182305572406,
+      "loss": 1.2902,
+      "step": 2801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016372236963680322,
+      "loss": 0.9632,
+      "step": 2802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016369290690779632,
+      "loss": 0.5061,
+      "step": 2803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016366343487300797,
+      "loss": 1.3191,
+      "step": 2804
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001636339535367441,
+      "loss": 0.8973,
+      "step": 2805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016360446290331212,
+      "loss": 1.0332,
+      "step": 2806
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016357496297702065,
+      "loss": 0.6581,
+      "step": 2807
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016354545376217976,
+      "loss": 0.6799,
+      "step": 2808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016351593526310092,
+      "loss": 1.4566,
+      "step": 2809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016348640748409685,
+      "loss": 1.1186,
+      "step": 2810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016345687042948167,
+      "loss": 0.9118,
+      "step": 2811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016342732410357092,
+      "loss": 1.4659,
+      "step": 2812
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016339776851068133,
+      "loss": 1.3496,
+      "step": 2813
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016336820365513117,
+      "loss": 1.1805,
+      "step": 2814
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016333862954123993,
+      "loss": 1.3646,
+      "step": 2815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016330904617332855,
+      "loss": 0.9162,
+      "step": 2816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016327945355571925,
+      "loss": 1.475,
+      "step": 2817
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016324985169273562,
+      "loss": 1.3266,
+      "step": 2818
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016322024058870258,
+      "loss": 1.4433,
+      "step": 2819
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016319062024794652,
+      "loss": 0.7048,
+      "step": 2820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016316099067479497,
+      "loss": 0.9028,
+      "step": 2821
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016313135187357702,
+      "loss": 1.3014,
+      "step": 2822
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000163101703848623,
+      "loss": 0.7114,
+      "step": 2823
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016307204660426453,
+      "loss": 1.303,
+      "step": 2824
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016304238014483473,
+      "loss": 1.4008,
+      "step": 2825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016301270447466795,
+      "loss": 1.3013,
+      "step": 2826
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016298301959809993,
+      "loss": 0.5372,
+      "step": 2827
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016295332551946774,
+      "loss": 1.6065,
+      "step": 2828
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016292362224310978,
+      "loss": 0.7911,
+      "step": 2829
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016289390977336586,
+      "loss": 1.0997,
+      "step": 2830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016286418811457706,
+      "loss": 1.2201,
+      "step": 2831
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016283445727108582,
+      "loss": 1.2122,
+      "step": 2832
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016280471724723595,
+      "loss": 0.8672,
+      "step": 2833
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001627749680473726,
+      "loss": 0.7349,
+      "step": 2834
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016274520967584217,
+      "loss": 1.3736,
+      "step": 2835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016271544213699256,
+      "loss": 1.4817,
+      "step": 2836
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001626856654351729,
+      "loss": 1.0919,
+      "step": 2837
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016265587957473368,
+      "loss": 1.3196,
+      "step": 2838
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001626260845600267,
+      "loss": 1.5712,
+      "step": 2839
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016259628039540512,
+      "loss": 1.251,
+      "step": 2840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016256646708522353,
+      "loss": 1.4586,
+      "step": 2841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016253664463383767,
+      "loss": 0.7145,
+      "step": 2842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016250681304560478,
+      "loss": 1.0035,
+      "step": 2843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016247697232488334,
+      "loss": 0.4776,
+      "step": 2844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016244712247603325,
+      "loss": 0.8244,
+      "step": 2845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016241726350341558,
+      "loss": 0.9405,
+      "step": 2846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016238739541139295,
+      "loss": 1.3279,
+      "step": 2847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016235751820432913,
+      "loss": 1.5384,
+      "step": 2848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016232763188658934,
+      "loss": 0.8446,
+      "step": 2849
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016229773646254006,
+      "loss": 0.85,
+      "step": 2850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001622678319365492,
+      "loss": 1.2955,
+      "step": 2851
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016223791831298582,
+      "loss": 1.6576,
+      "step": 2852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001622079955962205,
+      "loss": 1.2233,
+      "step": 2853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000162178063790625,
+      "loss": 1.4217,
+      "step": 2854
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016214812290057254,
+      "loss": 0.9839,
+      "step": 2855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016211817293043757,
+      "loss": 0.8275,
+      "step": 2856
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016208821388459584,
+      "loss": 1.4969,
+      "step": 2857
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016205824576742456,
+      "loss": 1.3117,
+      "step": 2858
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016202826858330218,
+      "loss": 1.4274,
+      "step": 2859
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016199828233660844,
+      "loss": 1.1045,
+      "step": 2860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001619682870317245,
+      "loss": 0.6756,
+      "step": 2861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016193828267303277,
+      "loss": 0.8502,
+      "step": 2862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000161908269264917,
+      "loss": 0.6668,
+      "step": 2863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016187824681176228,
+      "loss": 1.3094,
+      "step": 2864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016184821531795498,
+      "loss": 1.5859,
+      "step": 2865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001618181747878828,
+      "loss": 1.1809,
+      "step": 2866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016178812522593485,
+      "loss": 1.3051,
+      "step": 2867
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001617580666365014,
+      "loss": 0.5233,
+      "step": 2868
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016172799902397422,
+      "loss": 0.6164,
+      "step": 2869
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016169792239274626,
+      "loss": 0.4375,
+      "step": 2870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001616678367472118,
+      "loss": 1.4147,
+      "step": 2871
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016163774209176652,
+      "loss": 0.8915,
+      "step": 2872
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016160763843080735,
+      "loss": 1.247,
+      "step": 2873
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016157752576873252,
+      "loss": 1.3163,
+      "step": 2874
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016154740410994167,
+      "loss": 1.1957,
+      "step": 2875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016151727345883567,
+      "loss": 1.2975,
+      "step": 2876
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016148713381981667,
+      "loss": 1.447,
+      "step": 2877
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016145698519728826,
+      "loss": 1.0976,
+      "step": 2878
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016142682759565525,
+      "loss": 1.2313,
+      "step": 2879
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001613966610193238,
+      "loss": 1.2612,
+      "step": 2880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001613664854727013,
+      "loss": 1.2263,
+      "step": 2881
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016133630096019653,
+      "loss": 1.152,
+      "step": 2882
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001613061074862196,
+      "loss": 0.9841,
+      "step": 2883
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016127590505518193,
+      "loss": 1.0402,
+      "step": 2884
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016124569367149616,
+      "loss": 0.7632,
+      "step": 2885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016121547333957628,
+      "loss": 0.9971,
+      "step": 2886
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001611852440638376,
+      "loss": 1.0,
+      "step": 2887
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016115500584869676,
+      "loss": 0.7239,
+      "step": 2888
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001611247586985717,
+      "loss": 1.3056,
+      "step": 2889
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016109450261788156,
+      "loss": 1.075,
+      "step": 2890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016106423761104698,
+      "loss": 1.15,
+      "step": 2891
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016103396368248973,
+      "loss": 0.9597,
+      "step": 2892
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016100368083663295,
+      "loss": 0.9073,
+      "step": 2893
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001609733890779011,
+      "loss": 1.0338,
+      "step": 2894
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016094308841071991,
+      "loss": 1.1787,
+      "step": 2895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001609127788395165,
+      "loss": 1.2733,
+      "step": 2896
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016088246036871903,
+      "loss": 1.2558,
+      "step": 2897
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016085213300275736,
+      "loss": 0.5463,
+      "step": 2898
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001608217967460623,
+      "loss": 1.0275,
+      "step": 2899
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016079145160306614,
+      "loss": 1.2117,
+      "step": 2900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016076109757820242,
+      "loss": 0.9483,
+      "step": 2901
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016073073467590604,
+      "loss": 1.0338,
+      "step": 2902
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016070036290061302,
+      "loss": 1.2488,
+      "step": 2903
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016066998225676085,
+      "loss": 0.7744,
+      "step": 2904
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016063959274878825,
+      "loss": 1.4042,
+      "step": 2905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016060919438113533,
+      "loss": 1.2316,
+      "step": 2906
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016057878715824327,
+      "loss": 1.3325,
+      "step": 2907
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001605483710845548,
+      "loss": 0.9129,
+      "step": 2908
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016051794616451376,
+      "loss": 1.0679,
+      "step": 2909
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016048751240256538,
+      "loss": 0.6617,
+      "step": 2910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001604570698031561,
+      "loss": 1.4214,
+      "step": 2911
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001604266183707338,
+      "loss": 1.5201,
+      "step": 2912
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016039615810974745,
+      "loss": 0.647,
+      "step": 2913
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001603656890246475,
+      "loss": 0.7108,
+      "step": 2914
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016033521111988552,
+      "loss": 0.8685,
+      "step": 2915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016030472439991449,
+      "loss": 1.4118,
+      "step": 2916
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016027422886918865,
+      "loss": 1.4032,
+      "step": 2917
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016024372453216353,
+      "loss": 1.331,
+      "step": 2918
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016021321139329586,
+      "loss": 0.68,
+      "step": 2919
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001601826894570438,
+      "loss": 1.1312,
+      "step": 2920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001601521587278667,
+      "loss": 1.1966,
+      "step": 2921
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001601216192102252,
+      "loss": 0.607,
+      "step": 2922
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016009107090858126,
+      "loss": 0.5178,
+      "step": 2923
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016006051382739812,
+      "loss": 0.4596,
+      "step": 2924
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016002994797114026,
+      "loss": 1.5752,
+      "step": 2925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015999937334427352,
+      "loss": 1.8363,
+      "step": 2926
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015996878995126488,
+      "loss": 1.428,
+      "step": 2927
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001599381977965828,
+      "loss": 1.2626,
+      "step": 2928
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015990759688469682,
+      "loss": 1.2347,
+      "step": 2929
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015987698722007792,
+      "loss": 1.2654,
+      "step": 2930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015984636880719824,
+      "loss": 0.9864,
+      "step": 2931
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001598157416505313,
+      "loss": 0.5263,
+      "step": 2932
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015978510575455178,
+      "loss": 1.1225,
+      "step": 2933
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015975446112373578,
+      "loss": 0.5434,
+      "step": 2934
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015972380776256055,
+      "loss": 1.1787,
+      "step": 2935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015969314567550468,
+      "loss": 0.6362,
+      "step": 2936
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000159662474867048,
+      "loss": 1.6106,
+      "step": 2937
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015963179534167173,
+      "loss": 0.9673,
+      "step": 2938
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001596011071038581,
+      "loss": 1.4531,
+      "step": 2939
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015957041015809091,
+      "loss": 1.0878,
+      "step": 2940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015953970450885508,
+      "loss": 1.4749,
+      "step": 2941
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001595089901606368,
+      "loss": 1.2258,
+      "step": 2942
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015947826711792354,
+      "loss": 0.5567,
+      "step": 2943
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001594475353852041,
+      "loss": 1.2835,
+      "step": 2944
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015941679496696848,
+      "loss": 0.6622,
+      "step": 2945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000159386045867708,
+      "loss": 0.831,
+      "step": 2946
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001593552880919152,
+      "loss": 1.5408,
+      "step": 2947
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001593245216440839,
+      "loss": 0.8217,
+      "step": 2948
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015929374652870926,
+      "loss": 1.2905,
+      "step": 2949
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015926296275028755,
+      "loss": 0.6041,
+      "step": 2950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015923217031331647,
+      "loss": 1.4685,
+      "step": 2951
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001592013692222949,
+      "loss": 0.5885,
+      "step": 2952
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015917055948172298,
+      "loss": 0.6123,
+      "step": 2953
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015913974109610214,
+      "loss": 1.1941,
+      "step": 2954
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015910891406993507,
+      "loss": 0.9442,
+      "step": 2955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001590780784077257,
+      "loss": 0.6016,
+      "step": 2956
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001590472341139793,
+      "loss": 0.9582,
+      "step": 2957
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015901638119320224,
+      "loss": 1.1663,
+      "step": 2958
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001589855196499024,
+      "loss": 0.9168,
+      "step": 2959
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015895464948858864,
+      "loss": 0.6019,
+      "step": 2960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015892377071377126,
+      "loss": 1.2332,
+      "step": 2961
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015889288332996174,
+      "loss": 1.2479,
+      "step": 2962
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001588619873416729,
+      "loss": 1.3851,
+      "step": 2963
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001588310827534187,
+      "loss": 0.6627,
+      "step": 2964
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001588001695697145,
+      "loss": 1.2197,
+      "step": 2965
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015876924779507676,
+      "loss": 0.8913,
+      "step": 2966
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015873831743402333,
+      "loss": 1.2281,
+      "step": 2967
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001587073784910732,
+      "loss": 1.1915,
+      "step": 2968
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001586764309707467,
+      "loss": 0.936,
+      "step": 2969
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015864547487756542,
+      "loss": 1.5063,
+      "step": 2970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001586145102160521,
+      "loss": 1.252,
+      "step": 2971
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015858353699073084,
+      "loss": 0.7311,
+      "step": 2972
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015855255520612693,
+      "loss": 1.1122,
+      "step": 2973
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001585215648667669,
+      "loss": 2.0358,
+      "step": 2974
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015849056597717863,
+      "loss": 0.5938,
+      "step": 2975
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001584595585418911,
+      "loss": 1.3749,
+      "step": 2976
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001584285425654347,
+      "loss": 0.5284,
+      "step": 2977
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015839751805234094,
+      "loss": 1.1362,
+      "step": 2978
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001583664850071426,
+      "loss": 1.1679,
+      "step": 2979
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015833544343437378,
+      "loss": 1.5219,
+      "step": 2980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015830439333856976,
+      "loss": 1.4951,
+      "step": 2981
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015827333472426703,
+      "loss": 1.3213,
+      "step": 2982
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015824226759600341,
+      "loss": 1.5773,
+      "step": 2983
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015821119195831798,
+      "loss": 1.4029,
+      "step": 2984
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001581801078157509,
+      "loss": 1.0252,
+      "step": 2985
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015814901517284378,
+      "loss": 0.8317,
+      "step": 2986
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015811791403413933,
+      "loss": 1.1849,
+      "step": 2987
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015808680440418157,
+      "loss": 1.1552,
+      "step": 2988
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001580556862875157,
+      "loss": 1.5816,
+      "step": 2989
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001580245596886883,
+      "loss": 1.3383,
+      "step": 2990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015799342461224694,
+      "loss": 0.4859,
+      "step": 2991
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015796228106274067,
+      "loss": 0.9666,
+      "step": 2992
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015793112904471967,
+      "loss": 0.9355,
+      "step": 2993
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015789996856273533,
+      "loss": 1.8104,
+      "step": 2994
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001578687996213404,
+      "loss": 1.0226,
+      "step": 2995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001578376222250887,
+      "loss": 1.2936,
+      "step": 2996
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001578064363785354,
+      "loss": 1.2584,
+      "step": 2997
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015777524208623688,
+      "loss": 1.423,
+      "step": 2998
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015774403935275077,
+      "loss": 0.5895,
+      "step": 2999
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015771282818263584,
+      "loss": 1.3878,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015768160858045226,
+      "loss": 1.1896,
+      "step": 3001
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015765038055076124,
+      "loss": 1.2517,
+      "step": 3002
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015761914409812542,
+      "loss": 1.214,
+      "step": 3003
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015758789922710843,
+      "loss": 0.6666,
+      "step": 3004
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015755664594227538,
+      "loss": 1.1379,
+      "step": 3005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001575253842481925,
+      "loss": 1.3259,
+      "step": 3006
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015749411414942715,
+      "loss": 1.1195,
+      "step": 3007
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015746283565054809,
+      "loss": 1.4416,
+      "step": 3008
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015743154875612523,
+      "loss": 0.4813,
+      "step": 3009
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015740025347072965,
+      "loss": 1.1165,
+      "step": 3010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015736894979893372,
+      "loss": 1.0614,
+      "step": 3011
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015733763774531114,
+      "loss": 0.6484,
+      "step": 3012
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001573063173144366,
+      "loss": 1.128,
+      "step": 3013
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015727498851088618,
+      "loss": 1.3315,
+      "step": 3014
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015724365133923716,
+      "loss": 1.0212,
+      "step": 3015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015721230580406794,
+      "loss": 0.641,
+      "step": 3016
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015718095190995834,
+      "loss": 1.0289,
+      "step": 3017
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015714958966148922,
+      "loss": 0.7772,
+      "step": 3018
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001571182190632427,
+      "loss": 0.605,
+      "step": 3019
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015708684011980222,
+      "loss": 0.5201,
+      "step": 3020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015705545283575231,
+      "loss": 0.7251,
+      "step": 3021
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001570240572156788,
+      "loss": 1.0089,
+      "step": 3022
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015699265326416876,
+      "loss": 1.0661,
+      "step": 3023
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015696124098581028,
+      "loss": 0.6162,
+      "step": 3024
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015692982038519298,
+      "loss": 0.9702,
+      "step": 3025
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015689839146690746,
+      "loss": 0.4761,
+      "step": 3026
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015686695423554556,
+      "loss": 1.6877,
+      "step": 3027
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015683550869570047,
+      "loss": 1.3583,
+      "step": 3028
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015680405485196647,
+      "loss": 1.062,
+      "step": 3029
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001567725927089391,
+      "loss": 0.9184,
+      "step": 3030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015674112227121505,
+      "loss": 1.0833,
+      "step": 3031
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015670964354339232,
+      "loss": 1.2873,
+      "step": 3032
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015667815653007007,
+      "loss": 1.304,
+      "step": 3033
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015664666123584869,
+      "loss": 1.1198,
+      "step": 3034
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015661515766532972,
+      "loss": 0.4328,
+      "step": 3035
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000156583645823116,
+      "loss": 1.3226,
+      "step": 3036
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015655212571381153,
+      "loss": 1.2959,
+      "step": 3037
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015652059734202147,
+      "loss": 1.7843,
+      "step": 3038
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015648906071235227,
+      "loss": 1.1802,
+      "step": 3039
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015645751582941158,
+      "loss": 1.2294,
+      "step": 3040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015642596269780818,
+      "loss": 0.6835,
+      "step": 3041
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015639440132215215,
+      "loss": 1.321,
+      "step": 3042
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001563628317070547,
+      "loss": 0.969,
+      "step": 3043
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001563312538571283,
+      "loss": 0.6389,
+      "step": 3044
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015629966777698658,
+      "loss": 1.0299,
+      "step": 3045
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001562680734712444,
+      "loss": 1.404,
+      "step": 3046
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015623647094451778,
+      "loss": 1.4444,
+      "step": 3047
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015620486020142402,
+      "loss": 0.5698,
+      "step": 3048
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001561732412465815,
+      "loss": 1.1135,
+      "step": 3049
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015614161408460995,
+      "loss": 1.1778,
+      "step": 3050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015610997872013017,
+      "loss": 1.445,
+      "step": 3051
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015607833515776426,
+      "loss": 0.4946,
+      "step": 3052
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001560466834021354,
+      "loss": 1.6607,
+      "step": 3053
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015601502345786812,
+      "loss": 1.3508,
+      "step": 3054
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015598335532958797,
+      "loss": 0.7434,
+      "step": 3055
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015595167902192187,
+      "loss": 0.9753,
+      "step": 3056
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001559199945394978,
+      "loss": 0.7419,
+      "step": 3057
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015588830188694496,
+      "loss": 1.2492,
+      "step": 3058
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015585660106889386,
+      "loss": 0.5954,
+      "step": 3059
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015582489208997602,
+      "loss": 1.4789,
+      "step": 3060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001557931749548243,
+      "loss": 0.7348,
+      "step": 3061
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001557614496680727,
+      "loss": 1.0996,
+      "step": 3062
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015572971623435631,
+      "loss": 0.9135,
+      "step": 3063
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015569797465831168,
+      "loss": 1.5153,
+      "step": 3064
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015566622494457625,
+      "loss": 1.1685,
+      "step": 3065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015563446709778881,
+      "loss": 0.7169,
+      "step": 3066
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015560270112258934,
+      "loss": 1.5494,
+      "step": 3067
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015557092702361892,
+      "loss": 0.8714,
+      "step": 3068
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015553914480551987,
+      "loss": 1.2219,
+      "step": 3069
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015550735447293575,
+      "loss": 1.9014,
+      "step": 3070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015547555603051118,
+      "loss": 1.1015,
+      "step": 3071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001554437494828921,
+      "loss": 1.5018,
+      "step": 3072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015541193483472552,
+      "loss": 1.0444,
+      "step": 3073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015538011209065972,
+      "loss": 1.0394,
+      "step": 3074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015534828125534413,
+      "loss": 1.3978,
+      "step": 3075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001553164423334293,
+      "loss": 1.1363,
+      "step": 3076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015528459532956708,
+      "loss": 1.242,
+      "step": 3077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015525274024841049,
+      "loss": 1.388,
+      "step": 3078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001552208770946135,
+      "loss": 1.1561,
+      "step": 3079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015518900587283162,
+      "loss": 0.8973,
+      "step": 3080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015515712658772132,
+      "loss": 1.0817,
+      "step": 3081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015512523924394021,
+      "loss": 1.1497,
+      "step": 3082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015509334384614722,
+      "loss": 1.4022,
+      "step": 3083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001550614403990024,
+      "loss": 1.0427,
+      "step": 3084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015502952890716692,
+      "loss": 0.9536,
+      "step": 3085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001549976093753032,
+      "loss": 0.618,
+      "step": 3086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015496568180807486,
+      "loss": 1.2976,
+      "step": 3087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015493374621014655,
+      "loss": 1.3588,
+      "step": 3088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015490180258618424,
+      "loss": 1.5647,
+      "step": 3089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000154869850940855,
+      "loss": 0.8995,
+      "step": 3090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001548378912788271,
+      "loss": 1.0097,
+      "step": 3091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015480592360476994,
+      "loss": 1.0692,
+      "step": 3092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015477394792335413,
+      "loss": 0.5671,
+      "step": 3093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001547419642392515,
+      "loss": 0.9349,
+      "step": 3094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001547099725571349,
+      "loss": 1.3083,
+      "step": 3095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015467797288167852,
+      "loss": 0.9987,
+      "step": 3096
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001546459652175576,
+      "loss": 1.2859,
+      "step": 3097
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001546139495694486,
+      "loss": 1.3668,
+      "step": 3098
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001545819259420291,
+      "loss": 1.5231,
+      "step": 3099
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015454989433997786,
+      "loss": 1.5196,
+      "step": 3100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015451785476797492,
+      "loss": 0.6676,
+      "step": 3101
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001544858072307013,
+      "loss": 1.2568,
+      "step": 3102
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015445375173283926,
+      "loss": 1.1986,
+      "step": 3103
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015442168827907232,
+      "loss": 1.3238,
+      "step": 3104
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000154389616874085,
+      "loss": 1.1449,
+      "step": 3105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015435753752256307,
+      "loss": 0.8147,
+      "step": 3106
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015432545022919344,
+      "loss": 1.5691,
+      "step": 3107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001542933549986642,
+      "loss": 0.6418,
+      "step": 3108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015426125183566467,
+      "loss": 0.7273,
+      "step": 3109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015422914074488507,
+      "loss": 0.3836,
+      "step": 3110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015419702173101712,
+      "loss": 0.8158,
+      "step": 3111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015416489479875344,
+      "loss": 1.2225,
+      "step": 3112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015413275995278792,
+      "loss": 1.5106,
+      "step": 3113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015410061719781562,
+      "loss": 1.3339,
+      "step": 3114
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015406846653853268,
+      "loss": 1.419,
+      "step": 3115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015403630797963645,
+      "loss": 0.848,
+      "step": 3116
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015400414152582544,
+      "loss": 1.2816,
+      "step": 3117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015397196718179923,
+      "loss": 1.169,
+      "step": 3118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001539397849522587,
+      "loss": 1.2858,
+      "step": 3119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015390759484190577,
+      "loss": 0.8611,
+      "step": 3120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001538753968554435,
+      "loss": 1.0256,
+      "step": 3121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015384319099757618,
+      "loss": 0.5626,
+      "step": 3122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015381097727300923,
+      "loss": 0.6994,
+      "step": 3123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015377875568644913,
+      "loss": 1.0004,
+      "step": 3124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015374652624260363,
+      "loss": 1.138,
+      "step": 3125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015371428894618162,
+      "loss": 0.8973,
+      "step": 3126
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000153682043801893,
+      "loss": 1.3033,
+      "step": 3127
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015364979081444898,
+      "loss": 1.2814,
+      "step": 3128
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015361752998856182,
+      "loss": 1.0587,
+      "step": 3129
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015358526132894492,
+      "loss": 0.6765,
+      "step": 3130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015355298484031295,
+      "loss": 1.6455,
+      "step": 3131
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015352070052738153,
+      "loss": 1.0331,
+      "step": 3132
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015348840839486758,
+      "loss": 1.2032,
+      "step": 3133
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015345610844748908,
+      "loss": 0.6608,
+      "step": 3134
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015342380068996523,
+      "loss": 0.5981,
+      "step": 3135
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015339148512701623,
+      "loss": 0.7905,
+      "step": 3136
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015335916176336358,
+      "loss": 1.3599,
+      "step": 3137
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015332683060372983,
+      "loss": 0.5052,
+      "step": 3138
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015329449165283867,
+      "loss": 1.3213,
+      "step": 3139
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015326214491541497,
+      "loss": 1.5564,
+      "step": 3140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015322979039618473,
+      "loss": 1.209,
+      "step": 3141
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015319742809987504,
+      "loss": 1.2946,
+      "step": 3142
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015316505803121414,
+      "loss": 1.1277,
+      "step": 3143
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015313268019493146,
+      "loss": 0.8092,
+      "step": 3144
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015310029459575755,
+      "loss": 1.3811,
+      "step": 3145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000153067901238424,
+      "loss": 1.1967,
+      "step": 3146
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015303550012766364,
+      "loss": 1.1975,
+      "step": 3147
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001530030912682104,
+      "loss": 1.2239,
+      "step": 3148
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015297067466479936,
+      "loss": 1.5014,
+      "step": 3149
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001529382503221667,
+      "loss": 1.1525,
+      "step": 3150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015290581824504973,
+      "loss": 0.4387,
+      "step": 3151
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015287337843818691,
+      "loss": 1.4184,
+      "step": 3152
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001528409309063178,
+      "loss": 0.6288,
+      "step": 3153
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015280847565418316,
+      "loss": 1.3653,
+      "step": 3154
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001527760126865248,
+      "loss": 1.287,
+      "step": 3155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015274354200808565,
+      "loss": 0.518,
+      "step": 3156
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001527110636236099,
+      "loss": 1.1942,
+      "step": 3157
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015267857753784261,
+      "loss": 0.6913,
+      "step": 3158
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015264608375553033,
+      "loss": 1.5105,
+      "step": 3159
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015261358228142036,
+      "loss": 1.3756,
+      "step": 3160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001525810731202614,
+      "loss": 1.3796,
+      "step": 3161
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015254855627680307,
+      "loss": 1.7794,
+      "step": 3162
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015251603175579626,
+      "loss": 1.1942,
+      "step": 3163
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001524834995619929,
+      "loss": 1.3036,
+      "step": 3164
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001524509597001461,
+      "loss": 0.8931,
+      "step": 3165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015241841217501008,
+      "loss": 1.3495,
+      "step": 3166
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015238585699134011,
+      "loss": 1.7978,
+      "step": 3167
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001523532941538927,
+      "loss": 0.6263,
+      "step": 3168
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015232072366742528,
+      "loss": 0.5379,
+      "step": 3169
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015228814553669667,
+      "loss": 1.1719,
+      "step": 3170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015225555976646656,
+      "loss": 1.1997,
+      "step": 3171
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015222296636149588,
+      "loss": 0.6073,
+      "step": 3172
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001521903653265467,
+      "loss": 1.039,
+      "step": 3173
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001521577566663821,
+      "loss": 0.6261,
+      "step": 3174
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015212514038576638,
+      "loss": 1.1896,
+      "step": 3175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015209251648946485,
+      "loss": 1.2502,
+      "step": 3176
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015205988498224405,
+      "loss": 1.2149,
+      "step": 3177
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015202724586887149,
+      "loss": 1.8624,
+      "step": 3178
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000151994599154116,
+      "loss": 0.7325,
+      "step": 3179
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015196194484274727,
+      "loss": 0.7859,
+      "step": 3180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015192928293953628,
+      "loss": 1.2215,
+      "step": 3181
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015189661344925501,
+      "loss": 0.6649,
+      "step": 3182
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015186393637667666,
+      "loss": 0.4453,
+      "step": 3183
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015183125172657546,
+      "loss": 1.4598,
+      "step": 3184
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015179855950372674,
+      "loss": 0.891,
+      "step": 3185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015176585971290698,
+      "loss": 0.8279,
+      "step": 3186
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015173315235889376,
+      "loss": 0.7136,
+      "step": 3187
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015170043744646574,
+      "loss": 1.3768,
+      "step": 3188
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015166771498040265,
+      "loss": 1.2368,
+      "step": 3189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001516349849654855,
+      "loss": 1.0922,
+      "step": 3190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015160224740649612,
+      "loss": 0.5379,
+      "step": 3191
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015156950230821767,
+      "loss": 1.1956,
+      "step": 3192
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015153674967543433,
+      "loss": 0.9629,
+      "step": 3193
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015150398951293135,
+      "loss": 1.288,
+      "step": 3194
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015147122182549523,
+      "loss": 1.0242,
+      "step": 3195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015143844661791333,
+      "loss": 1.4923,
+      "step": 3196
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015140566389497425,
+      "loss": 0.9661,
+      "step": 3197
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015137287366146773,
+      "loss": 0.93,
+      "step": 3198
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015134007592218453,
+      "loss": 1.534,
+      "step": 3199
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015130727068191652,
+      "loss": 0.7132,
+      "step": 3200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015127445794545672,
+      "loss": 1.4412,
+      "step": 3201
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001512416377175991,
+      "loss": 1.436,
+      "step": 3202
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001512088100031389,
+      "loss": 1.5648,
+      "step": 3203
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015117597480687235,
+      "loss": 0.6514,
+      "step": 3204
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001511431321335968,
+      "loss": 1.1875,
+      "step": 3205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001511102819881107,
+      "loss": 1.4532,
+      "step": 3206
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001510774243752136,
+      "loss": 1.3111,
+      "step": 3207
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001510445592997061,
+      "loss": 1.2556,
+      "step": 3208
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001510116867663899,
+      "loss": 1.3487,
+      "step": 3209
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015097880678006785,
+      "loss": 1.2006,
+      "step": 3210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001509459193455438,
+      "loss": 1.3607,
+      "step": 3211
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015091302446762278,
+      "loss": 0.5516,
+      "step": 3212
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015088012215111078,
+      "loss": 1.0179,
+      "step": 3213
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015084721240081508,
+      "loss": 1.1706,
+      "step": 3214
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001508142952215438,
+      "loss": 1.3482,
+      "step": 3215
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015078137061810636,
+      "loss": 1.3041,
+      "step": 3216
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001507484385953131,
+      "loss": 0.6073,
+      "step": 3217
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015071549915797553,
+      "loss": 1.5582,
+      "step": 3218
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015068255231090628,
+      "loss": 1.0055,
+      "step": 3219
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015064959805891897,
+      "loss": 1.196,
+      "step": 3220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015061663640682837,
+      "loss": 0.7192,
+      "step": 3221
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015058366735945026,
+      "loss": 0.719,
+      "step": 3222
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015055069092160164,
+      "loss": 1.2037,
+      "step": 3223
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015051770709810036,
+      "loss": 1.0519,
+      "step": 3224
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001504847158937656,
+      "loss": 1.1349,
+      "step": 3225
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015045171731341745,
+      "loss": 1.2707,
+      "step": 3226
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015041871136187713,
+      "loss": 1.1508,
+      "step": 3227
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015038569804396695,
+      "loss": 1.0014,
+      "step": 3228
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015035267736451032,
+      "loss": 1.2996,
+      "step": 3229
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015031964932833163,
+      "loss": 1.3729,
+      "step": 3230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015028661394025643,
+      "loss": 0.9841,
+      "step": 3231
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015025357120511133,
+      "loss": 1.4781,
+      "step": 3232
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015022052112772394,
+      "loss": 1.1136,
+      "step": 3233
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015018746371292313,
+      "loss": 1.8112,
+      "step": 3234
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015015439896553856,
+      "loss": 1.2773,
+      "step": 3235
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015012132689040124,
+      "loss": 1.6025,
+      "step": 3236
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015008824749234312,
+      "loss": 0.6473,
+      "step": 3237
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015005516077619716,
+      "loss": 1.3564,
+      "step": 3238
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001500220667467975,
+      "loss": 1.4169,
+      "step": 3239
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014998896540897925,
+      "loss": 1.3496,
+      "step": 3240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014995585676757877,
+      "loss": 0.5606,
+      "step": 3241
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014992274082743329,
+      "loss": 1.2476,
+      "step": 3242
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014988961759338114,
+      "loss": 0.6319,
+      "step": 3243
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014985648707026175,
+      "loss": 1.0174,
+      "step": 3244
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014982334926291575,
+      "loss": 1.4318,
+      "step": 3245
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014979020417618455,
+      "loss": 1.2722,
+      "step": 3246
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014975705181491085,
+      "loss": 0.9972,
+      "step": 3247
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001497238921839383,
+      "loss": 1.2686,
+      "step": 3248
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001496907252881117,
+      "loss": 1.1424,
+      "step": 3249
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001496575511322768,
+      "loss": 1.2306,
+      "step": 3250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014962436972128056,
+      "loss": 0.5978,
+      "step": 3251
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014959118105997086,
+      "loss": 0.8271,
+      "step": 3252
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014955798515319665,
+      "loss": 0.9023,
+      "step": 3253
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001495247820058081,
+      "loss": 0.9473,
+      "step": 3254
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014949157162265616,
+      "loss": 1.0376,
+      "step": 3255
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014945835400859316,
+      "loss": 1.5359,
+      "step": 3256
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001494251291684722,
+      "loss": 1.2968,
+      "step": 3257
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001493918971071476,
+      "loss": 1.1561,
+      "step": 3258
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014935865782947475,
+      "loss": 1.1676,
+      "step": 3259
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014932541134030992,
+      "loss": 1.6306,
+      "step": 3260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014929215764451062,
+      "loss": 0.5774,
+      "step": 3261
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014925889674693537,
+      "loss": 0.6243,
+      "step": 3262
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001492256286524437,
+      "loss": 0.6513,
+      "step": 3263
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014919235336589615,
+      "loss": 1.2124,
+      "step": 3264
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014915907089215445,
+      "loss": 0.676,
+      "step": 3265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014912578123608126,
+      "loss": 1.5666,
+      "step": 3266
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014909248440254033,
+      "loss": 0.9731,
+      "step": 3267
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014905918039639644,
+      "loss": 1.2374,
+      "step": 3268
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014902586922251547,
+      "loss": 0.5725,
+      "step": 3269
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014899255088576428,
+      "loss": 0.6651,
+      "step": 3270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014895922539101084,
+      "loss": 1.2395,
+      "step": 3271
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014892589274312408,
+      "loss": 1.4102,
+      "step": 3272
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001488925529469741,
+      "loss": 1.4987,
+      "step": 3273
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014885920600743194,
+      "loss": 1.3386,
+      "step": 3274
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001488258519293697,
+      "loss": 0.6721,
+      "step": 3275
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001487924907176606,
+      "loss": 1.2491,
+      "step": 3276
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014875912237717877,
+      "loss": 0.9955,
+      "step": 3277
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014872574691279948,
+      "loss": 1.1782,
+      "step": 3278
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014869236432939905,
+      "loss": 1.4925,
+      "step": 3279
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014865897463185477,
+      "loss": 0.7151,
+      "step": 3280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000148625577825045,
+      "loss": 1.045,
+      "step": 3281
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014859217391384916,
+      "loss": 0.8033,
+      "step": 3282
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001485587629031477,
+      "loss": 1.3293,
+      "step": 3283
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014852534479782204,
+      "loss": 1.535,
+      "step": 3284
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014849191960275476,
+      "loss": 0.8499,
+      "step": 3285
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001484584873228294,
+      "loss": 1.4907,
+      "step": 3286
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014842504796293057,
+      "loss": 0.655,
+      "step": 3287
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001483916015279438,
+      "loss": 0.9483,
+      "step": 3288
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014835814802275585,
+      "loss": 1.3875,
+      "step": 3289
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014832468745225433,
+      "loss": 1.714,
+      "step": 3290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014829121982132802,
+      "loss": 1.261,
+      "step": 3291
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001482577451348666,
+      "loss": 0.5668,
+      "step": 3292
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001482242633977609,
+      "loss": 0.612,
+      "step": 3293
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014819077461490274,
+      "loss": 1.6702,
+      "step": 3294
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014815727879118496,
+      "loss": 0.8754,
+      "step": 3295
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014812377593150137,
+      "loss": 1.05,
+      "step": 3296
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014809026604074694,
+      "loss": 0.7156,
+      "step": 3297
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014805674912381756,
+      "loss": 1.3965,
+      "step": 3298
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001480232251856102,
+      "loss": 1.349,
+      "step": 3299
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014798969423102283,
+      "loss": 1.3446,
+      "step": 3300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014795615626495446,
+      "loss": 0.4464,
+      "step": 3301
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001479226112923051,
+      "loss": 0.8533,
+      "step": 3302
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014788905931797578,
+      "loss": 0.9356,
+      "step": 3303
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014785550034686863,
+      "loss": 1.0157,
+      "step": 3304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014782193438388674,
+      "loss": 1.3821,
+      "step": 3305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001477883614339342,
+      "loss": 1.2007,
+      "step": 3306
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014775478150191615,
+      "loss": 1.0442,
+      "step": 3307
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014772119459273872,
+      "loss": 1.1354,
+      "step": 3308
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014768760071130917,
+      "loss": 1.1135,
+      "step": 3309
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014765399986253565,
+      "loss": 1.163,
+      "step": 3310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014762039205132736,
+      "loss": 1.2761,
+      "step": 3311
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014758677728259455,
+      "loss": 1.3855,
+      "step": 3312
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014755315556124848,
+      "loss": 1.1716,
+      "step": 3313
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014751952689220135,
+      "loss": 1.1368,
+      "step": 3314
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014748589128036653,
+      "loss": 1.3275,
+      "step": 3315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014745224873065826,
+      "loss": 1.3034,
+      "step": 3316
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014741859924799185,
+      "loss": 0.8637,
+      "step": 3317
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014738494283728367,
+      "loss": 1.2168,
+      "step": 3318
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014735127950345095,
+      "loss": 0.9901,
+      "step": 3319
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014731760925141218,
+      "loss": 1.4359,
+      "step": 3320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014728393208608654,
+      "loss": 0.9868,
+      "step": 3321
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014725024801239456,
+      "loss": 1.5185,
+      "step": 3322
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014721655703525753,
+      "loss": 0.5836,
+      "step": 3323
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001471828591595978,
+      "loss": 1.1981,
+      "step": 3324
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014714915439033882,
+      "loss": 1.3701,
+      "step": 3325
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014711544273240498,
+      "loss": 0.5473,
+      "step": 3326
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014708172419072168,
+      "loss": 1.3778,
+      "step": 3327
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001470479987702153,
+      "loss": 0.5594,
+      "step": 3328
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014701426647581337,
+      "loss": 1.6069,
+      "step": 3329
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014698052731244415,
+      "loss": 0.5439,
+      "step": 3330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014694678128503714,
+      "loss": 1.3728,
+      "step": 3331
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014691302839852274,
+      "loss": 0.6267,
+      "step": 3332
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014687926865783243,
+      "loss": 1.2325,
+      "step": 3333
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014684550206789862,
+      "loss": 1.345,
+      "step": 3334
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001468117286336547,
+      "loss": 0.5709,
+      "step": 3335
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014677794836003512,
+      "loss": 1.1131,
+      "step": 3336
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014674416125197532,
+      "loss": 1.547,
+      "step": 3337
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014671036731441175,
+      "loss": 1.3425,
+      "step": 3338
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001466765665522818,
+      "loss": 1.1648,
+      "step": 3339
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001466427589705239,
+      "loss": 1.0392,
+      "step": 3340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001466089445740775,
+      "loss": 0.9568,
+      "step": 3341
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014657512336788294,
+      "loss": 0.9129,
+      "step": 3342
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001465412953568817,
+      "loss": 1.0273,
+      "step": 3343
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014650746054601616,
+      "loss": 1.0083,
+      "step": 3344
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014647361894022974,
+      "loss": 1.2917,
+      "step": 3345
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014643977054446677,
+      "loss": 0.8655,
+      "step": 3346
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014640591536367267,
+      "loss": 1.0752,
+      "step": 3347
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014637205340279385,
+      "loss": 0.9944,
+      "step": 3348
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014633818466677762,
+      "loss": 0.7613,
+      "step": 3349
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014630430916057235,
+      "loss": 1.592,
+      "step": 3350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014627042688912742,
+      "loss": 0.6731,
+      "step": 3351
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001462365378573931,
+      "loss": 0.8843,
+      "step": 3352
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014620264207032072,
+      "loss": 1.1251,
+      "step": 3353
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014616873953286264,
+      "loss": 1.2286,
+      "step": 3354
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001461348302499721,
+      "loss": 1.4584,
+      "step": 3355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001461009142266034,
+      "loss": 1.469,
+      "step": 3356
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001460669914677118,
+      "loss": 1.0859,
+      "step": 3357
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014603306197825353,
+      "loss": 1.1912,
+      "step": 3358
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014599912576318587,
+      "loss": 1.453,
+      "step": 3359
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014596518282746697,
+      "loss": 1.0936,
+      "step": 3360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014593123317605601,
+      "loss": 1.6956,
+      "step": 3361
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001458972768139133,
+      "loss": 0.879,
+      "step": 3362
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014586331374599984,
+      "loss": 0.6636,
+      "step": 3363
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014582934397727786,
+      "loss": 1.3421,
+      "step": 3364
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014579536751271045,
+      "loss": 0.6277,
+      "step": 3365
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001457613843572617,
+      "loss": 1.2367,
+      "step": 3366
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014572739451589666,
+      "loss": 1.1889,
+      "step": 3367
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001456933979935814,
+      "loss": 1.2736,
+      "step": 3368
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000145659394795283,
+      "loss": 1.6333,
+      "step": 3369
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014562538492596936,
+      "loss": 1.2255,
+      "step": 3370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001455913683906095,
+      "loss": 1.1171,
+      "step": 3371
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014555734519417334,
+      "loss": 0.8188,
+      "step": 3372
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014552331534163187,
+      "loss": 1.2136,
+      "step": 3373
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014548927883795693,
+      "loss": 0.7255,
+      "step": 3374
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001454552356881214,
+      "loss": 0.9899,
+      "step": 3375
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001454211858970991,
+      "loss": 0.5037,
+      "step": 3376
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001453871294698649,
+      "loss": 0.7515,
+      "step": 3377
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014535306641139445,
+      "loss": 1.3368,
+      "step": 3378
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014531899672666463,
+      "loss": 1.449,
+      "step": 3379
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014528492042065308,
+      "loss": 1.6661,
+      "step": 3380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014525083749833848,
+      "loss": 0.3439,
+      "step": 3381
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014521674796470052,
+      "loss": 1.3508,
+      "step": 3382
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014518265182471976,
+      "loss": 1.4104,
+      "step": 3383
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014514854908337786,
+      "loss": 1.3407,
+      "step": 3384
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014511443974565724,
+      "loss": 1.0644,
+      "step": 3385
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014508032381654152,
+      "loss": 1.1174,
+      "step": 3386
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014504620130101513,
+      "loss": 1.3077,
+      "step": 3387
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001450120722040635,
+      "loss": 0.9202,
+      "step": 3388
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014497793653067296,
+      "loss": 1.0915,
+      "step": 3389
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000144943794285831,
+      "loss": 1.0521,
+      "step": 3390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014490964547452583,
+      "loss": 1.3731,
+      "step": 3391
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001448754901017468,
+      "loss": 0.8714,
+      "step": 3392
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014484132817248405,
+      "loss": 1.5718,
+      "step": 3393
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014480715969172883,
+      "loss": 1.3252,
+      "step": 3394
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014477298466447332,
+      "loss": 1.4127,
+      "step": 3395
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001447388030957105,
+      "loss": 0.7492,
+      "step": 3396
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014470461499043458,
+      "loss": 0.6455,
+      "step": 3397
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001446704203536405,
+      "loss": 1.0479,
+      "step": 3398
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014463621919032422,
+      "loss": 0.6786,
+      "step": 3399
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014460201150548268,
+      "loss": 1.3306,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014456779730411375,
+      "loss": 1.4337,
+      "step": 3401
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014453357659121628,
+      "loss": 0.8137,
+      "step": 3402
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014449934937179002,
+      "loss": 0.6462,
+      "step": 3403
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014446511565083573,
+      "loss": 1.1515,
+      "step": 3404
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014443087543335504,
+      "loss": 1.4608,
+      "step": 3405
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014439662872435066,
+      "loss": 1.2298,
+      "step": 3406
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001443623755288261,
+      "loss": 0.7589,
+      "step": 3407
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001443281158517859,
+      "loss": 0.8632,
+      "step": 3408
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014429384969823553,
+      "loss": 1.1335,
+      "step": 3409
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014425957707318144,
+      "loss": 1.0862,
+      "step": 3410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001442252979816309,
+      "loss": 0.4672,
+      "step": 3411
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014419101242859236,
+      "loss": 0.9303,
+      "step": 3412
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014415672041907497,
+      "loss": 0.6945,
+      "step": 3413
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014412242195808897,
+      "loss": 0.8216,
+      "step": 3414
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001440881170506455,
+      "loss": 0.6344,
+      "step": 3415
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014405380570175658,
+      "loss": 1.4211,
+      "step": 3416
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001440194879164353,
+      "loss": 1.7284,
+      "step": 3417
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001439851636996956,
+      "loss": 0.6136,
+      "step": 3418
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001439508330565524,
+      "loss": 1.2517,
+      "step": 3419
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014391649599202147,
+      "loss": 1.3652,
+      "step": 3420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001438821525111197,
+      "loss": 0.6306,
+      "step": 3421
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001438478026188647,
+      "loss": 1.0828,
+      "step": 3422
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014381344632027519,
+      "loss": 1.1802,
+      "step": 3423
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001437790836203707,
+      "loss": 1.2524,
+      "step": 3424
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014374471452417183,
+      "loss": 0.9963,
+      "step": 3425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014371033903670002,
+      "loss": 1.5574,
+      "step": 3426
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001436759571629776,
+      "loss": 1.1906,
+      "step": 3427
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014364156890802794,
+      "loss": 1.2475,
+      "step": 3428
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014360717427687532,
+      "loss": 1.3465,
+      "step": 3429
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001435727732745449,
+      "loss": 1.3034,
+      "step": 3430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014353836590606277,
+      "loss": 0.9733,
+      "step": 3431
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014350395217645607,
+      "loss": 0.4999,
+      "step": 3432
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014346953209075266,
+      "loss": 1.0687,
+      "step": 3433
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014343510565398157,
+      "loss": 1.2467,
+      "step": 3434
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014340067287117256,
+      "loss": 1.4691,
+      "step": 3435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001433662337473564,
+      "loss": 1.1954,
+      "step": 3436
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014333178828756485,
+      "loss": 1.1259,
+      "step": 3437
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014329733649683042,
+      "loss": 1.0926,
+      "step": 3438
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001432628783801867,
+      "loss": 1.2258,
+      "step": 3439
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014322841394266817,
+      "loss": 1.4608,
+      "step": 3440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014319394318931022,
+      "loss": 1.2738,
+      "step": 3441
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014315946612514907,
+      "loss": 1.5647,
+      "step": 3442
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001431249827552221,
+      "loss": 1.2663,
+      "step": 3443
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014309049308456736,
+      "loss": 1.2671,
+      "step": 3444
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014305599711822399,
+      "loss": 1.0767,
+      "step": 3445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014302149486123192,
+      "loss": 0.7504,
+      "step": 3446
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001429869863186321,
+      "loss": 1.5481,
+      "step": 3447
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001429524714954664,
+      "loss": 1.2208,
+      "step": 3448
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014291795039677747,
+      "loss": 1.1051,
+      "step": 3449
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014288342302760907,
+      "loss": 1.4983,
+      "step": 3450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014284888939300575,
+      "loss": 1.6187,
+      "step": 3451
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014281434949801297,
+      "loss": 1.2707,
+      "step": 3452
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001427798033476772,
+      "loss": 1.3746,
+      "step": 3453
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014274525094704575,
+      "loss": 1.1199,
+      "step": 3454
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014271069230116688,
+      "loss": 0.6674,
+      "step": 3455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014267612741508968,
+      "loss": 1.247,
+      "step": 3456
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001426415562938643,
+      "loss": 1.1855,
+      "step": 3457
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001426069789425416,
+      "loss": 1.1898,
+      "step": 3458
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001425723953661736,
+      "loss": 1.3823,
+      "step": 3459
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000142537805569813,
+      "loss": 0.8006,
+      "step": 3460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014250320955851354,
+      "loss": 0.7976,
+      "step": 3461
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014246860733732982,
+      "loss": 0.505,
+      "step": 3462
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014243399891131735,
+      "loss": 0.7165,
+      "step": 3463
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014239938428553257,
+      "loss": 1.466,
+      "step": 3464
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014236476346503283,
+      "loss": 0.8306,
+      "step": 3465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014233013645487634,
+      "loss": 1.3144,
+      "step": 3466
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014229550326012226,
+      "loss": 1.3732,
+      "step": 3467
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001422608638858306,
+      "loss": 1.2981,
+      "step": 3468
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014222621833706237,
+      "loss": 1.4654,
+      "step": 3469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014219156661887935,
+      "loss": 1.1463,
+      "step": 3470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001421569087363443,
+      "loss": 1.3933,
+      "step": 3471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014212224469452094,
+      "loss": 1.8564,
+      "step": 3472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014208757449847377,
+      "loss": 1.5531,
+      "step": 3473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014205289815326822,
+      "loss": 0.7431,
+      "step": 3474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014201821566397064,
+      "loss": 1.3508,
+      "step": 3475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014198352703564834,
+      "loss": 1.4955,
+      "step": 3476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014194883227336942,
+      "loss": 1.3484,
+      "step": 3477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001419141313822029,
+      "loss": 0.9739,
+      "step": 3478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014187942436721875,
+      "loss": 1.3252,
+      "step": 3479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001418447112334878,
+      "loss": 1.2254,
+      "step": 3480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014180999198608175,
+      "loss": 1.0601,
+      "step": 3481
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014177526663007323,
+      "loss": 0.855,
+      "step": 3482
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014174053517053574,
+      "loss": 1.2444,
+      "step": 3483
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014170579761254372,
+      "loss": 1.0314,
+      "step": 3484
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001416710539611724,
+      "loss": 1.2524,
+      "step": 3485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000141636304221498,
+      "loss": 1.1419,
+      "step": 3486
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014160154839859757,
+      "loss": 0.7096,
+      "step": 3487
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014156678649754914,
+      "loss": 1.0548,
+      "step": 3488
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014153201852343145,
+      "loss": 1.3405,
+      "step": 3489
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014149724448132435,
+      "loss": 0.5056,
+      "step": 3490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014146246437630837,
+      "loss": 1.1397,
+      "step": 3491
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014142767821346506,
+      "loss": 0.6925,
+      "step": 3492
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014139288599787682,
+      "loss": 0.8393,
+      "step": 3493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014135808773462692,
+      "loss": 1.1596,
+      "step": 3494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014132328342879948,
+      "loss": 1.5126,
+      "step": 3495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001412884730854796,
+      "loss": 1.5173,
+      "step": 3496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014125365670975318,
+      "loss": 1.2346,
+      "step": 3497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014121883430670708,
+      "loss": 1.4974,
+      "step": 3498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014118400588142892,
+      "loss": 1.1426,
+      "step": 3499
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014114917143900728,
+      "loss": 0.5416,
+      "step": 3500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014111433098453163,
+      "loss": 1.8018,
+      "step": 3501
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001410794845230923,
+      "loss": 1.2205,
+      "step": 3502
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014104463205978046,
+      "loss": 1.2032,
+      "step": 3503
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014100977359968823,
+      "loss": 1.0055,
+      "step": 3504
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014097490914790855,
+      "loss": 1.6017,
+      "step": 3505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014094003870953524,
+      "loss": 1.0147,
+      "step": 3506
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014090516228966302,
+      "loss": 1.2907,
+      "step": 3507
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014087027989338746,
+      "loss": 0.6137,
+      "step": 3508
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014083539152580504,
+      "loss": 1.2469,
+      "step": 3509
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014080049719201302,
+      "loss": 1.3357,
+      "step": 3510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001407655968971097,
+      "loss": 1.4358,
+      "step": 3511
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001407306906461941,
+      "loss": 1.2965,
+      "step": 3512
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001406957784443661,
+      "loss": 1.1838,
+      "step": 3513
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014066086029672657,
+      "loss": 1.426,
+      "step": 3514
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001406259362083772,
+      "loss": 0.9321,
+      "step": 3515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014059100618442054,
+      "loss": 0.7169,
+      "step": 3516
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001405560702299599,
+      "loss": 1.6354,
+      "step": 3517
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001405211283500997,
+      "loss": 1.2386,
+      "step": 3518
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000140486180549945,
+      "loss": 1.343,
+      "step": 3519
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014045122683460184,
+      "loss": 0.8382,
+      "step": 3520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001404162672091771,
+      "loss": 0.9295,
+      "step": 3521
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014038130167877844,
+      "loss": 1.0989,
+      "step": 3522
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014034633024851458,
+      "loss": 1.1875,
+      "step": 3523
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014031135292349488,
+      "loss": 0.8728,
+      "step": 3524
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014027636970882974,
+      "loss": 1.0946,
+      "step": 3525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014024138060963028,
+      "loss": 1.1213,
+      "step": 3526
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001402063856310086,
+      "loss": 0.691,
+      "step": 3527
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014017138477807753,
+      "loss": 1.3236,
+      "step": 3528
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014013637805595093,
+      "loss": 1.0415,
+      "step": 3529
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014010136546974334,
+      "loss": 0.9921,
+      "step": 3530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014006634702457026,
+      "loss": 1.4104,
+      "step": 3531
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014003132272554804,
+      "loss": 1.2007,
+      "step": 3532
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001399962925777938,
+      "loss": 1.816,
+      "step": 3533
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001399612565864257,
+      "loss": 1.573,
+      "step": 3534
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013992621475656248,
+      "loss": 1.4306,
+      "step": 3535
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013989116709332402,
+      "loss": 1.3326,
+      "step": 3536
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013985611360183087,
+      "loss": 1.2478,
+      "step": 3537
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001398210542872045,
+      "loss": 0.5452,
+      "step": 3538
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013978598915456714,
+      "loss": 1.309,
+      "step": 3539
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000139750918209042,
+      "loss": 1.0561,
+      "step": 3540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001397158414557531,
+      "loss": 0.9566,
+      "step": 3541
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013968075889982527,
+      "loss": 1.0566,
+      "step": 3542
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013964567054638421,
+      "loss": 1.2655,
+      "step": 3543
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013961057640055646,
+      "loss": 1.3582,
+      "step": 3544
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001395754764674694,
+      "loss": 1.2134,
+      "step": 3545
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013954037075225128,
+      "loss": 1.3312,
+      "step": 3546
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013950525926003117,
+      "loss": 1.2884,
+      "step": 3547
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013947014199593903,
+      "loss": 0.6093,
+      "step": 3548
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001394350189651056,
+      "loss": 0.9272,
+      "step": 3549
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013939989017266248,
+      "loss": 1.4207,
+      "step": 3550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013936475562374215,
+      "loss": 0.8521,
+      "step": 3551
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001393296153234779,
+      "loss": 0.5036,
+      "step": 3552
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013929446927700385,
+      "loss": 1.4263,
+      "step": 3553
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013925931748945498,
+      "loss": 1.4125,
+      "step": 3554
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001392241599659671,
+      "loss": 0.7822,
+      "step": 3555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013918899671167687,
+      "loss": 1.4175,
+      "step": 3556
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013915382773172177,
+      "loss": 1.3847,
+      "step": 3557
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013911865303124015,
+      "loss": 0.7712,
+      "step": 3558
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013908347261537112,
+      "loss": 1.4287,
+      "step": 3559
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013904828648925475,
+      "loss": 0.4644,
+      "step": 3560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013901309465803177,
+      "loss": 1.0395,
+      "step": 3561
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013897789712684396,
+      "loss": 1.3671,
+      "step": 3562
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001389426939008337,
+      "loss": 1.6466,
+      "step": 3563
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013890748498514437,
+      "loss": 1.0372,
+      "step": 3564
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001388722703849202,
+      "loss": 1.705,
+      "step": 3565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013883705010530603,
+      "loss": 1.2982,
+      "step": 3566
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001388018241514478,
+      "loss": 1.1944,
+      "step": 3567
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013876659252849214,
+      "loss": 1.2937,
+      "step": 3568
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013873135524158648,
+      "loss": 0.8376,
+      "step": 3569
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013869611229587917,
+      "loss": 1.1962,
+      "step": 3570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013866086369651934,
+      "loss": 1.3351,
+      "step": 3571
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013862560944865687,
+      "loss": 1.2006,
+      "step": 3572
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013859034955744268,
+      "loss": 1.4989,
+      "step": 3573
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013855508402802825,
+      "loss": 1.2118,
+      "step": 3574
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013851981286556607,
+      "loss": 0.4807,
+      "step": 3575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013848453607520945,
+      "loss": 0.6013,
+      "step": 3576
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001384492536621123,
+      "loss": 1.1581,
+      "step": 3577
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001384139656314297,
+      "loss": 1.3619,
+      "step": 3578
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013837867198831724,
+      "loss": 0.9393,
+      "step": 3579
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013834337273793156,
+      "loss": 0.5533,
+      "step": 3580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001383080678854299,
+      "loss": 1.2159,
+      "step": 3581
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013827275743597057,
+      "loss": 1.4205,
+      "step": 3582
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013823744139471246,
+      "loss": 1.0369,
+      "step": 3583
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013820211976681542,
+      "loss": 1.5095,
+      "step": 3584
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001381667925574401,
+      "loss": 1.3422,
+      "step": 3585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013813145977174788,
+      "loss": 1.4429,
+      "step": 3586
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001380961214149011,
+      "loss": 0.4549,
+      "step": 3587
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013806077749206272,
+      "loss": 0.6307,
+      "step": 3588
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013802542800839674,
+      "loss": 0.8519,
+      "step": 3589
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001379900729690678,
+      "loss": 1.3006,
+      "step": 3590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013795471237924143,
+      "loss": 1.0391,
+      "step": 3591
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013791934624408388,
+      "loss": 1.3803,
+      "step": 3592
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013788397456876244,
+      "loss": 1.2344,
+      "step": 3593
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013784859735844484,
+      "loss": 1.0388,
+      "step": 3594
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001378132146183,
+      "loss": 0.9066,
+      "step": 3595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013777782635349738,
+      "loss": 1.1316,
+      "step": 3596
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001377424325692074,
+      "loss": 1.3287,
+      "step": 3597
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013770703327060116,
+      "loss": 1.1434,
+      "step": 3598
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001376716284628507,
+      "loss": 1.2714,
+      "step": 3599
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013763621815112876,
+      "loss": 1.2107,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000137600802340609,
+      "loss": 1.3984,
+      "step": 3601
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001375653810364657,
+      "loss": 0.6199,
+      "step": 3602
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013752995424387408,
+      "loss": 0.9082,
+      "step": 3603
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001374945219680102,
+      "loss": 1.2067,
+      "step": 3604
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001374590842140508,
+      "loss": 1.2264,
+      "step": 3605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013742364098717346,
+      "loss": 1.1417,
+      "step": 3606
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001373881922925566,
+      "loss": 0.9328,
+      "step": 3607
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001373527381353794,
+      "loss": 1.185,
+      "step": 3608
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001373172785208219,
+      "loss": 1.5797,
+      "step": 3609
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001372818134540648,
+      "loss": 1.1092,
+      "step": 3610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013724634294028974,
+      "loss": 1.1851,
+      "step": 3611
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013721086698467906,
+      "loss": 0.5268,
+      "step": 3612
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000137175385592416,
+      "loss": 1.2174,
+      "step": 3613
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013713989876868444,
+      "loss": 1.6316,
+      "step": 3614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013710440651866924,
+      "loss": 1.1366,
+      "step": 3615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001370689088475559,
+      "loss": 1.6406,
+      "step": 3616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013703340576053078,
+      "loss": 1.0895,
+      "step": 3617
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013699789726278103,
+      "loss": 1.1824,
+      "step": 3618
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013696238335949453,
+      "loss": 0.8864,
+      "step": 3619
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001369268640558601,
+      "loss": 0.8103,
+      "step": 3620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013689133935706712,
+      "loss": 1.2356,
+      "step": 3621
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013685580926830598,
+      "loss": 1.0849,
+      "step": 3622
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013682027379476774,
+      "loss": 1.1585,
+      "step": 3623
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013678473294164424,
+      "loss": 0.8699,
+      "step": 3624
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013674918671412818,
+      "loss": 1.3925,
+      "step": 3625
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013671363511741298,
+      "loss": 1.2676,
+      "step": 3626
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001366780781566929,
+      "loss": 1.377,
+      "step": 3627
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001366425158371629,
+      "loss": 1.0536,
+      "step": 3628
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013660694816401882,
+      "loss": 1.3811,
+      "step": 3629
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013657137514245718,
+      "loss": 0.6424,
+      "step": 3630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013653579677767535,
+      "loss": 1.1025,
+      "step": 3631
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013650021307487153,
+      "loss": 0.7957,
+      "step": 3632
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013646462403924458,
+      "loss": 1.3798,
+      "step": 3633
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013642902967599422,
+      "loss": 1.2555,
+      "step": 3634
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001363934299903209,
+      "loss": 0.9199,
+      "step": 3635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013635782498742588,
+      "loss": 0.9298,
+      "step": 3636
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001363222146725112,
+      "loss": 0.6121,
+      "step": 3637
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013628659905077966,
+      "loss": 1.21,
+      "step": 3638
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013625097812743485,
+      "loss": 1.2872,
+      "step": 3639
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013621535190768113,
+      "loss": 0.8742,
+      "step": 3640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001361797203967236,
+      "loss": 0.9254,
+      "step": 3641
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013614408359976818,
+      "loss": 1.2749,
+      "step": 3642
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013610844152202155,
+      "loss": 0.7999,
+      "step": 3643
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013607279416869117,
+      "loss": 1.3309,
+      "step": 3644
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001360371415449852,
+      "loss": 0.9522,
+      "step": 3645
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001360014836561127,
+      "loss": 0.9534,
+      "step": 3646
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013596582050728336,
+      "loss": 1.5418,
+      "step": 3647
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013593015210370776,
+      "loss": 1.1597,
+      "step": 3648
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013589447845059716,
+      "loss": 0.9895,
+      "step": 3649
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001358587995531636,
+      "loss": 0.6066,
+      "step": 3650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013582311541662,
+      "loss": 1.2392,
+      "step": 3651
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013578742604617981,
+      "loss": 1.1908,
+      "step": 3652
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013575173144705753,
+      "loss": 0.7014,
+      "step": 3653
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013571603162446818,
+      "loss": 1.5275,
+      "step": 3654
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013568032658362766,
+      "loss": 1.0488,
+      "step": 3655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013564461632975262,
+      "loss": 1.3107,
+      "step": 3656
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013560890086806052,
+      "loss": 1.4628,
+      "step": 3657
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013557318020376947,
+      "loss": 1.1564,
+      "step": 3658
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001355374543420984,
+      "loss": 0.6066,
+      "step": 3659
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000135501723288267,
+      "loss": 1.2224,
+      "step": 3660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013546598704749572,
+      "loss": 1.3415,
+      "step": 3661
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001354302456250058,
+      "loss": 1.6489,
+      "step": 3662
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013539449902601915,
+      "loss": 0.9287,
+      "step": 3663
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001353587472557585,
+      "loss": 1.4047,
+      "step": 3664
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013532299031944735,
+      "loss": 1.4458,
+      "step": 3665
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013528722822230991,
+      "loss": 1.313,
+      "step": 3666
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001352514609695711,
+      "loss": 0.9502,
+      "step": 3667
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013521568856645678,
+      "loss": 1.4589,
+      "step": 3668
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013517991101819337,
+      "loss": 1.3977,
+      "step": 3669
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013514412833000807,
+      "loss": 0.6395,
+      "step": 3670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013510834050712892,
+      "loss": 1.4347,
+      "step": 3671
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013507254755478463,
+      "loss": 1.2034,
+      "step": 3672
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013503674947820478,
+      "loss": 0.4839,
+      "step": 3673
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013500094628261948,
+      "loss": 1.3749,
+      "step": 3674
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001349651379732598,
+      "loss": 0.8743,
+      "step": 3675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013492932455535744,
+      "loss": 0.6117,
+      "step": 3676
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013489350603414487,
+      "loss": 1.1701,
+      "step": 3677
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013485768241485534,
+      "loss": 0.9872,
+      "step": 3678
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001348218537027228,
+      "loss": 1.4796,
+      "step": 3679
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000134786019902982,
+      "loss": 1.2812,
+      "step": 3680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013475018102086835,
+      "loss": 1.3577,
+      "step": 3681
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013471433706161812,
+      "loss": 1.0573,
+      "step": 3682
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013467848803046814,
+      "loss": 0.7829,
+      "step": 3683
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001346426339326562,
+      "loss": 1.3363,
+      "step": 3684
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013460677477342066,
+      "loss": 0.9776,
+      "step": 3685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013457091055800075,
+      "loss": 1.2423,
+      "step": 3686
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001345350412916363,
+      "loss": 0.8137,
+      "step": 3687
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013449916697956798,
+      "loss": 0.8719,
+      "step": 3688
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013446328762703713,
+      "loss": 1.4454,
+      "step": 3689
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001344274032392859,
+      "loss": 0.5323,
+      "step": 3690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013439151382155717,
+      "loss": 0.9173,
+      "step": 3691
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013435561937909446,
+      "loss": 1.4684,
+      "step": 3692
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001343197199171421,
+      "loss": 0.9015,
+      "step": 3693
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001342838154409452,
+      "loss": 1.2313,
+      "step": 3694
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013424790595574946,
+      "loss": 1.5394,
+      "step": 3695
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013421199146680145,
+      "loss": 0.9269,
+      "step": 3696
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001341760719793484,
+      "loss": 1.0218,
+      "step": 3697
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001341401474986383,
+      "loss": 1.2335,
+      "step": 3698
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013410421802991984,
+      "loss": 1.2997,
+      "step": 3699
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013406828357844243,
+      "loss": 0.8451,
+      "step": 3700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001340323441494563,
+      "loss": 1.0803,
+      "step": 3701
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013399639974821228,
+      "loss": 0.439,
+      "step": 3702
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013396045037996203,
+      "loss": 1.2997,
+      "step": 3703
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013392449604995786,
+      "loss": 1.0155,
+      "step": 3704
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013388853676345283,
+      "loss": 0.7999,
+      "step": 3705
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001338525725257008,
+      "loss": 1.1791,
+      "step": 3706
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001338166033419562,
+      "loss": 0.7138,
+      "step": 3707
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013378062921747433,
+      "loss": 1.2471,
+      "step": 3708
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001337446501575111,
+      "loss": 1.0484,
+      "step": 3709
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013370866616732322,
+      "loss": 0.8727,
+      "step": 3710
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013367267725216807,
+      "loss": 0.7081,
+      "step": 3711
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013363668341730383,
+      "loss": 1.2661,
+      "step": 3712
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013360068466798926,
+      "loss": 1.377,
+      "step": 3713
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013356468100948394,
+      "loss": 1.1248,
+      "step": 3714
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013352867244704819,
+      "loss": 0.9691,
+      "step": 3715
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013349265898594294,
+      "loss": 1.3725,
+      "step": 3716
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013345664063142996,
+      "loss": 1.3143,
+      "step": 3717
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001334206173887716,
+      "loss": 1.0811,
+      "step": 3718
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013338458926323105,
+      "loss": 1.0443,
+      "step": 3719
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013334855626007214,
+      "loss": 1.246,
+      "step": 3720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013331251838455947,
+      "loss": 0.649,
+      "step": 3721
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001332764756419582,
+      "loss": 1.2989,
+      "step": 3722
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013324042803753444,
+      "loss": 1.4853,
+      "step": 3723
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013320437557655485,
+      "loss": 1.1391,
+      "step": 3724
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001331683182642868,
+      "loss": 1.2195,
+      "step": 3725
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013313225610599844,
+      "loss": 0.3667,
+      "step": 3726
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013309618910695855,
+      "loss": 1.1153,
+      "step": 3727
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001330601172724367,
+      "loss": 0.8853,
+      "step": 3728
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013302404060770313,
+      "loss": 0.6364,
+      "step": 3729
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013298795911802875,
+      "loss": 1.1181,
+      "step": 3730
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013295187280868518,
+      "loss": 1.3747,
+      "step": 3731
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001329157816849449,
+      "loss": 0.8897,
+      "step": 3732
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001328796857520808,
+      "loss": 0.8726,
+      "step": 3733
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013284358501536675,
+      "loss": 1.3722,
+      "step": 3734
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013280747948007715,
+      "loss": 0.9962,
+      "step": 3735
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013277136915148714,
+      "loss": 1.1402,
+      "step": 3736
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013273525403487268,
+      "loss": 1.0032,
+      "step": 3737
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013269913413551022,
+      "loss": 0.5742,
+      "step": 3738
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013266300945867707,
+      "loss": 0.5179,
+      "step": 3739
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013262688000965117,
+      "loss": 1.4468,
+      "step": 3740
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013259074579371121,
+      "loss": 1.421,
+      "step": 3741
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013255460681613646,
+      "loss": 0.8619,
+      "step": 3742
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001325184630822071,
+      "loss": 1.4835,
+      "step": 3743
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001324823145972037,
+      "loss": 0.6787,
+      "step": 3744
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013244616136640785,
+      "loss": 1.5762,
+      "step": 3745
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013241000339510157,
+      "loss": 0.9901,
+      "step": 3746
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001323738406885677,
+      "loss": 1.0368,
+      "step": 3747
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013233767325208982,
+      "loss": 1.6614,
+      "step": 3748
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013230150109095202,
+      "loss": 1.1342,
+      "step": 3749
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013226532421043932,
+      "loss": 0.7954,
+      "step": 3750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001322291426158372,
+      "loss": 1.2704,
+      "step": 3751
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000132192956312432,
+      "loss": 1.5107,
+      "step": 3752
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013215676530551061,
+      "loss": 0.9708,
+      "step": 3753
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013212056960036078,
+      "loss": 0.8533,
+      "step": 3754
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013208436920227073,
+      "loss": 1.0834,
+      "step": 3755
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013204816411652956,
+      "loss": 1.3925,
+      "step": 3756
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013201195434842694,
+      "loss": 0.8597,
+      "step": 3757
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013197573990325322,
+      "loss": 1.5058,
+      "step": 3758
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013193952078629958,
+      "loss": 1.0597,
+      "step": 3759
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013190329700285767,
+      "loss": 1.3573,
+      "step": 3760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013186706855821995,
+      "loss": 0.4875,
+      "step": 3761
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001318308354576796,
+      "loss": 0.9443,
+      "step": 3762
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013179459770653034,
+      "loss": 1.186,
+      "step": 3763
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013175835531006663,
+      "loss": 1.1179,
+      "step": 3764
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001317221082735837,
+      "loss": 1.5637,
+      "step": 3765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013168585660237737,
+      "loss": 0.9765,
+      "step": 3766
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013164960030174414,
+      "loss": 1.2949,
+      "step": 3767
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013161333937698118,
+      "loss": 0.7877,
+      "step": 3768
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001315770738333863,
+      "loss": 1.0563,
+      "step": 3769
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013154080367625817,
+      "loss": 1.2296,
+      "step": 3770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001315045289108959,
+      "loss": 1.0805,
+      "step": 3771
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013146824954259943,
+      "loss": 1.1258,
+      "step": 3772
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013143196557666926,
+      "loss": 0.5726,
+      "step": 3773
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013139567701840664,
+      "loss": 1.1577,
+      "step": 3774
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013135938387311347,
+      "loss": 0.5257,
+      "step": 3775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013132308614609238,
+      "loss": 0.6608,
+      "step": 3776
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013128678384264651,
+      "loss": 0.7274,
+      "step": 3777
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013125047696807985,
+      "loss": 1.2195,
+      "step": 3778
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013121416552769693,
+      "loss": 0.9887,
+      "step": 3779
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013117784952680297,
+      "loss": 0.801,
+      "step": 3780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013114152897070392,
+      "loss": 1.6369,
+      "step": 3781
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013110520386470638,
+      "loss": 1.1073,
+      "step": 3782
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013106887421411752,
+      "loss": 0.7993,
+      "step": 3783
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013103254002424532,
+      "loss": 1.352,
+      "step": 3784
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013099620130039828,
+      "loss": 1.1044,
+      "step": 3785
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013095985804788564,
+      "loss": 0.468,
+      "step": 3786
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013092351027201731,
+      "loss": 0.8664,
+      "step": 3787
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013088715797810385,
+      "loss": 0.6726,
+      "step": 3788
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013085080117145647,
+      "loss": 1.3472,
+      "step": 3789
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000130814439857387,
+      "loss": 1.3129,
+      "step": 3790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000130778074041208,
+      "loss": 0.8495,
+      "step": 3791
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013074170372823267,
+      "loss": 1.5314,
+      "step": 3792
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013070532892377483,
+      "loss": 1.2118,
+      "step": 3793
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000130668949633149,
+      "loss": 0.7243,
+      "step": 3794
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013063256586167028,
+      "loss": 0.8921,
+      "step": 3795
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013059617761465461,
+      "loss": 1.2946,
+      "step": 3796
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013055978489741828,
+      "loss": 1.1888,
+      "step": 3797
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013052338771527857,
+      "loss": 0.7205,
+      "step": 3798
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013048698607355313,
+      "loss": 1.3434,
+      "step": 3799
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013045057997756044,
+      "loss": 1.1305,
+      "step": 3800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001304141694326196,
+      "loss": 0.7359,
+      "step": 3801
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013037775444405027,
+      "loss": 1.7284,
+      "step": 3802
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013034133501717282,
+      "loss": 0.5038,
+      "step": 3803
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013030491115730835,
+      "loss": 1.1725,
+      "step": 3804
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013026848286977843,
+      "loss": 1.4565,
+      "step": 3805
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001302320501599054,
+      "loss": 1.3117,
+      "step": 3806
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001301956130330123,
+      "loss": 1.0505,
+      "step": 3807
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013015917149442264,
+      "loss": 0.9665,
+      "step": 3808
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013012272554946067,
+      "loss": 1.052,
+      "step": 3809
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013008627520345138,
+      "loss": 0.982,
+      "step": 3810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001300498204617202,
+      "loss": 0.8615,
+      "step": 3811
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013001336132959342,
+      "loss": 0.7118,
+      "step": 3812
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001299768978123977,
+      "loss": 1.0667,
+      "step": 3813
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012994042991546067,
+      "loss": 1.2188,
+      "step": 3814
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012990395764411034,
+      "loss": 1.124,
+      "step": 3815
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012986748100367548,
+      "loss": 0.8955,
+      "step": 3816
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012983099999948545,
+      "loss": 1.4722,
+      "step": 3817
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012979451463687028,
+      "loss": 1.074,
+      "step": 3818
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012975802492116062,
+      "loss": 0.7211,
+      "step": 3819
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012972153085768776,
+      "loss": 1.1367,
+      "step": 3820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012968503245178363,
+      "loss": 1.22,
+      "step": 3821
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012964852970878077,
+      "loss": 0.822,
+      "step": 3822
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012961202263401243,
+      "loss": 1.3043,
+      "step": 3823
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012957551123281231,
+      "loss": 1.506,
+      "step": 3824
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000129538995510515,
+      "loss": 0.9151,
+      "step": 3825
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012950247547245555,
+      "loss": 1.3493,
+      "step": 3826
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012946595112396964,
+      "loss": 0.3777,
+      "step": 3827
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012942942247039365,
+      "loss": 1.2185,
+      "step": 3828
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001293928895170646,
+      "loss": 0.5529,
+      "step": 3829
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012935635226931998,
+      "loss": 0.8848,
+      "step": 3830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012931981073249814,
+      "loss": 1.6748,
+      "step": 3831
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001292832649119379,
+      "loss": 1.414,
+      "step": 3832
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001292467148129787,
+      "loss": 1.0858,
+      "step": 3833
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012921016044096075,
+      "loss": 0.5566,
+      "step": 3834
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012917360180122472,
+      "loss": 0.6948,
+      "step": 3835
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012913703889911197,
+      "loss": 1.4523,
+      "step": 3836
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001291004717399645,
+      "loss": 0.9195,
+      "step": 3837
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001290639003291249,
+      "loss": 1.5175,
+      "step": 3838
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001290273246719364,
+      "loss": 1.508,
+      "step": 3839
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012899074477374284,
+      "loss": 1.3325,
+      "step": 3840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012895416063988873,
+      "loss": 0.8126,
+      "step": 3841
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012891757227571907,
+      "loss": 1.0111,
+      "step": 3842
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012888097968657966,
+      "loss": 0.5474,
+      "step": 3843
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012884438287781672,
+      "loss": 0.6344,
+      "step": 3844
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012880778185477725,
+      "loss": 0.4696,
+      "step": 3845
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012877117662280882,
+      "loss": 1.2381,
+      "step": 3846
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012873456718725957,
+      "loss": 1.4949,
+      "step": 3847
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012869795355347825,
+      "loss": 1.1268,
+      "step": 3848
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012866133572681426,
+      "loss": 0.9646,
+      "step": 3849
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012862471371261763,
+      "loss": 1.09,
+      "step": 3850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000128588087516239,
+      "loss": 0.9914,
+      "step": 3851
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001285514571430296,
+      "loss": 0.8526,
+      "step": 3852
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001285148225983412,
+      "loss": 1.3871,
+      "step": 3853
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001284781838875263,
+      "loss": 1.2693,
+      "step": 3854
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012844154101593794,
+      "loss": 0.6944,
+      "step": 3855
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012840489398892983,
+      "loss": 1.3797,
+      "step": 3856
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012836824281185622,
+      "loss": 1.4996,
+      "step": 3857
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012833158749007197,
+      "loss": 1.3781,
+      "step": 3858
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012829492802893255,
+      "loss": 1.3251,
+      "step": 3859
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012825826443379413,
+      "loss": 1.5488,
+      "step": 3860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001282215967100133,
+      "loss": 0.601,
+      "step": 3861
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012818492486294742,
+      "loss": 0.7355,
+      "step": 3862
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001281482488979544,
+      "loss": 0.8628,
+      "step": 3863
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001281115688203927,
+      "loss": 1.031,
+      "step": 3864
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012807488463562145,
+      "loss": 1.2378,
+      "step": 3865
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012803819634900034,
+      "loss": 1.3034,
+      "step": 3866
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012800150396588967,
+      "loss": 0.7761,
+      "step": 3867
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012796480749165036,
+      "loss": 0.8493,
+      "step": 3868
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012792810693164388,
+      "loss": 1.4773,
+      "step": 3869
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012789140229123232,
+      "loss": 0.6084,
+      "step": 3870
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001278546935757784,
+      "loss": 1.2976,
+      "step": 3871
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001278179807906454,
+      "loss": 1.2303,
+      "step": 3872
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012778126394119717,
+      "loss": 1.1383,
+      "step": 3873
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012774454303279823,
+      "loss": 1.3575,
+      "step": 3874
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012770781807081363,
+      "loss": 1.0411,
+      "step": 3875
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012767108906060903,
+      "loss": 1.4538,
+      "step": 3876
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012763435600755065,
+      "loss": 0.7042,
+      "step": 3877
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001275976189170054,
+      "loss": 1.2181,
+      "step": 3878
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012756087779434064,
+      "loss": 1.3842,
+      "step": 3879
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012752413264492443,
+      "loss": 1.2853,
+      "step": 3880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012748738347412533,
+      "loss": 1.0012,
+      "step": 3881
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012745063028731267,
+      "loss": 1.3185,
+      "step": 3882
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012741387308985605,
+      "loss": 0.7657,
+      "step": 3883
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012737711188712597,
+      "loss": 0.5592,
+      "step": 3884
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012734034668449335,
+      "loss": 1.228,
+      "step": 3885
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001273035774873297,
+      "loss": 1.085,
+      "step": 3886
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001272668043010072,
+      "loss": 1.2977,
+      "step": 3887
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001272300271308985,
+      "loss": 1.0449,
+      "step": 3888
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001271932459823769,
+      "loss": 1.313,
+      "step": 3889
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012715646086081632,
+      "loss": 0.4676,
+      "step": 3890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012711967177159115,
+      "loss": 1.7213,
+      "step": 3891
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012708287872007643,
+      "loss": 1.0564,
+      "step": 3892
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001270460817116478,
+      "loss": 0.6869,
+      "step": 3893
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001270092807516814,
+      "loss": 1.2874,
+      "step": 3894
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012697247584555403,
+      "loss": 0.73,
+      "step": 3895
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012693566699864298,
+      "loss": 0.9663,
+      "step": 3896
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012689885421632622,
+      "loss": 1.2774,
+      "step": 3897
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012686203750398228,
+      "loss": 0.8044,
+      "step": 3898
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012682521686699008,
+      "loss": 1.1271,
+      "step": 3899
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012678839231072936,
+      "loss": 1.3018,
+      "step": 3900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012675156384058035,
+      "loss": 0.6887,
+      "step": 3901
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001267147314619238,
+      "loss": 1.8673,
+      "step": 3902
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000126677895180141,
+      "loss": 1.2158,
+      "step": 3903
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012664105500061402,
+      "loss": 0.8386,
+      "step": 3904
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012660421092872522,
+      "loss": 1.1601,
+      "step": 3905
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012656736296985774,
+      "loss": 1.1865,
+      "step": 3906
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012653051112939515,
+      "loss": 1.3203,
+      "step": 3907
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001264936554127217,
+      "loss": 0.6546,
+      "step": 3908
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012645679582522212,
+      "loss": 1.1288,
+      "step": 3909
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012641993237228177,
+      "loss": 1.3376,
+      "step": 3910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012638306505928653,
+      "loss": 1.7796,
+      "step": 3911
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012634619389162284,
+      "loss": 1.3444,
+      "step": 3912
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012630931887467773,
+      "loss": 0.5071,
+      "step": 3913
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001262724400138388,
+      "loss": 1.1646,
+      "step": 3914
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012623555731449416,
+      "loss": 1.1353,
+      "step": 3915
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012619867078203257,
+      "loss": 1.0298,
+      "step": 3916
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012616178042184327,
+      "loss": 1.2956,
+      "step": 3917
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012612488623931607,
+      "loss": 0.9363,
+      "step": 3918
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012608798823984135,
+      "loss": 0.9958,
+      "step": 3919
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012605108642881006,
+      "loss": 0.9917,
+      "step": 3920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012601418081161373,
+      "loss": 1.3784,
+      "step": 3921
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012597727139364436,
+      "loss": 1.4839,
+      "step": 3922
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012594035818029462,
+      "loss": 0.7667,
+      "step": 3923
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012590344117695761,
+      "loss": 1.3007,
+      "step": 3924
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012586652038902712,
+      "loss": 1.2937,
+      "step": 3925
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012582959582189732,
+      "loss": 1.1725,
+      "step": 3926
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012579266748096314,
+      "loss": 1.442,
+      "step": 3927
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001257557353716199,
+      "loss": 1.3008,
+      "step": 3928
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012571879949926354,
+      "loss": 1.1931,
+      "step": 3929
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012568185986929048,
+      "loss": 0.8674,
+      "step": 3930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001256449164870978,
+      "loss": 0.8808,
+      "step": 3931
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012560796935808313,
+      "loss": 1.0435,
+      "step": 3932
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012557101848764446,
+      "loss": 0.611,
+      "step": 3933
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012553406388118052,
+      "loss": 1.3384,
+      "step": 3934
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012549710554409055,
+      "loss": 1.2179,
+      "step": 3935
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012546014348177424,
+      "loss": 1.0661,
+      "step": 3936
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012542317769963191,
+      "loss": 0.7076,
+      "step": 3937
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012538620820306447,
+      "loss": 1.6529,
+      "step": 3938
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012534923499747324,
+      "loss": 0.9867,
+      "step": 3939
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001253122580882602,
+      "loss": 1.4065,
+      "step": 3940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012527527748082773,
+      "loss": 1.0828,
+      "step": 3941
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012523829318057894,
+      "loss": 0.4896,
+      "step": 3942
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001252013051929173,
+      "loss": 1.0392,
+      "step": 3943
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012516431352324698,
+      "loss": 1.3755,
+      "step": 3944
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012512731817697253,
+      "loss": 1.0923,
+      "step": 3945
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001250903191594992,
+      "loss": 1.35,
+      "step": 3946
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001250533164762326,
+      "loss": 0.8913,
+      "step": 3947
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012501631013257897,
+      "loss": 1.2068,
+      "step": 3948
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012497930013394517,
+      "loss": 1.3245,
+      "step": 3949
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012494228648573844,
+      "loss": 1.045,
+      "step": 3950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012490526919336662,
+      "loss": 1.0739,
+      "step": 3951
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001248682482622381,
+      "loss": 1.2728,
+      "step": 3952
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012483122369776176,
+      "loss": 1.4185,
+      "step": 3953
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012479419550534703,
+      "loss": 1.2285,
+      "step": 3954
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012475716369040392,
+      "loss": 1.4967,
+      "step": 3955
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012472012825834284,
+      "loss": 1.1295,
+      "step": 3956
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012468308921457494,
+      "loss": 1.2554,
+      "step": 3957
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001246460465645116,
+      "loss": 0.9896,
+      "step": 3958
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000124609000313565,
+      "loss": 1.4384,
+      "step": 3959
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012457195046714777,
+      "loss": 0.7182,
+      "step": 3960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012453489703067295,
+      "loss": 1.438,
+      "step": 3961
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001244978400095543,
+      "loss": 0.5727,
+      "step": 3962
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012446077940920583,
+      "loss": 1.256,
+      "step": 3963
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001244237152350424,
+      "loss": 1.3088,
+      "step": 3964
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012438664749247916,
+      "loss": 0.5113,
+      "step": 3965
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012434957618693185,
+      "loss": 1.1331,
+      "step": 3966
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001243125013238167,
+      "loss": 1.1492,
+      "step": 3967
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001242754229085506,
+      "loss": 1.0783,
+      "step": 3968
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001242383409465507,
+      "loss": 0.7413,
+      "step": 3969
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012420125544323498,
+      "loss": 1.3744,
+      "step": 3970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012416416640402165,
+      "loss": 1.4526,
+      "step": 3971
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001241270738343296,
+      "loss": 1.4017,
+      "step": 3972
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012408997773957826,
+      "loss": 1.1909,
+      "step": 3973
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012405287812518743,
+      "loss": 0.7405,
+      "step": 3974
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012401577499657752,
+      "loss": 0.8646,
+      "step": 3975
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012397866835916948,
+      "loss": 1.0619,
+      "step": 3976
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012394155821838474,
+      "loss": 1.4589,
+      "step": 3977
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012390444457964516,
+      "loss": 0.7605,
+      "step": 3978
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001238673274483733,
+      "loss": 1.4144,
+      "step": 3979
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012383020682999197,
+      "loss": 1.2738,
+      "step": 3980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012379308272992477,
+      "loss": 0.8555,
+      "step": 3981
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012375595515359567,
+      "loss": 1.0247,
+      "step": 3982
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012371882410642902,
+      "loss": 1.0849,
+      "step": 3983
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012368168959384998,
+      "loss": 0.6897,
+      "step": 3984
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012364455162128395,
+      "loss": 1.3527,
+      "step": 3985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012360741019415693,
+      "loss": 0.5199,
+      "step": 3986
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012357026531789546,
+      "loss": 1.4316,
+      "step": 3987
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012353311699792654,
+      "loss": 1.2847,
+      "step": 3988
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012349596523967765,
+      "loss": 0.6023,
+      "step": 3989
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012345881004857686,
+      "loss": 1.0845,
+      "step": 3990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001234216514300527,
+      "loss": 1.3402,
+      "step": 3991
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012338448938953414,
+      "loss": 1.2876,
+      "step": 3992
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001233473239324507,
+      "loss": 1.3709,
+      "step": 3993
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012331015506423238,
+      "loss": 1.4799,
+      "step": 3994
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012327298279030975,
+      "loss": 0.3999,
+      "step": 3995
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012323580711611385,
+      "loss": 1.375,
+      "step": 3996
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001231986280470761,
+      "loss": 1.2592,
+      "step": 3997
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012316144558862852,
+      "loss": 1.1876,
+      "step": 3998
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012312425974620372,
+      "loss": 1.1077,
+      "step": 3999
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012308707052523455,
+      "loss": 1.1406,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012304987793115458,
+      "loss": 0.8431,
+      "step": 4001
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001230126819693978,
+      "loss": 0.64,
+      "step": 4002
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012297548264539866,
+      "loss": 1.0662,
+      "step": 4003
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012293827996459213,
+      "loss": 0.9659,
+      "step": 4004
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012290107393241363,
+      "loss": 0.6847,
+      "step": 4005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012286386455429918,
+      "loss": 0.9492,
+      "step": 4006
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012282665183568518,
+      "loss": 1.6054,
+      "step": 4007
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012278943578200853,
+      "loss": 1.0535,
+      "step": 4008
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012275221639870666,
+      "loss": 0.5265,
+      "step": 4009
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012271499369121753,
+      "loss": 0.7706,
+      "step": 4010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012267776766497937,
+      "loss": 1.0243,
+      "step": 4011
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001226405383254312,
+      "loss": 0.6689,
+      "step": 4012
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012260330567801227,
+      "loss": 1.3953,
+      "step": 4013
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012256606972816246,
+      "loss": 1.1803,
+      "step": 4014
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012252883048132212,
+      "loss": 1.4768,
+      "step": 4015
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012249158794293193,
+      "loss": 0.7395,
+      "step": 4016
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012245434211843333,
+      "loss": 1.5032,
+      "step": 4017
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012241709301326795,
+      "loss": 1.1408,
+      "step": 4018
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001223798406328781,
+      "loss": 0.9841,
+      "step": 4019
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012234258498270646,
+      "loss": 0.4372,
+      "step": 4020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012230532606819628,
+      "loss": 0.7607,
+      "step": 4021
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012226806389479113,
+      "loss": 1.3477,
+      "step": 4022
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012223079846793527,
+      "loss": 0.5516,
+      "step": 4023
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012219352979307322,
+      "loss": 0.7105,
+      "step": 4024
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001221562578756502,
+      "loss": 1.3057,
+      "step": 4025
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012211898272111168,
+      "loss": 0.6625,
+      "step": 4026
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012208170433490375,
+      "loss": 0.9341,
+      "step": 4027
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012204442272247292,
+      "loss": 1.0329,
+      "step": 4028
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012200713788926617,
+      "loss": 1.578,
+      "step": 4029
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012196984984073099,
+      "loss": 1.5931,
+      "step": 4030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012193255858231522,
+      "loss": 0.8765,
+      "step": 4031
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012189526411946741,
+      "loss": 0.9779,
+      "step": 4032
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012185796645763627,
+      "loss": 1.1483,
+      "step": 4033
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012182066560227125,
+      "loss": 1.4374,
+      "step": 4034
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012178336155882209,
+      "loss": 1.1841,
+      "step": 4035
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012174605433273903,
+      "loss": 1.1838,
+      "step": 4036
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001217087439294729,
+      "loss": 1.4586,
+      "step": 4037
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012167143035447479,
+      "loss": 0.8007,
+      "step": 4038
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012163411361319641,
+      "loss": 1.1979,
+      "step": 4039
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001215967937110899,
+      "loss": 1.6058,
+      "step": 4040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001215594706536078,
+      "loss": 1.2279,
+      "step": 4041
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012152214444620313,
+      "loss": 0.9046,
+      "step": 4042
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012148481509432952,
+      "loss": 1.2842,
+      "step": 4043
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012144748260344079,
+      "loss": 1.2449,
+      "step": 4044
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012141014697899143,
+      "loss": 1.4118,
+      "step": 4045
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001213728082264363,
+      "loss": 1.0585,
+      "step": 4046
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012133546635123076,
+      "loss": 1.0841,
+      "step": 4047
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012129812135883058,
+      "loss": 1.1773,
+      "step": 4048
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012126077325469202,
+      "loss": 1.338,
+      "step": 4049
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012122342204427182,
+      "loss": 0.9266,
+      "step": 4050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012118606773302707,
+      "loss": 1.151,
+      "step": 4051
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012114871032641542,
+      "loss": 1.3598,
+      "step": 4052
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001211113498298949,
+      "loss": 0.566,
+      "step": 4053
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012107398624892408,
+      "loss": 1.5052,
+      "step": 4054
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012103661958896184,
+      "loss": 0.4298,
+      "step": 4055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001209992498554677,
+      "loss": 1.101,
+      "step": 4056
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012096187705390142,
+      "loss": 1.2507,
+      "step": 4057
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012092450118972339,
+      "loss": 0.6487,
+      "step": 4058
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012088712226839434,
+      "loss": 1.5557,
+      "step": 4059
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012084974029537545,
+      "loss": 1.2847,
+      "step": 4060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012081235527612844,
+      "loss": 0.7225,
+      "step": 4061
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012077496721611534,
+      "loss": 1.0912,
+      "step": 4062
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012073757612079872,
+      "loss": 1.3593,
+      "step": 4063
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012070018199564153,
+      "loss": 1.3616,
+      "step": 4064
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012066278484610727,
+      "loss": 1.135,
+      "step": 4065
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012062538467765974,
+      "loss": 0.7751,
+      "step": 4066
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012058798149576327,
+      "loss": 0.911,
+      "step": 4067
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012055057530588263,
+      "loss": 1.0457,
+      "step": 4068
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012051316611348298,
+      "loss": 0.9022,
+      "step": 4069
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012047575392402999,
+      "loss": 1.1107,
+      "step": 4070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012043833874298969,
+      "loss": 1.1628,
+      "step": 4071
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012040092057582859,
+      "loss": 1.2293,
+      "step": 4072
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012036349942801362,
+      "loss": 0.6844,
+      "step": 4073
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012032607530501218,
+      "loss": 1.2543,
+      "step": 4074
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012028864821229207,
+      "loss": 0.5595,
+      "step": 4075
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012025121815532154,
+      "loss": 1.0849,
+      "step": 4076
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012021378513956927,
+      "loss": 1.0162,
+      "step": 4077
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012017634917050433,
+      "loss": 1.3507,
+      "step": 4078
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012013891025359631,
+      "loss": 0.5428,
+      "step": 4079
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012010146839431514,
+      "loss": 1.4532,
+      "step": 4080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012006402359813124,
+      "loss": 0.9829,
+      "step": 4081
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012002657587051546,
+      "loss": 1.5024,
+      "step": 4082
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011998912521693906,
+      "loss": 1.0457,
+      "step": 4083
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011995167164287366,
+      "loss": 1.0229,
+      "step": 4084
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011991421515379149,
+      "loss": 1.0764,
+      "step": 4085
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011987675575516495,
+      "loss": 0.7188,
+      "step": 4086
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011983929345246712,
+      "loss": 0.8005,
+      "step": 4087
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011980182825117132,
+      "loss": 0.8368,
+      "step": 4088
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011976436015675139,
+      "loss": 0.4844,
+      "step": 4089
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011972688917468156,
+      "loss": 1.4512,
+      "step": 4090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011968941531043644,
+      "loss": 1.8935,
+      "step": 4091
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001196519385694912,
+      "loss": 1.3139,
+      "step": 4092
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011961445895732128,
+      "loss": 1.2888,
+      "step": 4093
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011957697647940261,
+      "loss": 1.1873,
+      "step": 4094
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011953949114121148,
+      "loss": 0.7973,
+      "step": 4095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011950200294822476,
+      "loss": 1.2097,
+      "step": 4096
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011946451190591948,
+      "loss": 1.1861,
+      "step": 4097
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011942701801977332,
+      "loss": 1.2225,
+      "step": 4098
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011938952129526426,
+      "loss": 1.2325,
+      "step": 4099
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001193520217378707,
+      "loss": 1.1501,
+      "step": 4100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001193145193530715,
+      "loss": 1.1332,
+      "step": 4101
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011927701414634584,
+      "loss": 1.1629,
+      "step": 4102
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011923950612317347,
+      "loss": 1.0657,
+      "step": 4103
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011920199528903442,
+      "loss": 0.6495,
+      "step": 4104
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011916448164940914,
+      "loss": 1.0496,
+      "step": 4105
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011912696520977856,
+      "loss": 1.356,
+      "step": 4106
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011908944597562394,
+      "loss": 1.1095,
+      "step": 4107
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011905192395242701,
+      "loss": 1.1926,
+      "step": 4108
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001190143991456699,
+      "loss": 1.5492,
+      "step": 4109
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011897687156083508,
+      "loss": 1.5076,
+      "step": 4110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011893934120340549,
+      "loss": 1.6338,
+      "step": 4111
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011890180807886453,
+      "loss": 0.6439,
+      "step": 4112
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011886427219269582,
+      "loss": 1.2533,
+      "step": 4113
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011882673355038361,
+      "loss": 1.1795,
+      "step": 4114
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011878919215741238,
+      "loss": 1.4225,
+      "step": 4115
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011875164801926706,
+      "loss": 1.3576,
+      "step": 4116
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011871410114143303,
+      "loss": 1.344,
+      "step": 4117
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011867655152939605,
+      "loss": 1.65,
+      "step": 4118
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011863899918864221,
+      "loss": 1.6781,
+      "step": 4119
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001186014441246581,
+      "loss": 0.4347,
+      "step": 4120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011856388634293063,
+      "loss": 1.2993,
+      "step": 4121
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011852632584894715,
+      "loss": 0.6858,
+      "step": 4122
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011848876264819539,
+      "loss": 0.7732,
+      "step": 4123
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011845119674616352,
+      "loss": 1.4392,
+      "step": 4124
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011841362814834,
+      "loss": 0.5556,
+      "step": 4125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001183760568602138,
+      "loss": 0.444,
+      "step": 4126
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001183384828872742,
+      "loss": 0.9293,
+      "step": 4127
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011830090623501089,
+      "loss": 1.0948,
+      "step": 4128
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011826332690891404,
+      "loss": 1.2138,
+      "step": 4129
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011822574491447402,
+      "loss": 0.9618,
+      "step": 4130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011818816025718181,
+      "loss": 1.5282,
+      "step": 4131
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011815057294252862,
+      "loss": 1.4792,
+      "step": 4132
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001181129829760061,
+      "loss": 0.398,
+      "step": 4133
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011807539036310635,
+      "loss": 0.5809,
+      "step": 4134
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011803779510932173,
+      "loss": 1.3813,
+      "step": 4135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011800019722014508,
+      "loss": 0.6366,
+      "step": 4136
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011796259670106961,
+      "loss": 1.3,
+      "step": 4137
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011792499355758888,
+      "loss": 1.1455,
+      "step": 4138
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011788738779519684,
+      "loss": 1.5922,
+      "step": 4139
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011784977941938788,
+      "loss": 1.2844,
+      "step": 4140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001178121684356567,
+      "loss": 1.1687,
+      "step": 4141
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011777455484949845,
+      "loss": 1.3793,
+      "step": 4142
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011773693866640856,
+      "loss": 0.4941,
+      "step": 4143
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011769931989188292,
+      "loss": 0.8135,
+      "step": 4144
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011766169853141782,
+      "loss": 0.7423,
+      "step": 4145
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011762407459050986,
+      "loss": 1.0068,
+      "step": 4146
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011758644807465607,
+      "loss": 1.077,
+      "step": 4147
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011754881898935374,
+      "loss": 1.1398,
+      "step": 4148
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011751118734010076,
+      "loss": 1.3862,
+      "step": 4149
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011747355313239514,
+      "loss": 0.794,
+      "step": 4150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011743591637173545,
+      "loss": 0.7053,
+      "step": 4151
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011739827706362057,
+      "loss": 0.7695,
+      "step": 4152
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011736063521354972,
+      "loss": 0.6086,
+      "step": 4153
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011732299082702252,
+      "loss": 1.1479,
+      "step": 4154
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011728534390953896,
+      "loss": 1.225,
+      "step": 4155
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011724769446659945,
+      "loss": 0.5351,
+      "step": 4156
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011721004250370467,
+      "loss": 1.3097,
+      "step": 4157
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011717238802635573,
+      "loss": 1.3821,
+      "step": 4158
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011713473104005407,
+      "loss": 1.507,
+      "step": 4159
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011709707155030163,
+      "loss": 1.195,
+      "step": 4160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011705940956260045,
+      "loss": 0.6555,
+      "step": 4161
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001170217450824532,
+      "loss": 1.4407,
+      "step": 4162
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011698407811536279,
+      "loss": 1.3071,
+      "step": 4163
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011694640866683249,
+      "loss": 1.4002,
+      "step": 4164
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011690873674236598,
+      "loss": 0.6759,
+      "step": 4165
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011687106234746721,
+      "loss": 0.6663,
+      "step": 4166
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011683338548764063,
+      "loss": 1.2781,
+      "step": 4167
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011679570616839096,
+      "loss": 1.3954,
+      "step": 4168
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011675802439522325,
+      "loss": 1.2878,
+      "step": 4169
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011672034017364299,
+      "loss": 0.7614,
+      "step": 4170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011668265350915604,
+      "loss": 0.7607,
+      "step": 4171
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011664496440726847,
+      "loss": 1.0806,
+      "step": 4172
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011660727287348684,
+      "loss": 1.2558,
+      "step": 4173
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011656957891331808,
+      "loss": 1.1499,
+      "step": 4174
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011653188253226939,
+      "loss": 0.6879,
+      "step": 4175
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011649418373584833,
+      "loss": 1.4108,
+      "step": 4176
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011645648252956283,
+      "loss": 1.2976,
+      "step": 4177
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011641877891892126,
+      "loss": 1.3204,
+      "step": 4178
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001163810729094322,
+      "loss": 1.2364,
+      "step": 4179
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011634336450660468,
+      "loss": 0.5048,
+      "step": 4180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011630565371594801,
+      "loss": 1.3407,
+      "step": 4181
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011626794054297195,
+      "loss": 0.7996,
+      "step": 4182
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011623022499318645,
+      "loss": 0.8976,
+      "step": 4183
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011619250707210195,
+      "loss": 0.9963,
+      "step": 4184
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001161547867852292,
+      "loss": 1.0117,
+      "step": 4185
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001161170641380792,
+      "loss": 1.0743,
+      "step": 4186
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011607933913616349,
+      "loss": 1.2007,
+      "step": 4187
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011604161178499375,
+      "loss": 1.7145,
+      "step": 4188
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011600388209008218,
+      "loss": 1.1512,
+      "step": 4189
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011596615005694114,
+      "loss": 0.5605,
+      "step": 4190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011592841569108346,
+      "loss": 0.8296,
+      "step": 4191
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011589067899802228,
+      "loss": 0.6066,
+      "step": 4192
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011585293998327111,
+      "loss": 1.2358,
+      "step": 4193
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011581519865234374,
+      "loss": 1.1333,
+      "step": 4194
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001157774550107543,
+      "loss": 1.0077,
+      "step": 4195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011573970906401734,
+      "loss": 1.0568,
+      "step": 4196
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011570196081764762,
+      "loss": 0.4328,
+      "step": 4197
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011566421027716037,
+      "loss": 1.4,
+      "step": 4198
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011562645744807106,
+      "loss": 1.4107,
+      "step": 4199
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011558870233589554,
+      "loss": 1.1714,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011555094494614995,
+      "loss": 1.8522,
+      "step": 4201
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011551318528435083,
+      "loss": 1.372,
+      "step": 4202
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011547542335601496,
+      "loss": 1.0987,
+      "step": 4203
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011543765916665958,
+      "loss": 1.0735,
+      "step": 4204
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011539989272180212,
+      "loss": 1.2985,
+      "step": 4205
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011536212402696041,
+      "loss": 0.7664,
+      "step": 4206
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011532435308765266,
+      "loss": 0.8839,
+      "step": 4207
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011528657990939726,
+      "loss": 1.2393,
+      "step": 4208
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011524880449771311,
+      "loss": 1.5338,
+      "step": 4209
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011521102685811929,
+      "loss": 0.937,
+      "step": 4210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001151732469961353,
+      "loss": 1.3069,
+      "step": 4211
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011513546491728085,
+      "loss": 1.3955,
+      "step": 4212
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011509768062707616,
+      "loss": 1.5551,
+      "step": 4213
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011505989413104154,
+      "loss": 1.2025,
+      "step": 4214
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011502210543469786,
+      "loss": 1.5162,
+      "step": 4215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001149843145435661,
+      "loss": 0.3825,
+      "step": 4216
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011494652146316771,
+      "loss": 0.5649,
+      "step": 4217
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001149087261990244,
+      "loss": 0.8356,
+      "step": 4218
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011487092875665817,
+      "loss": 1.2279,
+      "step": 4219
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011483312914159141,
+      "loss": 0.9862,
+      "step": 4220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001147953273593468,
+      "loss": 1.1937,
+      "step": 4221
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001147575234154473,
+      "loss": 1.3426,
+      "step": 4222
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001147197173154162,
+      "loss": 1.3119,
+      "step": 4223
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011468190906477719,
+      "loss": 1.0543,
+      "step": 4224
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001146440986690541,
+      "loss": 1.3651,
+      "step": 4225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011460628613377125,
+      "loss": 1.3501,
+      "step": 4226
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011456847146445318,
+      "loss": 0.58,
+      "step": 4227
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011453065466662476,
+      "loss": 1.5263,
+      "step": 4228
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011449283574581115,
+      "loss": 0.9542,
+      "step": 4229
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011445501470753788,
+      "loss": 0.7975,
+      "step": 4230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001144171915573307,
+      "loss": 1.2525,
+      "step": 4231
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011437936630071579,
+      "loss": 1.1413,
+      "step": 4232
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011434153894321953,
+      "loss": 1.2121,
+      "step": 4233
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011430370949036859,
+      "loss": 1.0867,
+      "step": 4234
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011426587794769012,
+      "loss": 0.7853,
+      "step": 4235
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011422804432071133,
+      "loss": 0.5766,
+      "step": 4236
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011419020861495994,
+      "loss": 1.0606,
+      "step": 4237
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011415237083596388,
+      "loss": 1.2003,
+      "step": 4238
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011411453098925137,
+      "loss": 0.9323,
+      "step": 4239
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011407668908035099,
+      "loss": 1.5182,
+      "step": 4240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011403884511479156,
+      "loss": 1.1021,
+      "step": 4241
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011400099909810227,
+      "loss": 1.2463,
+      "step": 4242
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011396315103581253,
+      "loss": 0.4175,
+      "step": 4243
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011392530093345213,
+      "loss": 0.9287,
+      "step": 4244
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011388744879655105,
+      "loss": 1.6537,
+      "step": 4245
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011384959463063973,
+      "loss": 1.1714,
+      "step": 4246
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001138117384412487,
+      "loss": 1.1484,
+      "step": 4247
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000113773880233909,
+      "loss": 1.894,
+      "step": 4248
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011373602001415182,
+      "loss": 1.0044,
+      "step": 4249
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011369815778750868,
+      "loss": 1.7639,
+      "step": 4250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011366029355951138,
+      "loss": 0.8863,
+      "step": 4251
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011362242733569204,
+      "loss": 1.1158,
+      "step": 4252
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001135845591215831,
+      "loss": 1.2951,
+      "step": 4253
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011354668892271725,
+      "loss": 1.3405,
+      "step": 4254
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011350881674462741,
+      "loss": 1.1435,
+      "step": 4255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001134709425928469,
+      "loss": 1.4152,
+      "step": 4256
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011343306647290933,
+      "loss": 0.6982,
+      "step": 4257
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011339518839034841,
+      "loss": 1.1772,
+      "step": 4258
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001133573083506984,
+      "loss": 1.4172,
+      "step": 4259
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011331942635949368,
+      "loss": 0.7253,
+      "step": 4260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011328154242226894,
+      "loss": 1.2132,
+      "step": 4261
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001132436565445592,
+      "loss": 1.2955,
+      "step": 4262
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011320576873189972,
+      "loss": 1.6288,
+      "step": 4263
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011316787898982606,
+      "loss": 1.3878,
+      "step": 4264
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011312998732387407,
+      "loss": 0.8712,
+      "step": 4265
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011309209373957985,
+      "loss": 0.5019,
+      "step": 4266
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011305419824247976,
+      "loss": 1.368,
+      "step": 4267
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011301630083811058,
+      "loss": 1.06,
+      "step": 4268
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011297840153200918,
+      "loss": 1.7112,
+      "step": 4269
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011294050032971287,
+      "loss": 1.1659,
+      "step": 4270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011290259723675908,
+      "loss": 1.5698,
+      "step": 4271
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011286469225868566,
+      "loss": 1.4303,
+      "step": 4272
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011282678540103066,
+      "loss": 1.2922,
+      "step": 4273
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011278887666933242,
+      "loss": 1.3941,
+      "step": 4274
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011275096606912954,
+      "loss": 1.2484,
+      "step": 4275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011271305360596094,
+      "loss": 1.1434,
+      "step": 4276
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011267513928536573,
+      "loss": 0.9927,
+      "step": 4277
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011263722311288334,
+      "loss": 0.9077,
+      "step": 4278
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011259930509405355,
+      "loss": 1.1052,
+      "step": 4279
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011256138523441623,
+      "loss": 1.4889,
+      "step": 4280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011252346353951169,
+      "loss": 1.2334,
+      "step": 4281
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011248554001488039,
+      "loss": 0.7027,
+      "step": 4282
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011244761466606313,
+      "loss": 1.2868,
+      "step": 4283
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011240968749860095,
+      "loss": 1.1576,
+      "step": 4284
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011237175851803514,
+      "loss": 0.9724,
+      "step": 4285
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001123338277299073,
+      "loss": 0.8655,
+      "step": 4286
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011229589513975924,
+      "loss": 0.9692,
+      "step": 4287
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011225796075313309,
+      "loss": 1.3466,
+      "step": 4288
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011222002457557117,
+      "loss": 1.3816,
+      "step": 4289
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011218208661261616,
+      "loss": 1.3304,
+      "step": 4290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011214414686981086,
+      "loss": 1.6115,
+      "step": 4291
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011210620535269847,
+      "loss": 1.4127,
+      "step": 4292
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001120682620668224,
+      "loss": 0.673,
+      "step": 4293
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011203031701772626,
+      "loss": 0.727,
+      "step": 4294
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011199237021095404,
+      "loss": 1.4936,
+      "step": 4295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011195442165204988,
+      "loss": 1.0303,
+      "step": 4296
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011191647134655821,
+      "loss": 1.3388,
+      "step": 4297
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011187851930002369,
+      "loss": 0.8102,
+      "step": 4298
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011184056551799136,
+      "loss": 1.066,
+      "step": 4299
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011180261000600626,
+      "loss": 0.7923,
+      "step": 4300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011176465276961403,
+      "loss": 1.3392,
+      "step": 4301
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011172669381436018,
+      "loss": 0.8766,
+      "step": 4302
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011168873314579079,
+      "loss": 1.0396,
+      "step": 4303
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011165077076945201,
+      "loss": 1.44,
+      "step": 4304
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011161280669089028,
+      "loss": 1.4071,
+      "step": 4305
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011157484091565231,
+      "loss": 0.7504,
+      "step": 4306
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011153687344928509,
+      "loss": 1.1211,
+      "step": 4307
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011149890429733578,
+      "loss": 1.3687,
+      "step": 4308
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011146093346535178,
+      "loss": 0.6937,
+      "step": 4309
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011142296095888087,
+      "loss": 1.1262,
+      "step": 4310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001113849867834709,
+      "loss": 1.4372,
+      "step": 4311
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011134701094467009,
+      "loss": 0.7007,
+      "step": 4312
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001113090334480268,
+      "loss": 1.0033,
+      "step": 4313
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001112710542990898,
+      "loss": 0.9322,
+      "step": 4314
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011123307350340785,
+      "loss": 1.6534,
+      "step": 4315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001111950910665302,
+      "loss": 1.5915,
+      "step": 4316
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011115710699400619,
+      "loss": 1.1631,
+      "step": 4317
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011111912129138547,
+      "loss": 1.0759,
+      "step": 4318
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011108113396421784,
+      "loss": 1.4372,
+      "step": 4319
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011104314501805343,
+      "loss": 1.2772,
+      "step": 4320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011100515445844262,
+      "loss": 1.0143,
+      "step": 4321
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001109671622909359,
+      "loss": 1.4839,
+      "step": 4322
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001109291685210841,
+      "loss": 0.464,
+      "step": 4323
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011089117315443829,
+      "loss": 0.6614,
+      "step": 4324
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011085317619654969,
+      "loss": 1.0919,
+      "step": 4325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011081517765296985,
+      "loss": 1.2478,
+      "step": 4326
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011077717752925046,
+      "loss": 1.2145,
+      "step": 4327
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011073917583094351,
+      "loss": 1.2899,
+      "step": 4328
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011070117256360122,
+      "loss": 1.3522,
+      "step": 4329
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011066316773277599,
+      "loss": 1.0487,
+      "step": 4330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011062516134402044,
+      "loss": 1.4965,
+      "step": 4331
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011058715340288752,
+      "loss": 1.3482,
+      "step": 4332
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011054914391493028,
+      "loss": 1.5091,
+      "step": 4333
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011051113288570209,
+      "loss": 1.2573,
+      "step": 4334
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001104731203207565,
+      "loss": 1.3789,
+      "step": 4335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001104351062256473,
+      "loss": 1.3051,
+      "step": 4336
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011039709060592845,
+      "loss": 1.4382,
+      "step": 4337
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011035907346715424,
+      "loss": 1.3029,
+      "step": 4338
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001103210548148791,
+      "loss": 1.4201,
+      "step": 4339
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011028303465465773,
+      "loss": 1.0996,
+      "step": 4340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011024501299204497,
+      "loss": 1.1957,
+      "step": 4341
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011020698983259594,
+      "loss": 1.1275,
+      "step": 4342
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011016896518186604,
+      "loss": 0.953,
+      "step": 4343
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001101309390454107,
+      "loss": 1.0963,
+      "step": 4344
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001100929114287858,
+      "loss": 0.8761,
+      "step": 4345
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011005488233754728,
+      "loss": 0.9139,
+      "step": 4346
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001100168517772513,
+      "loss": 1.4822,
+      "step": 4347
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010997881975345434,
+      "loss": 1.3561,
+      "step": 4348
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000109940786271713,
+      "loss": 1.3921,
+      "step": 4349
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010990275133758411,
+      "loss": 1.2603,
+      "step": 4350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010986471495662471,
+      "loss": 0.6063,
+      "step": 4351
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010982667713439214,
+      "loss": 1.2586,
+      "step": 4352
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010978863787644375,
+      "loss": 1.5288,
+      "step": 4353
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010975059718833734,
+      "loss": 1.8868,
+      "step": 4354
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010971255507563069,
+      "loss": 1.1957,
+      "step": 4355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000109674511543882,
+      "loss": 1.6328,
+      "step": 4356
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010963646659864955,
+      "loss": 1.206,
+      "step": 4357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010959842024549181,
+      "loss": 1.5468,
+      "step": 4358
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010956037248996753,
+      "loss": 0.5435,
+      "step": 4359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010952232333763568,
+      "loss": 1.5111,
+      "step": 4360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010948427279405533,
+      "loss": 1.1677,
+      "step": 4361
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010944622086478582,
+      "loss": 1.5213,
+      "step": 4362
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010940816755538671,
+      "loss": 1.3393,
+      "step": 4363
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001093701128714177,
+      "loss": 1.2444,
+      "step": 4364
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010933205681843879,
+      "loss": 0.8955,
+      "step": 4365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010929399940201004,
+      "loss": 1.3309,
+      "step": 4366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010925594062769185,
+      "loss": 0.8201,
+      "step": 4367
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010921788050104471,
+      "loss": 0.7883,
+      "step": 4368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010917981902762935,
+      "loss": 1.3214,
+      "step": 4369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010914175621300676,
+      "loss": 0.6145,
+      "step": 4370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010910369206273801,
+      "loss": 1.1544,
+      "step": 4371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010906562658238443,
+      "loss": 1.5098,
+      "step": 4372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010902755977750753,
+      "loss": 1.1695,
+      "step": 4373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001089894916536691,
+      "loss": 1.3544,
+      "step": 4374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010895142221643089,
+      "loss": 1.2156,
+      "step": 4375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001089133514713551,
+      "loss": 1.2511,
+      "step": 4376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000108875279424004,
+      "loss": 1.2266,
+      "step": 4377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010883720607994003,
+      "loss": 1.2577,
+      "step": 4378
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010879913144472589,
+      "loss": 0.6596,
+      "step": 4379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010876105552392439,
+      "loss": 1.0395,
+      "step": 4380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010872297832309863,
+      "loss": 1.3527,
+      "step": 4381
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010868489984781179,
+      "loss": 0.8755,
+      "step": 4382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001086468201036273,
+      "loss": 1.0062,
+      "step": 4383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010860873909610874,
+      "loss": 1.3532,
+      "step": 4384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010857065683081994,
+      "loss": 1.2096,
+      "step": 4385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010853257331332483,
+      "loss": 1.5998,
+      "step": 4386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010849448854918755,
+      "loss": 1.1029,
+      "step": 4387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010845640254397247,
+      "loss": 1.2737,
+      "step": 4388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010841831530324406,
+      "loss": 0.729,
+      "step": 4389
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010838022683256705,
+      "loss": 0.7786,
+      "step": 4390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001083421371375063,
+      "loss": 0.4961,
+      "step": 4391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010830404622362687,
+      "loss": 1.3978,
+      "step": 4392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010826595409649396,
+      "loss": 1.0472,
+      "step": 4393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010822786076167304,
+      "loss": 0.6097,
+      "step": 4394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010818976622472961,
+      "loss": 1.2743,
+      "step": 4395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010815167049122952,
+      "loss": 0.8992,
+      "step": 4396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010811357356673863,
+      "loss": 1.2396,
+      "step": 4397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010807547545682308,
+      "loss": 0.6004,
+      "step": 4398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010803737616704919,
+      "loss": 1.003,
+      "step": 4399
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010799927570298333,
+      "loss": 1.559,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001079611740701922,
+      "loss": 0.6676,
+      "step": 4401
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010792307127424255,
+      "loss": 0.9417,
+      "step": 4402
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010788496732070136,
+      "loss": 1.2644,
+      "step": 4403
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001078468622151358,
+      "loss": 0.4165,
+      "step": 4404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010780875596311315,
+      "loss": 0.9494,
+      "step": 4405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010777064857020085,
+      "loss": 0.9854,
+      "step": 4406
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001077325400419666,
+      "loss": 0.592,
+      "step": 4407
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010769443038397813,
+      "loss": 1.2563,
+      "step": 4408
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010765631960180346,
+      "loss": 1.4392,
+      "step": 4409
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010761820770101074,
+      "loss": 0.9922,
+      "step": 4410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001075800946871682,
+      "loss": 0.9488,
+      "step": 4411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010754198056584435,
+      "loss": 0.8928,
+      "step": 4412
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010750386534260781,
+      "loss": 1.3179,
+      "step": 4413
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010746574902302737,
+      "loss": 1.4245,
+      "step": 4414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010742763161267193,
+      "loss": 0.7002,
+      "step": 4415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010738951311711065,
+      "loss": 1.2137,
+      "step": 4416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001073513935419127,
+      "loss": 0.5881,
+      "step": 4417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010731327289264763,
+      "loss": 1.2031,
+      "step": 4418
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010727515117488489,
+      "loss": 0.6377,
+      "step": 4419
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010723702839419427,
+      "loss": 1.4469,
+      "step": 4420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010719890455614566,
+      "loss": 1.2417,
+      "step": 4421
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010716077966630906,
+      "loss": 1.3778,
+      "step": 4422
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010712265373025474,
+      "loss": 1.3145,
+      "step": 4423
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010708452675355299,
+      "loss": 1.5911,
+      "step": 4424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010704639874177432,
+      "loss": 1.344,
+      "step": 4425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010700826970048936,
+      "loss": 1.3428,
+      "step": 4426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010697013963526897,
+      "loss": 1.4272,
+      "step": 4427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010693200855168403,
+      "loss": 1.3863,
+      "step": 4428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010689387645530573,
+      "loss": 1.206,
+      "step": 4429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001068557433517052,
+      "loss": 0.7434,
+      "step": 4430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010681760924645392,
+      "loss": 1.176,
+      "step": 4431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010677947414512342,
+      "loss": 1.3612,
+      "step": 4432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010674133805328534,
+      "loss": 1.0143,
+      "step": 4433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010670320097651158,
+      "loss": 1.3365,
+      "step": 4434
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010666506292037408,
+      "loss": 1.4247,
+      "step": 4435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010662692389044497,
+      "loss": 1.1522,
+      "step": 4436
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010658878389229647,
+      "loss": 1.0155,
+      "step": 4437
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010655064293150105,
+      "loss": 1.1559,
+      "step": 4438
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010651250101363118,
+      "loss": 1.1447,
+      "step": 4439
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010647435814425961,
+      "loss": 1.1397,
+      "step": 4440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010643621432895909,
+      "loss": 1.5383,
+      "step": 4441
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010639806957330265,
+      "loss": 1.431,
+      "step": 4442
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010635992388286335,
+      "loss": 1.5004,
+      "step": 4443
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010632177726321442,
+      "loss": 0.5701,
+      "step": 4444
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010628362971992925,
+      "loss": 0.9518,
+      "step": 4445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010624548125858134,
+      "loss": 1.6611,
+      "step": 4446
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010620733188474432,
+      "loss": 1.0857,
+      "step": 4447
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010616918160399194,
+      "loss": 1.0086,
+      "step": 4448
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010613103042189819,
+      "loss": 1.3331,
+      "step": 4449
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010609287834403696,
+      "loss": 1.0163,
+      "step": 4450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010605472537598256,
+      "loss": 1.2165,
+      "step": 4451
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010601657152330921,
+      "loss": 1.2088,
+      "step": 4452
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010597841679159137,
+      "loss": 1.3216,
+      "step": 4453
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010594026118640359,
+      "loss": 1.0858,
+      "step": 4454
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010590210471332051,
+      "loss": 1.0956,
+      "step": 4455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000105863947377917,
+      "loss": 0.5437,
+      "step": 4456
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010582578918576795,
+      "loss": 1.4579,
+      "step": 4457
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010578763014244846,
+      "loss": 1.089,
+      "step": 4458
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010574947025353366,
+      "loss": 1.449,
+      "step": 4459
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010571130952459894,
+      "loss": 0.907,
+      "step": 4460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010567314796121961,
+      "loss": 1.4754,
+      "step": 4461
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010563498556897134,
+      "loss": 1.0834,
+      "step": 4462
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010559682235342976,
+      "loss": 0.8766,
+      "step": 4463
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010555865832017064,
+      "loss": 1.1015,
+      "step": 4464
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010552049347476994,
+      "loss": 1.7936,
+      "step": 4465
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010548232782280366,
+      "loss": 0.7802,
+      "step": 4466
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010544416136984793,
+      "loss": 1.6473,
+      "step": 4467
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010540599412147909,
+      "loss": 1.5922,
+      "step": 4468
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010536782608327346,
+      "loss": 0.7202,
+      "step": 4469
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010532965726080754,
+      "loss": 1.1386,
+      "step": 4470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010529148765965802,
+      "loss": 0.6178,
+      "step": 4471
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010525331728540151,
+      "loss": 0.9924,
+      "step": 4472
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010521514614361491,
+      "loss": 1.2265,
+      "step": 4473
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001051769742398752,
+      "loss": 1.3619,
+      "step": 4474
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010513880157975942,
+      "loss": 0.5312,
+      "step": 4475
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010510062816884469,
+      "loss": 0.4953,
+      "step": 4476
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001050624540127084,
+      "loss": 1.1461,
+      "step": 4477
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010502427911692787,
+      "loss": 0.5324,
+      "step": 4478
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001049861034870806,
+      "loss": 0.569,
+      "step": 4479
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010494792712874424,
+      "loss": 1.2699,
+      "step": 4480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010490975004749644,
+      "loss": 0.9572,
+      "step": 4481
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001048715722489151,
+      "loss": 1.2446,
+      "step": 4482
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010483339373857808,
+      "loss": 1.2546,
+      "step": 4483
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010479521452206345,
+      "loss": 1.2058,
+      "step": 4484
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001047570346049493,
+      "loss": 0.3622,
+      "step": 4485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010471885399281392,
+      "loss": 0.9939,
+      "step": 4486
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010468067269123558,
+      "loss": 0.708,
+      "step": 4487
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010464249070579278,
+      "loss": 1.1782,
+      "step": 4488
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000104604308042064,
+      "loss": 1.1752,
+      "step": 4489
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010456612470562795,
+      "loss": 1.2594,
+      "step": 4490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010452794070206328,
+      "loss": 1.236,
+      "step": 4491
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010448975603694886,
+      "loss": 1.0878,
+      "step": 4492
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010445157071586364,
+      "loss": 1.1861,
+      "step": 4493
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001044133847443866,
+      "loss": 1.5235,
+      "step": 4494
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010437519812809689,
+      "loss": 1.0779,
+      "step": 4495
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010433701087257373,
+      "loss": 0.9555,
+      "step": 4496
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010429882298339637,
+      "loss": 1.4661,
+      "step": 4497
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010426063446614426,
+      "loss": 0.9831,
+      "step": 4498
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010422244532639691,
+      "loss": 1.2294,
+      "step": 4499
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010418425556973386,
+      "loss": 0.4908,
+      "step": 4500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010414606520173477,
+      "loss": 1.584,
+      "step": 4501
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010410787422797948,
+      "loss": 0.9866,
+      "step": 4502
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010406968265404773,
+      "loss": 0.6424,
+      "step": 4503
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010403149048551957,
+      "loss": 1.6066,
+      "step": 4504
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001039932977279749,
+      "loss": 0.9533,
+      "step": 4505
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010395510438699394,
+      "loss": 0.9437,
+      "step": 4506
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010391691046815685,
+      "loss": 1.368,
+      "step": 4507
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010387871597704386,
+      "loss": 0.9404,
+      "step": 4508
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010384052091923542,
+      "loss": 1.4017,
+      "step": 4509
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010380232530031194,
+      "loss": 1.0484,
+      "step": 4510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010376412912585392,
+      "loss": 1.6886,
+      "step": 4511
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010372593240144197,
+      "loss": 0.5605,
+      "step": 4512
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010368773513265685,
+      "loss": 0.6276,
+      "step": 4513
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010364953732507924,
+      "loss": 0.4732,
+      "step": 4514
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010361133898429007,
+      "loss": 0.5106,
+      "step": 4515
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010357314011587018,
+      "loss": 1.3974,
+      "step": 4516
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010353494072540067,
+      "loss": 0.5943,
+      "step": 4517
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010349674081846255,
+      "loss": 1.2444,
+      "step": 4518
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010345854040063696,
+      "loss": 0.6145,
+      "step": 4519
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001034203394775052,
+      "loss": 1.0823,
+      "step": 4520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010338213805464852,
+      "loss": 0.6576,
+      "step": 4521
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010334393613764832,
+      "loss": 1.5262,
+      "step": 4522
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010330573373208604,
+      "loss": 0.6079,
+      "step": 4523
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010326753084354323,
+      "loss": 1.2305,
+      "step": 4524
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010322932747760142,
+      "loss": 0.8346,
+      "step": 4525
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010319112363984231,
+      "loss": 0.6902,
+      "step": 4526
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010315291933584763,
+      "loss": 0.6719,
+      "step": 4527
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010311471457119917,
+      "loss": 1.1186,
+      "step": 4528
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010307650935147879,
+      "loss": 1.2494,
+      "step": 4529
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010303830368226842,
+      "loss": 0.9235,
+      "step": 4530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010300009756915006,
+      "loss": 0.5979,
+      "step": 4531
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010296189101770579,
+      "loss": 1.0472,
+      "step": 4532
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010292368403351771,
+      "loss": 1.2446,
+      "step": 4533
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000102885476622168,
+      "loss": 1.468,
+      "step": 4534
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010284726878923901,
+      "loss": 0.5612,
+      "step": 4535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010280906054031287,
+      "loss": 1.1832,
+      "step": 4536
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010277085188097213,
+      "loss": 0.9726,
+      "step": 4537
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010273264281679913,
+      "loss": 0.5543,
+      "step": 4538
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010269443335337638,
+      "loss": 1.4925,
+      "step": 4539
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010265622349628643,
+      "loss": 1.5253,
+      "step": 4540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010261801325111192,
+      "loss": 1.5735,
+      "step": 4541
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010257980262343546,
+      "loss": 1.2558,
+      "step": 4542
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010254159161883981,
+      "loss": 1.2885,
+      "step": 4543
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010250338024290775,
+      "loss": 0.9623,
+      "step": 4544
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010246516850122208,
+      "loss": 1.3258,
+      "step": 4545
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010242695639936572,
+      "loss": 0.5972,
+      "step": 4546
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010238874394292157,
+      "loss": 0.79,
+      "step": 4547
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010235053113747266,
+      "loss": 0.9231,
+      "step": 4548
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000102312317988602,
+      "loss": 1.215,
+      "step": 4549
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010227410450189267,
+      "loss": 1.3583,
+      "step": 4550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010223589068292783,
+      "loss": 1.4309,
+      "step": 4551
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010219767653729069,
+      "loss": 1.3659,
+      "step": 4552
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010215946207056444,
+      "loss": 0.5725,
+      "step": 4553
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010212124728833238,
+      "loss": 1.1551,
+      "step": 4554
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010208303219617786,
+      "loss": 1.332,
+      "step": 4555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010204481679968418,
+      "loss": 0.4442,
+      "step": 4556
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010200660110443488,
+      "loss": 1.2234,
+      "step": 4557
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001019683851160133,
+      "loss": 0.8738,
+      "step": 4558
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010193016884000301,
+      "loss": 1.1026,
+      "step": 4559
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010189195228198756,
+      "loss": 1.3768,
+      "step": 4560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010185373544755052,
+      "loss": 1.4138,
+      "step": 4561
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001018155183422755,
+      "loss": 1.3767,
+      "step": 4562
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010177730097174621,
+      "loss": 1.2035,
+      "step": 4563
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010173908334154632,
+      "loss": 1.4801,
+      "step": 4564
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010170086545725956,
+      "loss": 1.0119,
+      "step": 4565
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010166264732446981,
+      "loss": 1.3504,
+      "step": 4566
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010162442894876076,
+      "loss": 0.2876,
+      "step": 4567
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010158621033571639,
+      "loss": 0.767,
+      "step": 4568
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010154799149092045,
+      "loss": 0.7557,
+      "step": 4569
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010150977241995694,
+      "loss": 1.1692,
+      "step": 4570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010147155312840983,
+      "loss": 1.0567,
+      "step": 4571
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010143333362186305,
+      "loss": 1.2101,
+      "step": 4572
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010139511390590069,
+      "loss": 0.912,
+      "step": 4573
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010135689398610675,
+      "loss": 1.5642,
+      "step": 4574
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010131867386806533,
+      "loss": 1.1356,
+      "step": 4575
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010128045355736048,
+      "loss": 1.0277,
+      "step": 4576
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010124223305957646,
+      "loss": 0.5879,
+      "step": 4577
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010120401238029728,
+      "loss": 1.2208,
+      "step": 4578
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010116579152510728,
+      "loss": 1.009,
+      "step": 4579
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010112757049959056,
+      "loss": 1.2088,
+      "step": 4580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001010893493093314,
+      "loss": 1.0818,
+      "step": 4581
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001010511279599141,
+      "loss": 0.6239,
+      "step": 4582
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010101290645692287,
+      "loss": 1.3737,
+      "step": 4583
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001009746848059421,
+      "loss": 0.7459,
+      "step": 4584
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010093646301255608,
+      "loss": 1.49,
+      "step": 4585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001008982410823492,
+      "loss": 1.5106,
+      "step": 4586
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010086001902090573,
+      "loss": 1.784,
+      "step": 4587
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010082179683381023,
+      "loss": 1.2833,
+      "step": 4588
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010078357452664695,
+      "loss": 1.4522,
+      "step": 4589
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001007453521050004,
+      "loss": 1.27,
+      "step": 4590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000100707129574455,
+      "loss": 1.5425,
+      "step": 4591
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010066890694059525,
+      "loss": 1.4996,
+      "step": 4592
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010063068420900556,
+      "loss": 1.5592,
+      "step": 4593
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010059246138527046,
+      "loss": 0.4767,
+      "step": 4594
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010055423847497445,
+      "loss": 1.3004,
+      "step": 4595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010051601548370206,
+      "loss": 1.2653,
+      "step": 4596
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010047779241703778,
+      "loss": 1.0936,
+      "step": 4597
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010043956928056617,
+      "loss": 1.3896,
+      "step": 4598
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001004013460798718,
+      "loss": 0.7083,
+      "step": 4599
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010036312282053917,
+      "loss": 0.8702,
+      "step": 4600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010032489950815291,
+      "loss": 1.02,
+      "step": 4601
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010028667614829757,
+      "loss": 1.6163,
+      "step": 4602
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010024845274655773,
+      "loss": 1.3256,
+      "step": 4603
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010021022930851795,
+      "loss": 1.1126,
+      "step": 4604
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010017200583976285,
+      "loss": 1.4823,
+      "step": 4605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010013378234587707,
+      "loss": 0.6245,
+      "step": 4606
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010009555883244515,
+      "loss": 1.2797,
+      "step": 4607
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001000573353050517,
+      "loss": 1.2689,
+      "step": 4608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010001911176928134,
+      "loss": 1.1147,
+      "step": 4609
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.998088823071867e-05,
+      "loss": 1.1462,
+      "step": 4610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.994266469494832e-05,
+      "loss": 1.5849,
+      "step": 4611
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.99044411675549e-05,
+      "loss": 1.0867,
+      "step": 4612
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.986621765412294e-05,
+      "loss": 0.6683,
+      "step": 4613
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.982799416023716e-05,
+      "loss": 1.3189,
+      "step": 4614
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.978977069148208e-05,
+      "loss": 0.6161,
+      "step": 4615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.97515472534423e-05,
+      "loss": 1.0485,
+      "step": 4616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.971332385170245e-05,
+      "loss": 1.333,
+      "step": 4617
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.967510049184711e-05,
+      "loss": 1.6283,
+      "step": 4618
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.963687717946083e-05,
+      "loss": 1.0939,
+      "step": 4619
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.959865392012823e-05,
+      "loss": 0.9348,
+      "step": 4620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.956043071943386e-05,
+      "loss": 1.472,
+      "step": 4621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.952220758296221e-05,
+      "loss": 1.485,
+      "step": 4622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.948398451629796e-05,
+      "loss": 1.0464,
+      "step": 4623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.944576152502557e-05,
+      "loss": 1.1826,
+      "step": 4624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.940753861472956e-05,
+      "loss": 1.6056,
+      "step": 4625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.936931579099446e-05,
+      "loss": 0.6314,
+      "step": 4626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.933109305940477e-05,
+      "loss": 1.156,
+      "step": 4627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.929287042554502e-05,
+      "loss": 1.2987,
+      "step": 4628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.92546478949996e-05,
+      "loss": 1.002,
+      "step": 4629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.921642547335308e-05,
+      "loss": 1.0009,
+      "step": 4630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.917820316618982e-05,
+      "loss": 0.4295,
+      "step": 4631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.913998097909425e-05,
+      "loss": 1.4767,
+      "step": 4632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.910175891765084e-05,
+      "loss": 0.5461,
+      "step": 4633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.906353698744393e-05,
+      "loss": 0.5088,
+      "step": 4634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.902531519405789e-05,
+      "loss": 1.1774,
+      "step": 4635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.898709354307715e-05,
+      "loss": 1.6742,
+      "step": 4636
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.894887204008595e-05,
+      "loss": 1.3176,
+      "step": 4637
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.891065069066861e-05,
+      "loss": 0.676,
+      "step": 4638
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.887242950040947e-05,
+      "loss": 0.48,
+      "step": 4639
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.883420847489275e-05,
+      "loss": 1.4992,
+      "step": 4640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.87959876197027e-05,
+      "loss": 0.9895,
+      "step": 4641
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.875776694042357e-05,
+      "loss": 1.4539,
+      "step": 4642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.871954644263954e-05,
+      "loss": 0.7125,
+      "step": 4643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.868132613193468e-05,
+      "loss": 0.7034,
+      "step": 4644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.864310601389327e-05,
+      "loss": 0.9984,
+      "step": 4645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.860488609409933e-05,
+      "loss": 1.1531,
+      "step": 4646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.856666637813693e-05,
+      "loss": 1.2298,
+      "step": 4647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.852844687159019e-05,
+      "loss": 1.2912,
+      "step": 4648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.849022758004307e-05,
+      "loss": 1.3066,
+      "step": 4649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.84520085090796e-05,
+      "loss": 1.3783,
+      "step": 4650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.841378966428365e-05,
+      "loss": 1.3331,
+      "step": 4651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.837557105123926e-05,
+      "loss": 1.3535,
+      "step": 4652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.833735267553024e-05,
+      "loss": 1.1687,
+      "step": 4653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.829913454274044e-05,
+      "loss": 1.7179,
+      "step": 4654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.82609166584537e-05,
+      "loss": 1.2177,
+      "step": 4655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.822269902825383e-05,
+      "loss": 1.6911,
+      "step": 4656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.818448165772451e-05,
+      "loss": 1.4084,
+      "step": 4657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.81462645524495e-05,
+      "loss": 0.9718,
+      "step": 4658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.810804771801249e-05,
+      "loss": 1.28,
+      "step": 4659
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.8069831159997e-05,
+      "loss": 1.1968,
+      "step": 4660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.803161488398673e-05,
+      "loss": 0.881,
+      "step": 4661
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.799339889556517e-05,
+      "loss": 0.5677,
+      "step": 4662
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.795518320031582e-05,
+      "loss": 1.2834,
+      "step": 4663
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.791696780382218e-05,
+      "loss": 0.6249,
+      "step": 4664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.787875271166767e-05,
+      "loss": 0.9507,
+      "step": 4665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.784053792943556e-05,
+      "loss": 1.3359,
+      "step": 4666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.780232346270935e-05,
+      "loss": 1.1823,
+      "step": 4667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.77641093170722e-05,
+      "loss": 1.4811,
+      "step": 4668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.772589549810734e-05,
+      "loss": 0.5585,
+      "step": 4669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.768768201139803e-05,
+      "loss": 0.9386,
+      "step": 4670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.764946886252738e-05,
+      "loss": 0.925,
+      "step": 4671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.761125605707842e-05,
+      "loss": 1.7084,
+      "step": 4672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.757304360063429e-05,
+      "loss": 1.084,
+      "step": 4673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.753483149877796e-05,
+      "loss": 1.2936,
+      "step": 4674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.74966197570923e-05,
+      "loss": 1.2243,
+      "step": 4675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.74584083811602e-05,
+      "loss": 1.0167,
+      "step": 4676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.742019737656457e-05,
+      "loss": 1.032,
+      "step": 4677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.738198674888812e-05,
+      "loss": 0.6062,
+      "step": 4678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.734377650371358e-05,
+      "loss": 1.1436,
+      "step": 4679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.730556664662363e-05,
+      "loss": 1.0563,
+      "step": 4680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.726735718320092e-05,
+      "loss": 1.3448,
+      "step": 4681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.722914811902791e-05,
+      "loss": 1.3306,
+      "step": 4682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.719093945968714e-05,
+      "loss": 1.2503,
+      "step": 4683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.715273121076106e-05,
+      "loss": 0.9808,
+      "step": 4684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.711452337783199e-05,
+      "loss": 1.4149,
+      "step": 4685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.70763159664823e-05,
+      "loss": 0.7711,
+      "step": 4686
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.703810898229425e-05,
+      "loss": 0.5427,
+      "step": 4687
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.699990243084995e-05,
+      "loss": 1.5245,
+      "step": 4688
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.696169631773162e-05,
+      "loss": 1.3065,
+      "step": 4689
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.692349064852125e-05,
+      "loss": 1.3153,
+      "step": 4690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.688528542880085e-05,
+      "loss": 1.1766,
+      "step": 4691
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.68470806641524e-05,
+      "loss": 1.2586,
+      "step": 4692
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.680887636015771e-05,
+      "loss": 1.0848,
+      "step": 4693
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.677067252239858e-05,
+      "loss": 1.239,
+      "step": 4694
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.67324691564568e-05,
+      "loss": 1.3194,
+      "step": 4695
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.669426626791399e-05,
+      "loss": 1.465,
+      "step": 4696
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.665606386235168e-05,
+      "loss": 1.4455,
+      "step": 4697
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.661786194535149e-05,
+      "loss": 1.0368,
+      "step": 4698
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.657966052249482e-05,
+      "loss": 1.3943,
+      "step": 4699
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.654145959936307e-05,
+      "loss": 1.3481,
+      "step": 4700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.650325918153749e-05,
+      "loss": 0.485,
+      "step": 4701
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.646505927459937e-05,
+      "loss": 0.3262,
+      "step": 4702
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.642685988412984e-05,
+      "loss": 0.9591,
+      "step": 4703
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.638866101570994e-05,
+      "loss": 0.9148,
+      "step": 4704
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.635046267492078e-05,
+      "loss": 1.3004,
+      "step": 4705
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.63122648673432e-05,
+      "loss": 1.2146,
+      "step": 4706
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.627406759855804e-05,
+      "loss": 0.6556,
+      "step": 4707
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.623587087414612e-05,
+      "loss": 1.0422,
+      "step": 4708
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.61976746996881e-05,
+      "loss": 1.3639,
+      "step": 4709
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.615947908076459e-05,
+      "loss": 1.245,
+      "step": 4710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.612128402295615e-05,
+      "loss": 0.8175,
+      "step": 4711
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.60830895318432e-05,
+      "loss": 1.3539,
+      "step": 4712
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.604489561300607e-05,
+      "loss": 1.1291,
+      "step": 4713
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.600670227202511e-05,
+      "loss": 1.4541,
+      "step": 4714
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.596850951448048e-05,
+      "loss": 1.26,
+      "step": 4715
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.593031734595226e-05,
+      "loss": 0.9753,
+      "step": 4716
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.589212577202053e-05,
+      "loss": 1.5795,
+      "step": 4717
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.585393479826525e-05,
+      "loss": 1.3213,
+      "step": 4718
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.581574443026613e-05,
+      "loss": 0.5028,
+      "step": 4719
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.577755467360311e-05,
+      "loss": 1.0137,
+      "step": 4720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.573936553385575e-05,
+      "loss": 1.2159,
+      "step": 4721
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.570117701660364e-05,
+      "loss": 1.1019,
+      "step": 4722
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.566298912742631e-05,
+      "loss": 0.7314,
+      "step": 4723
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.562480187190312e-05,
+      "loss": 1.3598,
+      "step": 4724
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.558661525561344e-05,
+      "loss": 0.6252,
+      "step": 4725
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.554842928413637e-05,
+      "loss": 1.0617,
+      "step": 4726
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.551024396305118e-05,
+      "loss": 1.5936,
+      "step": 4727
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.547205929793677e-05,
+      "loss": 1.6052,
+      "step": 4728
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.543387529437209e-05,
+      "loss": 0.6551,
+      "step": 4729
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.539569195793602e-05,
+      "loss": 0.4887,
+      "step": 4730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.535750929420726e-05,
+      "loss": 1.2218,
+      "step": 4731
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.531932730876443e-05,
+      "loss": 1.1014,
+      "step": 4732
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.52811460071861e-05,
+      "loss": 0.5101,
+      "step": 4733
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.524296539505073e-05,
+      "loss": 0.7521,
+      "step": 4734
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.520478547793659e-05,
+      "loss": 1.1217,
+      "step": 4735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.516660626142194e-05,
+      "loss": 1.13,
+      "step": 4736
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.512842775108493e-05,
+      "loss": 0.4239,
+      "step": 4737
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.509024995250356e-05,
+      "loss": 1.0475,
+      "step": 4738
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.505207287125578e-05,
+      "loss": 0.7472,
+      "step": 4739
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.501389651291944e-05,
+      "loss": 1.451,
+      "step": 4740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.497572088307214e-05,
+      "loss": 1.3069,
+      "step": 4741
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.493754598729162e-05,
+      "loss": 0.9357,
+      "step": 4742
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.489937183115532e-05,
+      "loss": 0.8946,
+      "step": 4743
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.48611984202406e-05,
+      "loss": 1.2386,
+      "step": 4744
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.482302576012481e-05,
+      "loss": 0.533,
+      "step": 4745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.47848538563851e-05,
+      "loss": 1.0707,
+      "step": 4746
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.47466827145985e-05,
+      "loss": 1.0815,
+      "step": 4747
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.470851234034201e-05,
+      "loss": 1.053,
+      "step": 4748
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.467034273919247e-05,
+      "loss": 1.2194,
+      "step": 4749
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.463217391672659e-05,
+      "loss": 0.6114,
+      "step": 4750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.459400587852093e-05,
+      "loss": 1.076,
+      "step": 4751
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.455583863015208e-05,
+      "loss": 0.9972,
+      "step": 4752
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.451767217719638e-05,
+      "loss": 1.392,
+      "step": 4753
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.447950652523007e-05,
+      "loss": 1.4358,
+      "step": 4754
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.444134167982938e-05,
+      "loss": 1.3885,
+      "step": 4755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.440317764657029e-05,
+      "loss": 1.056,
+      "step": 4756
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.436501443102869e-05,
+      "loss": 1.1763,
+      "step": 4757
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.432685203878041e-05,
+      "loss": 0.4751,
+      "step": 4758
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.428869047540111e-05,
+      "loss": 0.6383,
+      "step": 4759
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.425052974646635e-05,
+      "loss": 1.0649,
+      "step": 4760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.421236985755158e-05,
+      "loss": 1.0861,
+      "step": 4761
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.417421081423209e-05,
+      "loss": 0.8243,
+      "step": 4762
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.413605262208301e-05,
+      "loss": 1.2481,
+      "step": 4763
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.409789528667951e-05,
+      "loss": 1.1069,
+      "step": 4764
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.405973881359646e-05,
+      "loss": 0.9215,
+      "step": 4765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.402158320840864e-05,
+      "loss": 1.0484,
+      "step": 4766
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.39834284766908e-05,
+      "loss": 1.3096,
+      "step": 4767
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.394527462401745e-05,
+      "loss": 0.7485,
+      "step": 4768
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.390712165596302e-05,
+      "loss": 0.5454,
+      "step": 4769
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.386896957810185e-05,
+      "loss": 1.1547,
+      "step": 4770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.383081839600808e-05,
+      "loss": 1.172,
+      "step": 4771
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.379266811525569e-05,
+      "loss": 1.3335,
+      "step": 4772
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.375451874141869e-05,
+      "loss": 1.1112,
+      "step": 4773
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.371637028007077e-05,
+      "loss": 1.0611,
+      "step": 4774
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.36782227367856e-05,
+      "loss": 1.3135,
+      "step": 4775
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.364007611713666e-05,
+      "loss": 1.187,
+      "step": 4776
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.360193042669737e-05,
+      "loss": 1.2734,
+      "step": 4777
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.356378567104095e-05,
+      "loss": 0.6209,
+      "step": 4778
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.352564185574042e-05,
+      "loss": 0.5503,
+      "step": 4779
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.348749898636885e-05,
+      "loss": 1.2066,
+      "step": 4780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.3449357068499e-05,
+      "loss": 1.494,
+      "step": 4781
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.341121610770354e-05,
+      "loss": 1.2182,
+      "step": 4782
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.337307610955506e-05,
+      "loss": 1.1698,
+      "step": 4783
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.333493707962594e-05,
+      "loss": 1.5581,
+      "step": 4784
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.329679902348841e-05,
+      "loss": 1.6516,
+      "step": 4785
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.325866194671468e-05,
+      "loss": 1.5261,
+      "step": 4786
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.322052585487662e-05,
+      "loss": 1.5753,
+      "step": 4787
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.31823907535461e-05,
+      "loss": 1.3735,
+      "step": 4788
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.314425664829482e-05,
+      "loss": 1.2701,
+      "step": 4789
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.310612354469432e-05,
+      "loss": 0.5945,
+      "step": 4790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.306799144831598e-05,
+      "loss": 1.536,
+      "step": 4791
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.302986036473104e-05,
+      "loss": 1.0786,
+      "step": 4792
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.299173029951068e-05,
+      "loss": 1.2492,
+      "step": 4793
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.295360125822569e-05,
+      "loss": 0.5249,
+      "step": 4794
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.291547324644704e-05,
+      "loss": 1.0358,
+      "step": 4795
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.28773462697453e-05,
+      "loss": 1.4495,
+      "step": 4796
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.283922033369093e-05,
+      "loss": 0.9674,
+      "step": 4797
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.280109544385435e-05,
+      "loss": 1.4182,
+      "step": 4798
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.276297160580574e-05,
+      "loss": 1.2374,
+      "step": 4799
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.272484882511517e-05,
+      "loss": 0.6757,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.26867271073524e-05,
+      "loss": 1.0234,
+      "step": 4801
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.264860645808732e-05,
+      "loss": 0.992,
+      "step": 4802
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.26104868828894e-05,
+      "loss": 0.8628,
+      "step": 4803
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.257236838732808e-05,
+      "loss": 1.2785,
+      "step": 4804
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.253425097697266e-05,
+      "loss": 1.6597,
+      "step": 4805
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.24961346573922e-05,
+      "loss": 0.6307,
+      "step": 4806
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.245801943415563e-05,
+      "loss": 0.9101,
+      "step": 4807
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.24199053128318e-05,
+      "loss": 1.2274,
+      "step": 4808
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.238179229898931e-05,
+      "loss": 0.6627,
+      "step": 4809
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.234368039819656e-05,
+      "loss": 1.5196,
+      "step": 4810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.230556961602189e-05,
+      "loss": 0.7488,
+      "step": 4811
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.226745995803344e-05,
+      "loss": 1.3026,
+      "step": 4812
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.222935142979916e-05,
+      "loss": 1.319,
+      "step": 4813
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.219124403688687e-05,
+      "loss": 1.423,
+      "step": 4814
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.215313778486423e-05,
+      "loss": 0.9437,
+      "step": 4815
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.211503267929862e-05,
+      "loss": 1.1087,
+      "step": 4816
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.207692872575748e-05,
+      "loss": 1.5801,
+      "step": 4817
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.203882592980784e-05,
+      "loss": 1.3152,
+      "step": 4818
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.200072429701668e-05,
+      "loss": 1.3348,
+      "step": 4819
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.196262383295084e-05,
+      "loss": 1.1494,
+      "step": 4820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.192452454317693e-05,
+      "loss": 1.5415,
+      "step": 4821
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.188642643326138e-05,
+      "loss": 1.4276,
+      "step": 4822
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.18483295087705e-05,
+      "loss": 1.492,
+      "step": 4823
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.181023377527041e-05,
+      "loss": 0.8507,
+      "step": 4824
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.177213923832702e-05,
+      "loss": 0.6148,
+      "step": 4825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.173404590350605e-05,
+      "loss": 1.255,
+      "step": 4826
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.169595377637316e-05,
+      "loss": 0.7815,
+      "step": 4827
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.165786286249373e-05,
+      "loss": 1.5752,
+      "step": 4828
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.161977316743295e-05,
+      "loss": 1.3078,
+      "step": 4829
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.158168469675596e-05,
+      "loss": 1.2276,
+      "step": 4830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.154359745602758e-05,
+      "loss": 1.2083,
+      "step": 4831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.150551145081246e-05,
+      "loss": 1.4356,
+      "step": 4832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.14674266866752e-05,
+      "loss": 0.5934,
+      "step": 4833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.142934316918009e-05,
+      "loss": 0.9743,
+      "step": 4834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.139126090389126e-05,
+      "loss": 1.4155,
+      "step": 4835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.135317989637272e-05,
+      "loss": 0.9963,
+      "step": 4836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.131510015218825e-05,
+      "loss": 1.5002,
+      "step": 4837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.127702167690138e-05,
+      "loss": 1.0367,
+      "step": 4838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.123894447607563e-05,
+      "loss": 1.1457,
+      "step": 4839
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.120086855527415e-05,
+      "loss": 1.6308,
+      "step": 4840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.116279392005999e-05,
+      "loss": 1.4013,
+      "step": 4841
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.112472057599602e-05,
+      "loss": 0.4614,
+      "step": 4842
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.108664852864492e-05,
+      "loss": 0.8257,
+      "step": 4843
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.104857778356912e-05,
+      "loss": 0.4819,
+      "step": 4844
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.101050834633094e-05,
+      "loss": 0.5583,
+      "step": 4845
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.097244022249247e-05,
+      "loss": 0.6569,
+      "step": 4846
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.093437341761556e-05,
+      "loss": 1.3802,
+      "step": 4847
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.0896307937262e-05,
+      "loss": 1.1382,
+      "step": 4848
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.085824378699325e-05,
+      "loss": 1.316,
+      "step": 4849
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.082018097237066e-05,
+      "loss": 1.4031,
+      "step": 4850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.078211949895531e-05,
+      "loss": 1.3045,
+      "step": 4851
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.074405937230817e-05,
+      "loss": 1.1989,
+      "step": 4852
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.070600059799001e-05,
+      "loss": 1.6047,
+      "step": 4853
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.066794318156122e-05,
+      "loss": 1.209,
+      "step": 4854
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.062988712858231e-05,
+      "loss": 1.2493,
+      "step": 4855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.059183244461332e-05,
+      "loss": 1.2144,
+      "step": 4856
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.055377913521419e-05,
+      "loss": 0.5709,
+      "step": 4857
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.05157272059447e-05,
+      "loss": 1.1031,
+      "step": 4858
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.047767666236435e-05,
+      "loss": 1.7047,
+      "step": 4859
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.043962751003245e-05,
+      "loss": 1.0815,
+      "step": 4860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.040157975450823e-05,
+      "loss": 1.4615,
+      "step": 4861
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.03635334013505e-05,
+      "loss": 1.0538,
+      "step": 4862
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.032548845611801e-05,
+      "loss": 1.1978,
+      "step": 4863
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.028744492436932e-05,
+      "loss": 1.5347,
+      "step": 4864
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.024940281166271e-05,
+      "loss": 1.5475,
+      "step": 4865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.021136212355626e-05,
+      "loss": 1.2702,
+      "step": 4866
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.017332286560788e-05,
+      "loss": 1.2367,
+      "step": 4867
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.01352850433753e-05,
+      "loss": 1.4004,
+      "step": 4868
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.009724866241588e-05,
+      "loss": 1.0642,
+      "step": 4869
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.005921372828702e-05,
+      "loss": 0.7137,
+      "step": 4870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.002118024654567e-05,
+      "loss": 0.8178,
+      "step": 4871
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.998314822274869e-05,
+      "loss": 1.7564,
+      "step": 4872
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.994511766245275e-05,
+      "loss": 1.2487,
+      "step": 4873
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.990708857121421e-05,
+      "loss": 1.0974,
+      "step": 4874
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.986906095458933e-05,
+      "loss": 1.0775,
+      "step": 4875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.9831034818134e-05,
+      "loss": 1.3748,
+      "step": 4876
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.97930101674041e-05,
+      "loss": 1.3243,
+      "step": 4877
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.975498700795508e-05,
+      "loss": 1.2613,
+      "step": 4878
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.971696534534231e-05,
+      "loss": 1.3858,
+      "step": 4879
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.967894518512092e-05,
+      "loss": 1.1599,
+      "step": 4880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.964092653284577e-05,
+      "loss": 1.3755,
+      "step": 4881
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.960290939407155e-05,
+      "loss": 1.44,
+      "step": 4882
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.956489377435272e-05,
+      "loss": 1.1524,
+      "step": 4883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.952687967924353e-05,
+      "loss": 0.9812,
+      "step": 4884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.948886711429792e-05,
+      "loss": 1.4389,
+      "step": 4885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.945085608506973e-05,
+      "loss": 0.5435,
+      "step": 4886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.941284659711251e-05,
+      "loss": 1.2478,
+      "step": 4887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.937483865597957e-05,
+      "loss": 1.4812,
+      "step": 4888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.933683226722405e-05,
+      "loss": 1.2754,
+      "step": 4889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.929882743639883e-05,
+      "loss": 1.1837,
+      "step": 4890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.926082416905648e-05,
+      "loss": 0.9716,
+      "step": 4891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.922282247074957e-05,
+      "loss": 0.9569,
+      "step": 4892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.91848223470302e-05,
+      "loss": 0.5757,
+      "step": 4893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.914682380345033e-05,
+      "loss": 1.4764,
+      "step": 4894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.910882684556175e-05,
+      "loss": 1.2912,
+      "step": 4895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.907083147891591e-05,
+      "loss": 0.7152,
+      "step": 4896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.903283770906411e-05,
+      "loss": 1.9175,
+      "step": 4897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.89948455415574e-05,
+      "loss": 1.556,
+      "step": 4898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.895685498194659e-05,
+      "loss": 0.8436,
+      "step": 4899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.89188660357822e-05,
+      "loss": 0.7904,
+      "step": 4900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.888087870861457e-05,
+      "loss": 1.1848,
+      "step": 4901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.884289300599383e-05,
+      "loss": 1.1241,
+      "step": 4902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.880490893346982e-05,
+      "loss": 1.3957,
+      "step": 4903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.876692649659216e-05,
+      "loss": 0.8004,
+      "step": 4904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.872894570091024e-05,
+      "loss": 1.3515,
+      "step": 4905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.869096655197322e-05,
+      "loss": 0.9982,
+      "step": 4906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.865298905532995e-05,
+      "loss": 1.2588,
+      "step": 4907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.861501321652913e-05,
+      "loss": 1.118,
+      "step": 4908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.857703904111916e-05,
+      "loss": 0.9335,
+      "step": 4909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.853906653464821e-05,
+      "loss": 1.3572,
+      "step": 4910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.850109570266424e-05,
+      "loss": 1.3338,
+      "step": 4911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.846312655071495e-05,
+      "loss": 0.5733,
+      "step": 4912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.842515908434767e-05,
+      "loss": 0.654,
+      "step": 4913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.838719330910975e-05,
+      "loss": 1.3317,
+      "step": 4914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.834922923054805e-05,
+      "loss": 0.475,
+      "step": 4915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.831126685420924e-05,
+      "loss": 1.2126,
+      "step": 4916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.827330618563983e-05,
+      "loss": 1.1002,
+      "step": 4917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.823534723038602e-05,
+      "loss": 1.5289,
+      "step": 4918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.819738999399372e-05,
+      "loss": 0.8256,
+      "step": 4919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.815943448200867e-05,
+      "loss": 0.7107,
+      "step": 4920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.812148069997633e-05,
+      "loss": 1.3524,
+      "step": 4921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.80835286534418e-05,
+      "loss": 0.4543,
+      "step": 4922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.804557834795015e-05,
+      "loss": 1.3468,
+      "step": 4923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.800762978904599e-05,
+      "loss": 1.74,
+      "step": 4924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.796968298227377e-05,
+      "loss": 1.0938,
+      "step": 4925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.793173793317764e-05,
+      "loss": 0.893,
+      "step": 4926
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.789379464730157e-05,
+      "loss": 0.6804,
+      "step": 4927
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.78558531301892e-05,
+      "loss": 1.4153,
+      "step": 4928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.781791338738388e-05,
+      "loss": 0.9904,
+      "step": 4929
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.777997542442887e-05,
+      "loss": 0.683,
+      "step": 4930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.774203924686695e-05,
+      "loss": 1.2389,
+      "step": 4931
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.770410486024077e-05,
+      "loss": 1.3576,
+      "step": 4932
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.766617227009272e-05,
+      "loss": 1.2,
+      "step": 4933
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.762824148196489e-05,
+      "loss": 1.2952,
+      "step": 4934
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.759031250139906e-05,
+      "loss": 0.9633,
+      "step": 4935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.755238533393691e-05,
+      "loss": 1.3072,
+      "step": 4936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.751445998511964e-05,
+      "loss": 1.4508,
+      "step": 4937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.747653646048835e-05,
+      "loss": 0.6323,
+      "step": 4938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.743861476558379e-05,
+      "loss": 0.883,
+      "step": 4939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.740069490594649e-05,
+      "loss": 1.2285,
+      "step": 4940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.736277688711666e-05,
+      "loss": 1.4167,
+      "step": 4941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.732486071463431e-05,
+      "loss": 0.8705,
+      "step": 4942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.728694639403911e-05,
+      "loss": 1.3538,
+      "step": 4943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.724903393087045e-05,
+      "loss": 1.1167,
+      "step": 4944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.721112333066761e-05,
+      "loss": 1.2518,
+      "step": 4945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.717321459896936e-05,
+      "loss": 1.2132,
+      "step": 4946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.713530774131434e-05,
+      "loss": 0.5156,
+      "step": 4947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.709740276324093e-05,
+      "loss": 1.4302,
+      "step": 4948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.705949967028716e-05,
+      "loss": 1.2878,
+      "step": 4949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.702159846799085e-05,
+      "loss": 1.3911,
+      "step": 4950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.698369916188943e-05,
+      "loss": 1.4149,
+      "step": 4951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.694580175752025e-05,
+      "loss": 1.1815,
+      "step": 4952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.69079062604202e-05,
+      "loss": 1.6928,
+      "step": 4953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.687001267612595e-05,
+      "loss": 1.4901,
+      "step": 4954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.683212101017395e-05,
+      "loss": 1.375,
+      "step": 4955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.679423126810029e-05,
+      "loss": 1.3639,
+      "step": 4956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.675634345544079e-05,
+      "loss": 0.3408,
+      "step": 4957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.671845757773108e-05,
+      "loss": 0.8128,
+      "step": 4958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.668057364050636e-05,
+      "loss": 1.0611,
+      "step": 4959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.664269164930162e-05,
+      "loss": 1.2668,
+      "step": 4960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.660481160965161e-05,
+      "loss": 1.3519,
+      "step": 4961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.656693352709074e-05,
+      "loss": 1.5225,
+      "step": 4962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.65290574071531e-05,
+      "loss": 1.3122,
+      "step": 4963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.64911832553726e-05,
+      "loss": 1.0814,
+      "step": 4964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.645331107728281e-05,
+      "loss": 0.5607,
+      "step": 4965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.641544087841688e-05,
+      "loss": 0.6713,
+      "step": 4966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.637757266430797e-05,
+      "loss": 1.1882,
+      "step": 4967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.633970644048865e-05,
+      "loss": 0.8809,
+      "step": 4968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.630184221249136e-05,
+      "loss": 1.2467,
+      "step": 4969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.62639799858482e-05,
+      "loss": 0.8936,
+      "step": 4970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.622611976609102e-05,
+      "loss": 1.0268,
+      "step": 4971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.618826155875128e-05,
+      "loss": 1.4237,
+      "step": 4972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.615040536936029e-05,
+      "loss": 0.852,
+      "step": 4973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.611255120344898e-05,
+      "loss": 0.5399,
+      "step": 4974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.607469906654788e-05,
+      "loss": 1.1939,
+      "step": 4975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.603684896418748e-05,
+      "loss": 1.3827,
+      "step": 4976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.599900090189777e-05,
+      "loss": 1.3508,
+      "step": 4977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.596115488520846e-05,
+      "loss": 0.7682,
+      "step": 4978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.592331091964902e-05,
+      "loss": 0.7048,
+      "step": 4979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.588546901074864e-05,
+      "loss": 1.177,
+      "step": 4980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.584762916403617e-05,
+      "loss": 0.7242,
+      "step": 4981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.580979138504008e-05,
+      "loss": 1.3798,
+      "step": 4982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.577195567928869e-05,
+      "loss": 1.1196,
+      "step": 4983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.573412205230993e-05,
+      "loss": 1.1678,
+      "step": 4984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.569629050963141e-05,
+      "loss": 1.1231,
+      "step": 4985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.565846105678051e-05,
+      "loss": 1.289,
+      "step": 4986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.562063369928425e-05,
+      "loss": 1.1821,
+      "step": 4987
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.558280844266928e-05,
+      "loss": 0.5142,
+      "step": 4988
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.554498529246215e-05,
+      "loss": 1.0491,
+      "step": 4989
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.550716425418887e-05,
+      "loss": 1.3895,
+      "step": 4990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.546934533337526e-05,
+      "loss": 0.8103,
+      "step": 4991
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.543152853554685e-05,
+      "loss": 1.4903,
+      "step": 4992
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.539371386622877e-05,
+      "loss": 0.6686,
+      "step": 4993
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.535590133094591e-05,
+      "loss": 0.6976,
+      "step": 4994
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.531809093522284e-05,
+      "loss": 1.2285,
+      "step": 4995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.528028268458383e-05,
+      "loss": 1.5005,
+      "step": 4996
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.52424765845527e-05,
+      "loss": 1.1944,
+      "step": 4997
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.520467264065322e-05,
+      "loss": 1.2036,
+      "step": 4998
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.516687085840861e-05,
+      "loss": 0.788,
+      "step": 4999
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.512907124334184e-05,
+      "loss": 0.9861,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.509127380097563e-05,
+      "loss": 0.7836,
+      "step": 5001
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.505347853683231e-05,
+      "loss": 1.3453,
+      "step": 5002
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.501568545643395e-05,
+      "loss": 1.703,
+      "step": 5003
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.497789456530216e-05,
+      "loss": 0.5666,
+      "step": 5004
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.494010586895848e-05,
+      "loss": 0.5736,
+      "step": 5005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.49023193729239e-05,
+      "loss": 0.952,
+      "step": 5006
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.486453508271915e-05,
+      "loss": 0.574,
+      "step": 5007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.482675300386474e-05,
+      "loss": 0.9137,
+      "step": 5008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.478897314188072e-05,
+      "loss": 0.5994,
+      "step": 5009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.47511955022869e-05,
+      "loss": 0.9405,
+      "step": 5010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.471342009060275e-05,
+      "loss": 1.2054,
+      "step": 5011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.467564691234739e-05,
+      "loss": 1.42,
+      "step": 5012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.46378759730396e-05,
+      "loss": 1.2028,
+      "step": 5013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.460010727819792e-05,
+      "loss": 1.428,
+      "step": 5014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.456234083334045e-05,
+      "loss": 1.4916,
+      "step": 5015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.452457664398504e-05,
+      "loss": 0.7511,
+      "step": 5016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.44868147156492e-05,
+      "loss": 1.3447,
+      "step": 5017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.444905505385009e-05,
+      "loss": 1.3812,
+      "step": 5018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.441129766410447e-05,
+      "loss": 1.2759,
+      "step": 5019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.437354255192896e-05,
+      "loss": 1.1101,
+      "step": 5020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.433578972283965e-05,
+      "loss": 1.0381,
+      "step": 5021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.429803918235239e-05,
+      "loss": 1.3032,
+      "step": 5022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.426029093598269e-05,
+      "loss": 0.4681,
+      "step": 5023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.422254498924572e-05,
+      "loss": 1.0604,
+      "step": 5024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.418480134765627e-05,
+      "loss": 1.0888,
+      "step": 5025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.41470600167289e-05,
+      "loss": 1.1134,
+      "step": 5026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.410932100197774e-05,
+      "loss": 1.2337,
+      "step": 5027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.407158430891659e-05,
+      "loss": 0.5141,
+      "step": 5028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.40338499430589e-05,
+      "loss": 0.6219,
+      "step": 5029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.399611790991786e-05,
+      "loss": 1.0669,
+      "step": 5030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.395838821500626e-05,
+      "loss": 1.277,
+      "step": 5031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.392066086383652e-05,
+      "loss": 1.3042,
+      "step": 5032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.388293586192082e-05,
+      "loss": 0.8095,
+      "step": 5033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.384521321477085e-05,
+      "loss": 1.2209,
+      "step": 5034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.380749292789806e-05,
+      "loss": 1.0615,
+      "step": 5035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.376977500681358e-05,
+      "loss": 0.8545,
+      "step": 5036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.373205945702809e-05,
+      "loss": 1.1828,
+      "step": 5037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.369434628405199e-05,
+      "loss": 1.1675,
+      "step": 5038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.365663549339533e-05,
+      "loss": 1.1062,
+      "step": 5039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.361892709056783e-05,
+      "loss": 0.6596,
+      "step": 5040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.358122108107875e-05,
+      "loss": 1.1221,
+      "step": 5041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.354351747043718e-05,
+      "loss": 1.5798,
+      "step": 5042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.350581626415172e-05,
+      "loss": 0.6071,
+      "step": 5043
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.346811746773065e-05,
+      "loss": 1.1378,
+      "step": 5044
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.343042108668193e-05,
+      "loss": 1.32,
+      "step": 5045
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.339272712651317e-05,
+      "loss": 1.5133,
+      "step": 5046
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.335503559273154e-05,
+      "loss": 0.8241,
+      "step": 5047
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.331734649084398e-05,
+      "loss": 0.8813,
+      "step": 5048
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.327965982635702e-05,
+      "loss": 1.4069,
+      "step": 5049
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.324197560477673e-05,
+      "loss": 0.7377,
+      "step": 5050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.320429383160907e-05,
+      "loss": 1.4521,
+      "step": 5051
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.31666145123594e-05,
+      "loss": 1.0898,
+      "step": 5052
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.312893765253283e-05,
+      "loss": 1.0264,
+      "step": 5053
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.309126325763406e-05,
+      "loss": 1.2497,
+      "step": 5054
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.305359133316752e-05,
+      "loss": 1.2456,
+      "step": 5055
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.301592188463725e-05,
+      "loss": 1.3125,
+      "step": 5056
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.297825491754682e-05,
+      "loss": 1.0549,
+      "step": 5057
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.294059043739956e-05,
+      "loss": 0.6569,
+      "step": 5058
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.290292844969843e-05,
+      "loss": 1.2818,
+      "step": 5059
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.286526895994592e-05,
+      "loss": 1.7267,
+      "step": 5060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.28276119736443e-05,
+      "loss": 0.5053,
+      "step": 5061
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.278995749629538e-05,
+      "loss": 1.0081,
+      "step": 5062
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.275230553340056e-05,
+      "loss": 1.0746,
+      "step": 5063
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.271465609046106e-05,
+      "loss": 1.255,
+      "step": 5064
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.267700917297753e-05,
+      "loss": 1.2385,
+      "step": 5065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.26393647864503e-05,
+      "loss": 1.349,
+      "step": 5066
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.260172293637947e-05,
+      "loss": 1.124,
+      "step": 5067
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.256408362826458e-05,
+      "loss": 1.0144,
+      "step": 5068
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.252644686760487e-05,
+      "loss": 0.8455,
+      "step": 5069
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.248881265989927e-05,
+      "loss": 0.9467,
+      "step": 5070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.245118101064628e-05,
+      "loss": 0.8273,
+      "step": 5071
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.241355192534394e-05,
+      "loss": 0.6476,
+      "step": 5072
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.237592540949015e-05,
+      "loss": 0.7796,
+      "step": 5073
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.233830146858219e-05,
+      "loss": 1.3942,
+      "step": 5074
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.230068010811708e-05,
+      "loss": 1.393,
+      "step": 5075
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.226306133359147e-05,
+      "loss": 0.5162,
+      "step": 5076
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.222544515050159e-05,
+      "loss": 1.3866,
+      "step": 5077
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.218783156434335e-05,
+      "loss": 0.3802,
+      "step": 5078
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.215022058061213e-05,
+      "loss": 1.6505,
+      "step": 5079
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.21126122048032e-05,
+      "loss": 1.028,
+      "step": 5080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.207500644241117e-05,
+      "loss": 0.8364,
+      "step": 5081
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.203740329893041e-05,
+      "loss": 1.4342,
+      "step": 5082
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.199980277985493e-05,
+      "loss": 1.1875,
+      "step": 5083
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.196220489067829e-05,
+      "loss": 1.4484,
+      "step": 5084
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.192460963689366e-05,
+      "loss": 1.1016,
+      "step": 5085
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.18870170239939e-05,
+      "loss": 1.3629,
+      "step": 5086
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.184942705747141e-05,
+      "loss": 0.6204,
+      "step": 5087
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.18118397428182e-05,
+      "loss": 0.7584,
+      "step": 5088
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.177425508552599e-05,
+      "loss": 0.713,
+      "step": 5089
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.173667309108602e-05,
+      "loss": 1.1094,
+      "step": 5090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.16990937649891e-05,
+      "loss": 0.7523,
+      "step": 5091
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.166151711272583e-05,
+      "loss": 1.0885,
+      "step": 5092
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.162394313978624e-05,
+      "loss": 1.2417,
+      "step": 5093
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.158637185165999e-05,
+      "loss": 1.2348,
+      "step": 5094
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.15488032538365e-05,
+      "loss": 1.2256,
+      "step": 5095
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.151123735180462e-05,
+      "loss": 1.0275,
+      "step": 5096
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.147367415105285e-05,
+      "loss": 1.169,
+      "step": 5097
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.143611365706938e-05,
+      "loss": 1.3444,
+      "step": 5098
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.139855587534193e-05,
+      "loss": 0.6426,
+      "step": 5099
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.136100081135779e-05,
+      "loss": 1.1361,
+      "step": 5100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.132344847060396e-05,
+      "loss": 1.2204,
+      "step": 5101
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.128589885856698e-05,
+      "loss": 0.684,
+      "step": 5102
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.124835198073298e-05,
+      "loss": 1.7436,
+      "step": 5103
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.121080784258765e-05,
+      "loss": 0.5716,
+      "step": 5104
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.117326644961643e-05,
+      "loss": 1.2925,
+      "step": 5105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.11357278073042e-05,
+      "loss": 1.1208,
+      "step": 5106
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.10981919211355e-05,
+      "loss": 1.2837,
+      "step": 5107
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.106065879659452e-05,
+      "loss": 0.8173,
+      "step": 5108
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.102312843916497e-05,
+      "loss": 0.6469,
+      "step": 5109
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.098560085433013e-05,
+      "loss": 1.1546,
+      "step": 5110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.094807604757301e-05,
+      "loss": 1.1798,
+      "step": 5111
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.091055402437608e-05,
+      "loss": 1.079,
+      "step": 5112
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.087303479022145e-05,
+      "loss": 1.1091,
+      "step": 5113
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.083551835059087e-05,
+      "loss": 1.624,
+      "step": 5114
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.079800471096562e-05,
+      "loss": 1.2981,
+      "step": 5115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.076049387682651e-05,
+      "loss": 1.2909,
+      "step": 5116
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.072298585365417e-05,
+      "loss": 1.2848,
+      "step": 5117
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.068548064692855e-05,
+      "loss": 1.4832,
+      "step": 5118
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.064797826212932e-05,
+      "loss": 1.5546,
+      "step": 5119
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.061047870473577e-05,
+      "loss": 1.3325,
+      "step": 5120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.057298198022671e-05,
+      "loss": 1.1056,
+      "step": 5121
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.053548809408052e-05,
+      "loss": 0.6579,
+      "step": 5122
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.049799705177527e-05,
+      "loss": 1.1942,
+      "step": 5123
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.046050885878853e-05,
+      "loss": 0.7596,
+      "step": 5124
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.04230235205974e-05,
+      "loss": 1.1996,
+      "step": 5125
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.038554104267874e-05,
+      "loss": 1.1322,
+      "step": 5126
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.034806143050883e-05,
+      "loss": 0.7403,
+      "step": 5127
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.031058468956358e-05,
+      "loss": 1.4827,
+      "step": 5128
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.027311082531848e-05,
+      "loss": 1.1382,
+      "step": 5129
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.023563984324864e-05,
+      "loss": 1.4097,
+      "step": 5130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.019817174882873e-05,
+      "loss": 1.179,
+      "step": 5131
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.016070654753291e-05,
+      "loss": 0.4451,
+      "step": 5132
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.012324424483506e-05,
+      "loss": 1.0968,
+      "step": 5133
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.008578484620856e-05,
+      "loss": 0.9318,
+      "step": 5134
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.004832835712632e-05,
+      "loss": 1.2959,
+      "step": 5135
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.001087478306096e-05,
+      "loss": 1.2254,
+      "step": 5136
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.997342412948456e-05,
+      "loss": 0.7313,
+      "step": 5137
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.993597640186874e-05,
+      "loss": 0.8859,
+      "step": 5138
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.989853160568489e-05,
+      "loss": 1.5824,
+      "step": 5139
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.986108974640374e-05,
+      "loss": 0.9463,
+      "step": 5140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.98236508294957e-05,
+      "loss": 0.8362,
+      "step": 5141
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.978621486043077e-05,
+      "loss": 0.5671,
+      "step": 5142
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.974878184467849e-05,
+      "loss": 1.5,
+      "step": 5143
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.971135178770793e-05,
+      "loss": 0.342,
+      "step": 5144
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.967392469498782e-05,
+      "loss": 1.1656,
+      "step": 5145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.963650057198642e-05,
+      "loss": 0.9711,
+      "step": 5146
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.959907942417142e-05,
+      "loss": 1.6174,
+      "step": 5147
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.956166125701035e-05,
+      "loss": 0.948,
+      "step": 5148
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.952424607597005e-05,
+      "loss": 1.0617,
+      "step": 5149
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.948683388651703e-05,
+      "loss": 0.9402,
+      "step": 5150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.944942469411738e-05,
+      "loss": 1.2246,
+      "step": 5151
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.941201850423674e-05,
+      "loss": 1.3236,
+      "step": 5152
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.937461532234031e-05,
+      "loss": 1.1103,
+      "step": 5153
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.933721515389274e-05,
+      "loss": 1.1653,
+      "step": 5154
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.929981800435848e-05,
+      "loss": 1.3929,
+      "step": 5155
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.926242387920133e-05,
+      "loss": 1.2009,
+      "step": 5156
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.922503278388468e-05,
+      "loss": 1.063,
+      "step": 5157
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.918764472387159e-05,
+      "loss": 0.6409,
+      "step": 5158
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.915025970462457e-05,
+      "loss": 0.5869,
+      "step": 5159
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.911287773160567e-05,
+      "loss": 0.9067,
+      "step": 5160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.907549881027664e-05,
+      "loss": 1.3793,
+      "step": 5161
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.90381229460986e-05,
+      "loss": 1.1121,
+      "step": 5162
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.900075014453233e-05,
+      "loss": 1.5128,
+      "step": 5163
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.896338041103817e-05,
+      "loss": 1.4906,
+      "step": 5164
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.892601375107597e-05,
+      "loss": 1.0451,
+      "step": 5165
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.88886501701051e-05,
+      "loss": 0.6779,
+      "step": 5166
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.88512896735846e-05,
+      "loss": 1.4414,
+      "step": 5167
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.881393226697298e-05,
+      "loss": 1.0898,
+      "step": 5168
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.877657795572818e-05,
+      "loss": 1.2093,
+      "step": 5169
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.873922674530799e-05,
+      "loss": 1.2925,
+      "step": 5170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.870187864116943e-05,
+      "loss": 0.6818,
+      "step": 5171
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.866453364876925e-05,
+      "loss": 1.3029,
+      "step": 5172
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.862719177356371e-05,
+      "loss": 1.3099,
+      "step": 5173
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.85898530210086e-05,
+      "loss": 1.2295,
+      "step": 5174
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.855251739655922e-05,
+      "loss": 1.4541,
+      "step": 5175
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.851518490567052e-05,
+      "loss": 0.8875,
+      "step": 5176
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.847785555379688e-05,
+      "loss": 0.7665,
+      "step": 5177
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.844052934639224e-05,
+      "loss": 1.2452,
+      "step": 5178
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.840320628891012e-05,
+      "loss": 1.4721,
+      "step": 5179
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.83658863868036e-05,
+      "loss": 0.9695,
+      "step": 5180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.832856964552523e-05,
+      "loss": 1.2316,
+      "step": 5181
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.829125607052711e-05,
+      "loss": 0.6535,
+      "step": 5182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.825394566726098e-05,
+      "loss": 0.5789,
+      "step": 5183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.821663844117796e-05,
+      "loss": 0.7986,
+      "step": 5184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.817933439772877e-05,
+      "loss": 1.1004,
+      "step": 5185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.814203354236374e-05,
+      "loss": 0.7129,
+      "step": 5186
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.810473588053263e-05,
+      "loss": 1.3209,
+      "step": 5187
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.806744141768476e-05,
+      "loss": 1.1133,
+      "step": 5188
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.803015015926905e-05,
+      "loss": 1.4928,
+      "step": 5189
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.799286211073387e-05,
+      "loss": 1.4914,
+      "step": 5190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.795557727752709e-05,
+      "loss": 1.0973,
+      "step": 5191
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.791829566509627e-05,
+      "loss": 0.4735,
+      "step": 5192
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.788101727888834e-05,
+      "loss": 0.74,
+      "step": 5193
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.784374212434981e-05,
+      "loss": 0.9403,
+      "step": 5194
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.780647020692677e-05,
+      "loss": 1.1335,
+      "step": 5195
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.776920153206478e-05,
+      "loss": 0.4304,
+      "step": 5196
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.773193610520887e-05,
+      "loss": 0.5016,
+      "step": 5197
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.769467393180376e-05,
+      "loss": 1.1051,
+      "step": 5198
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.765741501729357e-05,
+      "loss": 1.5452,
+      "step": 5199
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.762015936712191e-05,
+      "loss": 1.2459,
+      "step": 5200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.758290698673206e-05,
+      "loss": 0.6497,
+      "step": 5201
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.75456578815667e-05,
+      "loss": 0.977,
+      "step": 5202
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.750841205706806e-05,
+      "loss": 1.3367,
+      "step": 5203
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.747116951867791e-05,
+      "loss": 1.5225,
+      "step": 5204
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.743393027183754e-05,
+      "loss": 0.5595,
+      "step": 5205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.739669432198778e-05,
+      "loss": 0.498,
+      "step": 5206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.735946167456883e-05,
+      "loss": 1.1643,
+      "step": 5207
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.732223233502065e-05,
+      "loss": 1.387,
+      "step": 5208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.728500630878252e-05,
+      "loss": 1.5472,
+      "step": 5209
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.724778360129334e-05,
+      "loss": 1.1064,
+      "step": 5210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.721056421799147e-05,
+      "loss": 0.7615,
+      "step": 5211
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.717334816431486e-05,
+      "loss": 1.3292,
+      "step": 5212
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.713613544570081e-05,
+      "loss": 0.4639,
+      "step": 5213
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.709892606758638e-05,
+      "loss": 0.7466,
+      "step": 5214
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.706172003540791e-05,
+      "loss": 1.2894,
+      "step": 5215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.702451735460137e-05,
+      "loss": 1.4618,
+      "step": 5216
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.698731803060222e-05,
+      "loss": 1.5468,
+      "step": 5217
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.695012206884543e-05,
+      "loss": 1.436,
+      "step": 5218
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.691292947476545e-05,
+      "loss": 1.1056,
+      "step": 5219
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.687574025379632e-05,
+      "loss": 1.6019,
+      "step": 5220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.683855441137149e-05,
+      "loss": 0.6048,
+      "step": 5221
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.68013719529239e-05,
+      "loss": 1.7849,
+      "step": 5222
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.676419288388619e-05,
+      "loss": 1.1139,
+      "step": 5223
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.672701720969026e-05,
+      "loss": 1.0847,
+      "step": 5224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.668984493576762e-05,
+      "loss": 1.9001,
+      "step": 5225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.665267606754934e-05,
+      "loss": 1.2556,
+      "step": 5226
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.661551061046591e-05,
+      "loss": 1.4431,
+      "step": 5227
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.657834856994734e-05,
+      "loss": 0.8575,
+      "step": 5228
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.654118995142315e-05,
+      "loss": 0.494,
+      "step": 5229
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.650403476032236e-05,
+      "loss": 1.3437,
+      "step": 5230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.646688300207351e-05,
+      "loss": 1.3973,
+      "step": 5231
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.642973468210456e-05,
+      "loss": 1.0988,
+      "step": 5232
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.639258980584309e-05,
+      "loss": 1.4194,
+      "step": 5233
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.635544837871611e-05,
+      "loss": 1.2971,
+      "step": 5234
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.631831040615003e-05,
+      "loss": 1.3869,
+      "step": 5235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.628117589357097e-05,
+      "loss": 0.6773,
+      "step": 5236
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.624404484640439e-05,
+      "loss": 1.412,
+      "step": 5237
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.620691727007523e-05,
+      "loss": 1.1726,
+      "step": 5238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.616979317000803e-05,
+      "loss": 1.4786,
+      "step": 5239
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.613267255162676e-05,
+      "loss": 1.4203,
+      "step": 5240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.609555542035485e-05,
+      "loss": 1.0551,
+      "step": 5241
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.60584417816153e-05,
+      "loss": 1.1599,
+      "step": 5242
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.602133164083056e-05,
+      "loss": 0.7049,
+      "step": 5243
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.598422500342248e-05,
+      "loss": 1.4316,
+      "step": 5244
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.59471218748126e-05,
+      "loss": 0.8788,
+      "step": 5245
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.591002226042179e-05,
+      "loss": 0.6906,
+      "step": 5246
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.587292616567038e-05,
+      "loss": 0.63,
+      "step": 5247
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.583583359597837e-05,
+      "loss": 0.9832,
+      "step": 5248
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.579874455676506e-05,
+      "loss": 1.3016,
+      "step": 5249
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.576165905344927e-05,
+      "loss": 0.7041,
+      "step": 5250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.572457709144942e-05,
+      "loss": 1.0481,
+      "step": 5251
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.568749867618332e-05,
+      "loss": 1.3361,
+      "step": 5252
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.56504238130682e-05,
+      "loss": 0.7082,
+      "step": 5253
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.561335250752088e-05,
+      "loss": 1.5164,
+      "step": 5254
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.557628476495762e-05,
+      "loss": 0.8392,
+      "step": 5255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.553922059079418e-05,
+      "loss": 1.2462,
+      "step": 5256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.550215999044575e-05,
+      "loss": 1.0409,
+      "step": 5257
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.546510296932706e-05,
+      "loss": 1.422,
+      "step": 5258
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.542804953285226e-05,
+      "loss": 1.3327,
+      "step": 5259
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.5390999686435e-05,
+      "loss": 1.2897,
+      "step": 5260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.535395343548842e-05,
+      "loss": 0.4664,
+      "step": 5261
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.531691078542511e-05,
+      "loss": 1.2061,
+      "step": 5262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.527987174165715e-05,
+      "loss": 1.1146,
+      "step": 5263
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.524283630959611e-05,
+      "loss": 1.0716,
+      "step": 5264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.5205804494653e-05,
+      "loss": 1.5664,
+      "step": 5265
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.516877630223825e-05,
+      "loss": 1.3468,
+      "step": 5266
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.513175173776195e-05,
+      "loss": 0.841,
+      "step": 5267
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.509473080663341e-05,
+      "loss": 1.2619,
+      "step": 5268
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.505771351426159e-05,
+      "loss": 0.8841,
+      "step": 5269
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.502069986605485e-05,
+      "loss": 1.0008,
+      "step": 5270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.498368986742104e-05,
+      "loss": 1.4164,
+      "step": 5271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.494668352376742e-05,
+      "loss": 0.9726,
+      "step": 5272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.490968084050082e-05,
+      "loss": 1.5706,
+      "step": 5273
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.487268182302749e-05,
+      "loss": 1.0868,
+      "step": 5274
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.483568647675302e-05,
+      "loss": 0.766,
+      "step": 5275
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.47986948070827e-05,
+      "loss": 1.615,
+      "step": 5276
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.476170681942109e-05,
+      "loss": 0.8306,
+      "step": 5277
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.47247225191723e-05,
+      "loss": 1.1135,
+      "step": 5278
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.468774191173983e-05,
+      "loss": 1.1908,
+      "step": 5279
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.465076500252676e-05,
+      "loss": 1.2372,
+      "step": 5280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.461379179693557e-05,
+      "loss": 1.4697,
+      "step": 5281
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.457682230036808e-05,
+      "loss": 1.5119,
+      "step": 5282
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.45398565182258e-05,
+      "loss": 1.0594,
+      "step": 5283
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.45028944559095e-05,
+      "loss": 1.9135,
+      "step": 5284
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.446593611881949e-05,
+      "loss": 1.0747,
+      "step": 5285
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.442898151235556e-05,
+      "loss": 0.7858,
+      "step": 5286
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.439203064191693e-05,
+      "loss": 1.4125,
+      "step": 5287
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.435508351290218e-05,
+      "loss": 1.4036,
+      "step": 5288
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.431814013070954e-05,
+      "loss": 1.3763,
+      "step": 5289
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.428120050073651e-05,
+      "loss": 1.0895,
+      "step": 5290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.424426462838012e-05,
+      "loss": 0.9106,
+      "step": 5291
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.420733251903687e-05,
+      "loss": 1.0202,
+      "step": 5292
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.417040417810268e-05,
+      "loss": 1.1165,
+      "step": 5293
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.41334796109729e-05,
+      "loss": 1.4185,
+      "step": 5294
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.409655882304238e-05,
+      "loss": 0.9724,
+      "step": 5295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.405964181970542e-05,
+      "loss": 0.7794,
+      "step": 5296
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.402272860635562e-05,
+      "loss": 0.4433,
+      "step": 5297
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.398581918838629e-05,
+      "loss": 1.2563,
+      "step": 5298
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.394891357118996e-05,
+      "loss": 1.1234,
+      "step": 5299
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.391201176015866e-05,
+      "loss": 1.0666,
+      "step": 5300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.387511376068394e-05,
+      "loss": 1.3723,
+      "step": 5301
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.383821957815676e-05,
+      "loss": 1.2611,
+      "step": 5302
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.380132921796747e-05,
+      "loss": 1.169,
+      "step": 5303
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.376444268550584e-05,
+      "loss": 1.0343,
+      "step": 5304
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.372755998616124e-05,
+      "loss": 0.6349,
+      "step": 5305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.369068112532229e-05,
+      "loss": 1.4889,
+      "step": 5306
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.365380610837717e-05,
+      "loss": 1.2249,
+      "step": 5307
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.361693494071349e-05,
+      "loss": 1.3907,
+      "step": 5308
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.358006762771827e-05,
+      "loss": 1.1953,
+      "step": 5309
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.354320417477787e-05,
+      "loss": 1.3658,
+      "step": 5310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.350634458727832e-05,
+      "loss": 1.2608,
+      "step": 5311
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.346948887060487e-05,
+      "loss": 1.3258,
+      "step": 5312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.34326370301423e-05,
+      "loss": 1.2963,
+      "step": 5313
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.33957890712748e-05,
+      "loss": 1.4485,
+      "step": 5314
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.335894499938602e-05,
+      "loss": 1.4932,
+      "step": 5315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.332210481985897e-05,
+      "loss": 0.7159,
+      "step": 5316
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.328526853807623e-05,
+      "loss": 0.9934,
+      "step": 5317
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.324843615941968e-05,
+      "loss": 1.1837,
+      "step": 5318
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.321160768927062e-05,
+      "loss": 1.1016,
+      "step": 5319
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.317478313300995e-05,
+      "loss": 1.03,
+      "step": 5320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.313796249601778e-05,
+      "loss": 1.4876,
+      "step": 5321
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.310114578367378e-05,
+      "loss": 1.0009,
+      "step": 5322
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.306433300135702e-05,
+      "loss": 0.6216,
+      "step": 5323
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.3027524154446e-05,
+      "loss": 0.7453,
+      "step": 5324
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.299071924831861e-05,
+      "loss": 1.0175,
+      "step": 5325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.295391828835221e-05,
+      "loss": 1.3752,
+      "step": 5326
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.29171212799236e-05,
+      "loss": 1.0331,
+      "step": 5327
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.288032822840887e-05,
+      "loss": 0.9869,
+      "step": 5328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.284353913918369e-05,
+      "loss": 0.9926,
+      "step": 5329
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.280675401762309e-05,
+      "loss": 1.36,
+      "step": 5330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.276997286910152e-05,
+      "loss": 1.2726,
+      "step": 5331
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.273319569899281e-05,
+      "loss": 1.2545,
+      "step": 5332
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.269642251267033e-05,
+      "loss": 1.0031,
+      "step": 5333
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.26596533155067e-05,
+      "loss": 1.0765,
+      "step": 5334
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.262288811287405e-05,
+      "loss": 1.1658,
+      "step": 5335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.258612691014397e-05,
+      "loss": 0.6222,
+      "step": 5336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.25493697126874e-05,
+      "loss": 1.3594,
+      "step": 5337
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.251261652587466e-05,
+      "loss": 1.006,
+      "step": 5338
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.24758673550756e-05,
+      "loss": 1.1765,
+      "step": 5339
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.243912220565941e-05,
+      "loss": 0.6929,
+      "step": 5340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.240238108299462e-05,
+      "loss": 1.2033,
+      "step": 5341
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.236564399244938e-05,
+      "loss": 0.975,
+      "step": 5342
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.2328910939391e-05,
+      "loss": 0.6829,
+      "step": 5343
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.229218192918639e-05,
+      "loss": 0.6168,
+      "step": 5344
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.225545696720177e-05,
+      "loss": 1.0806,
+      "step": 5345
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.221873605880284e-05,
+      "loss": 1.2581,
+      "step": 5346
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.218201920935461e-05,
+      "loss": 0.9173,
+      "step": 5347
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.214530642422162e-05,
+      "loss": 1.111,
+      "step": 5348
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.210859770876772e-05,
+      "loss": 1.2528,
+      "step": 5349
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.207189306835613e-05,
+      "loss": 1.3038,
+      "step": 5350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.203519250834967e-05,
+      "loss": 1.1463,
+      "step": 5351
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.199849603411035e-05,
+      "loss": 1.734,
+      "step": 5352
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.196180365099967e-05,
+      "loss": 0.9934,
+      "step": 5353
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.192511536437856e-05,
+      "loss": 0.7733,
+      "step": 5354
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.18884311796073e-05,
+      "loss": 1.2428,
+      "step": 5355
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.185175110204565e-05,
+      "loss": 1.1469,
+      "step": 5356
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.181507513705259e-05,
+      "loss": 1.4936,
+      "step": 5357
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.177840328998672e-05,
+      "loss": 0.7579,
+      "step": 5358
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.174173556620592e-05,
+      "loss": 0.8652,
+      "step": 5359
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.170507197106744e-05,
+      "loss": 1.5349,
+      "step": 5360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.166841250992806e-05,
+      "loss": 0.6016,
+      "step": 5361
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.163175718814383e-05,
+      "loss": 1.0356,
+      "step": 5362
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.159510601107016e-05,
+      "loss": 0.7936,
+      "step": 5363
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.155845898406206e-05,
+      "loss": 0.6874,
+      "step": 5364
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.152181611247372e-05,
+      "loss": 1.358,
+      "step": 5365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.148517740165883e-05,
+      "loss": 0.643,
+      "step": 5366
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.144854285697045e-05,
+      "loss": 0.7883,
+      "step": 5367
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.141191248376102e-05,
+      "loss": 1.4715,
+      "step": 5368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.137528628738236e-05,
+      "loss": 0.6523,
+      "step": 5369
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.133866427318575e-05,
+      "loss": 1.1601,
+      "step": 5370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.13020464465218e-05,
+      "loss": 0.8623,
+      "step": 5371
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.126543281274046e-05,
+      "loss": 0.8574,
+      "step": 5372
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.12288233771912e-05,
+      "loss": 0.5489,
+      "step": 5373
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.119221814522274e-05,
+      "loss": 0.6401,
+      "step": 5374
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.115561712218328e-05,
+      "loss": 1.7321,
+      "step": 5375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.111902031342037e-05,
+      "loss": 1.4016,
+      "step": 5376
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.108242772428093e-05,
+      "loss": 0.8937,
+      "step": 5377
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.104583936011129e-05,
+      "loss": 1.0162,
+      "step": 5378
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.100925522625716e-05,
+      "loss": 1.3421,
+      "step": 5379
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.097267532806364e-05,
+      "loss": 0.968,
+      "step": 5380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.093609967087514e-05,
+      "loss": 1.211,
+      "step": 5381
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.089952826003551e-05,
+      "loss": 1.3438,
+      "step": 5382
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.086296110088806e-05,
+      "loss": 1.0328,
+      "step": 5383
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.082639819877532e-05,
+      "loss": 1.2893,
+      "step": 5384
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.078983955903925e-05,
+      "loss": 1.1996,
+      "step": 5385
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.075328518702132e-05,
+      "loss": 0.7376,
+      "step": 5386
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.071673508806214e-05,
+      "loss": 1.0593,
+      "step": 5387
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.068018926750188e-05,
+      "loss": 0.8299,
+      "step": 5388
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.064364773068004e-05,
+      "loss": 1.3381,
+      "step": 5389
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.060711048293546e-05,
+      "loss": 0.9872,
+      "step": 5390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.057057752960636e-05,
+      "loss": 1.1805,
+      "step": 5391
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.053404887603037e-05,
+      "loss": 0.8655,
+      "step": 5392
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.04975245275445e-05,
+      "loss": 1.1416,
+      "step": 5393
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.0461004489485e-05,
+      "loss": 0.6414,
+      "step": 5394
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.042448876718771e-05,
+      "loss": 0.7824,
+      "step": 5395
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.038797736598764e-05,
+      "loss": 0.8964,
+      "step": 5396
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.035147029121925e-05,
+      "loss": 1.4204,
+      "step": 5397
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.031496754821638e-05,
+      "loss": 0.7161,
+      "step": 5398
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.027846914231226e-05,
+      "loss": 0.5898,
+      "step": 5399
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.024197507883939e-05,
+      "loss": 1.2518,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.020548536312972e-05,
+      "loss": 0.9389,
+      "step": 5401
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.016900000051459e-05,
+      "loss": 1.3232,
+      "step": 5402
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.013251899632454e-05,
+      "loss": 1.311,
+      "step": 5403
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.009604235588967e-05,
+      "loss": 1.5195,
+      "step": 5404
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.005957008453935e-05,
+      "loss": 1.1959,
+      "step": 5405
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.002310218760231e-05,
+      "loss": 1.4517,
+      "step": 5406
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.998663867040662e-05,
+      "loss": 1.2935,
+      "step": 5407
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.995017953827982e-05,
+      "loss": 1.2056,
+      "step": 5408
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.991372479654866e-05,
+      "loss": 0.7983,
+      "step": 5409
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.987727445053933e-05,
+      "loss": 0.9302,
+      "step": 5410
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.98408285055774e-05,
+      "loss": 1.4163,
+      "step": 5411
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.980438696698775e-05,
+      "loss": 0.9467,
+      "step": 5412
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.97679498400946e-05,
+      "loss": 1.3262,
+      "step": 5413
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.973151713022159e-05,
+      "loss": 0.5321,
+      "step": 5414
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.969508884269171e-05,
+      "loss": 1.1163,
+      "step": 5415
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.965866498282716e-05,
+      "loss": 1.3978,
+      "step": 5416
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.962224555594977e-05,
+      "loss": 0.9673,
+      "step": 5417
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.958583056738043e-05,
+      "loss": 1.1779,
+      "step": 5418
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.954942002243955e-05,
+      "loss": 1.1184,
+      "step": 5419
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.951301392644687e-05,
+      "loss": 1.3094,
+      "step": 5420
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.947661228472146e-05,
+      "loss": 1.3975,
+      "step": 5421
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.94402151025817e-05,
+      "loss": 0.5867,
+      "step": 5422
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.940382238534542e-05,
+      "loss": 0.5764,
+      "step": 5423
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.936743413832974e-05,
+      "loss": 0.9783,
+      "step": 5424
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.933105036685101e-05,
+      "loss": 1.1841,
+      "step": 5425
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.929467107622519e-05,
+      "loss": 0.6781,
+      "step": 5426
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.925829627176737e-05,
+      "loss": 1.196,
+      "step": 5427
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.9221925958792e-05,
+      "loss": 1.5074,
+      "step": 5428
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.918556014261302e-05,
+      "loss": 1.5431,
+      "step": 5429
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.914919882854357e-05,
+      "loss": 1.0998,
+      "step": 5430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.911284202189619e-05,
+      "loss": 1.2834,
+      "step": 5431
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.907648972798271e-05,
+      "loss": 1.6971,
+      "step": 5432
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.904014195211439e-05,
+      "loss": 1.2302,
+      "step": 5433
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.900379869960177e-05,
+      "loss": 1.1585,
+      "step": 5434
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.896745997575472e-05,
+      "loss": 0.9244,
+      "step": 5435
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.893112578588248e-05,
+      "loss": 0.5236,
+      "step": 5436
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.889479613529366e-05,
+      "loss": 1.4259,
+      "step": 5437
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.885847102929606e-05,
+      "loss": 0.5358,
+      "step": 5438
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.882215047319705e-05,
+      "loss": 1.1004,
+      "step": 5439
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.878583447230312e-05,
+      "loss": 1.3358,
+      "step": 5440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.874952303192017e-05,
+      "loss": 0.9121,
+      "step": 5441
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.87132161573535e-05,
+      "loss": 1.5679,
+      "step": 5442
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.867691385390764e-05,
+      "loss": 1.2202,
+      "step": 5443
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.864061612688651e-05,
+      "loss": 0.9147,
+      "step": 5444
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.860432298159337e-05,
+      "loss": 1.6195,
+      "step": 5445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.85680344233308e-05,
+      "loss": 0.9446,
+      "step": 5446
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.85317504574006e-05,
+      "loss": 0.9619,
+      "step": 5447
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.849547108910413e-05,
+      "loss": 1.3734,
+      "step": 5448
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.845919632374185e-05,
+      "loss": 1.2927,
+      "step": 5449
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.842292616661369e-05,
+      "loss": 1.098,
+      "step": 5450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.838666062301886e-05,
+      "loss": 0.9206,
+      "step": 5451
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.835039969825589e-05,
+      "loss": 1.4374,
+      "step": 5452
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.831414339762263e-05,
+      "loss": 0.5025,
+      "step": 5453
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.827789172641631e-05,
+      "loss": 0.7914,
+      "step": 5454
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.82416446899334e-05,
+      "loss": 1.1398,
+      "step": 5455
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.820540229346972e-05,
+      "loss": 1.0834,
+      "step": 5456
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.816916454232045e-05,
+      "loss": 1.0905,
+      "step": 5457
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.813293144178006e-05,
+      "loss": 1.39,
+      "step": 5458
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.80967029971424e-05,
+      "loss": 1.3146,
+      "step": 5459
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.806047921370044e-05,
+      "loss": 0.8982,
+      "step": 5460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.802426009674679e-05,
+      "loss": 0.9824,
+      "step": 5461
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.798804565157311e-05,
+      "loss": 0.5543,
+      "step": 5462
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.795183588347047e-05,
+      "loss": 0.9923,
+      "step": 5463
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.79156307977293e-05,
+      "loss": 0.873,
+      "step": 5464
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.787943039963926e-05,
+      "loss": 1.5892,
+      "step": 5465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.784323469448938e-05,
+      "loss": 1.0499,
+      "step": 5466
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.780704368756802e-05,
+      "loss": 0.5987,
+      "step": 5467
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.777085738416282e-05,
+      "loss": 1.5441,
+      "step": 5468
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.773467578956068e-05,
+      "loss": 1.3325,
+      "step": 5469
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.769849890904799e-05,
+      "loss": 1.7256,
+      "step": 5470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.766232674791023e-05,
+      "loss": 0.7214,
+      "step": 5471
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.76261593114323e-05,
+      "loss": 1.6011,
+      "step": 5472
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.758999660489847e-05,
+      "loss": 1.1602,
+      "step": 5473
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.75538386335922e-05,
+      "loss": 1.1672,
+      "step": 5474
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.751768540279628e-05,
+      "loss": 0.6795,
+      "step": 5475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.748153691779293e-05,
+      "loss": 1.4759,
+      "step": 5476
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.744539318386353e-05,
+      "loss": 1.1769,
+      "step": 5477
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.740925420628881e-05,
+      "loss": 1.4365,
+      "step": 5478
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.737311999034883e-05,
+      "loss": 1.1243,
+      "step": 5479
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.733699054132295e-05,
+      "loss": 1.032,
+      "step": 5480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.73008658644898e-05,
+      "loss": 1.3137,
+      "step": 5481
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.726474596512735e-05,
+      "loss": 0.5814,
+      "step": 5482
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.722863084851287e-05,
+      "loss": 1.0139,
+      "step": 5483
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.71925205199229e-05,
+      "loss": 0.5703,
+      "step": 5484
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.715641498463328e-05,
+      "loss": 0.8808,
+      "step": 5485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.712031424791921e-05,
+      "loss": 1.0768,
+      "step": 5486
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.708421831505513e-05,
+      "loss": 1.2683,
+      "step": 5487
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.704812719131478e-05,
+      "loss": 0.487,
+      "step": 5488
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.701204088197126e-05,
+      "loss": 1.312,
+      "step": 5489
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.69759593922969e-05,
+      "loss": 1.4496,
+      "step": 5490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.693988272756329e-05,
+      "loss": 0.9291,
+      "step": 5491
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.690381089304147e-05,
+      "loss": 1.4735,
+      "step": 5492
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.68677438940016e-05,
+      "loss": 0.8627,
+      "step": 5493
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.683168173571321e-05,
+      "loss": 1.038,
+      "step": 5494
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.679562442344517e-05,
+      "loss": 0.54,
+      "step": 5495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.675957196246556e-05,
+      "loss": 1.111,
+      "step": 5496
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.672352435804179e-05,
+      "loss": 1.3604,
+      "step": 5497
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.668748161544056e-05,
+      "loss": 1.4653,
+      "step": 5498
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.665144373992788e-05,
+      "loss": 0.7775,
+      "step": 5499
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.661541073676894e-05,
+      "loss": 1.6561,
+      "step": 5500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.657938261122841e-05,
+      "loss": 0.9311,
+      "step": 5501
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.654335936857007e-05,
+      "loss": 0.8891,
+      "step": 5502
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.650734101405706e-05,
+      "loss": 1.2523,
+      "step": 5503
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.647132755295182e-05,
+      "loss": 1.4194,
+      "step": 5504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.64353189905161e-05,
+      "loss": 0.6823,
+      "step": 5505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.639931533201079e-05,
+      "loss": 1.331,
+      "step": 5506
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.636331658269622e-05,
+      "loss": 1.5955,
+      "step": 5507
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.632732274783195e-05,
+      "loss": 1.6477,
+      "step": 5508
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.629133383267683e-05,
+      "loss": 1.0299,
+      "step": 5509
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.625534984248893e-05,
+      "loss": 1.047,
+      "step": 5510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.621937078252572e-05,
+      "loss": 1.2229,
+      "step": 5511
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.618339665804385e-05,
+      "loss": 1.2977,
+      "step": 5512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.614742747429922e-05,
+      "loss": 0.8544,
+      "step": 5513
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.611146323654718e-05,
+      "loss": 1.1683,
+      "step": 5514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.607550395004219e-05,
+      "loss": 1.4301,
+      "step": 5515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.6039549620038e-05,
+      "loss": 0.5734,
+      "step": 5516
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.600360025178774e-05,
+      "loss": 1.2617,
+      "step": 5517
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.596765585054371e-05,
+      "loss": 1.3846,
+      "step": 5518
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.593171642155757e-05,
+      "loss": 0.7331,
+      "step": 5519
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.589578197008018e-05,
+      "loss": 1.4397,
+      "step": 5520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.585985250136175e-05,
+      "loss": 1.4224,
+      "step": 5521
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.582392802065159e-05,
+      "loss": 1.3046,
+      "step": 5522
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.578800853319856e-05,
+      "loss": 1.4734,
+      "step": 5523
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.575209404425055e-05,
+      "loss": 1.7259,
+      "step": 5524
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.571618455905481e-05,
+      "loss": 1.2514,
+      "step": 5525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.568028008285789e-05,
+      "loss": 0.7872,
+      "step": 5526
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.564438062090557e-05,
+      "loss": 1.0383,
+      "step": 5527
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.560848617844284e-05,
+      "loss": 1.376,
+      "step": 5528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.557259676071412e-05,
+      "loss": 0.6952,
+      "step": 5529
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.553671237296291e-05,
+      "loss": 0.9208,
+      "step": 5530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.550083302043207e-05,
+      "loss": 1.2834,
+      "step": 5531
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.546495870836374e-05,
+      "loss": 0.9909,
+      "step": 5532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.542908944199928e-05,
+      "loss": 0.9126,
+      "step": 5533
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.539322522657936e-05,
+      "loss": 1.4806,
+      "step": 5534
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.53573660673438e-05,
+      "loss": 0.55,
+      "step": 5535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.532151196953187e-05,
+      "loss": 1.2508,
+      "step": 5536
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.528566293838193e-05,
+      "loss": 0.5949,
+      "step": 5537
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.524981897913165e-05,
+      "loss": 1.1411,
+      "step": 5538
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.521398009701803e-05,
+      "loss": 0.6472,
+      "step": 5539
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.517814629727721e-05,
+      "loss": 0.9336,
+      "step": 5540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.514231758514466e-05,
+      "loss": 1.1909,
+      "step": 5541
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.510649396585515e-05,
+      "loss": 1.0727,
+      "step": 5542
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.50706754446426e-05,
+      "loss": 1.1319,
+      "step": 5543
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.50348620267402e-05,
+      "loss": 0.9118,
+      "step": 5544
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.499905371738054e-05,
+      "loss": 1.0568,
+      "step": 5545
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.496325052179527e-05,
+      "loss": 0.8718,
+      "step": 5546
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.492745244521534e-05,
+      "loss": 1.4803,
+      "step": 5547
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.489165949287108e-05,
+      "loss": 1.0724,
+      "step": 5548
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.485587166999194e-05,
+      "loss": 1.3769,
+      "step": 5549
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.482008898180664e-05,
+      "loss": 1.1446,
+      "step": 5550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.478431143354322e-05,
+      "loss": 1.6262,
+      "step": 5551
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.47485390304289e-05,
+      "loss": 1.4963,
+      "step": 5552
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.471277177769012e-05,
+      "loss": 1.1099,
+      "step": 5553
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.467700968055268e-05,
+      "loss": 1.1273,
+      "step": 5554
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.46412527442415e-05,
+      "loss": 0.8701,
+      "step": 5555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.460550097398088e-05,
+      "loss": 0.9825,
+      "step": 5556
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.456975437499422e-05,
+      "loss": 0.6065,
+      "step": 5557
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.453401295250429e-05,
+      "loss": 1.1336,
+      "step": 5558
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.449827671173304e-05,
+      "loss": 1.0256,
+      "step": 5559
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.446254565790164e-05,
+      "loss": 0.8783,
+      "step": 5560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.442681979623057e-05,
+      "loss": 1.4069,
+      "step": 5561
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.439109913193951e-05,
+      "loss": 1.1326,
+      "step": 5562
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.435538367024738e-05,
+      "loss": 0.9174,
+      "step": 5563
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.431967341637235e-05,
+      "loss": 0.5229,
+      "step": 5564
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.428396837553188e-05,
+      "loss": 1.38,
+      "step": 5565
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.424826855294248e-05,
+      "loss": 1.2944,
+      "step": 5566
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.421257395382018e-05,
+      "loss": 1.0624,
+      "step": 5567
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.417688458338005e-05,
+      "loss": 0.5906,
+      "step": 5568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.414120044683638e-05,
+      "loss": 0.9236,
+      "step": 5569
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.410552154940286e-05,
+      "loss": 1.3709,
+      "step": 5570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.406984789629227e-05,
+      "loss": 1.7667,
+      "step": 5571
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.403417949271663e-05,
+      "loss": 1.4723,
+      "step": 5572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.399851634388732e-05,
+      "loss": 0.5909,
+      "step": 5573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.396285845501482e-05,
+      "loss": 1.1453,
+      "step": 5574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.392720583130885e-05,
+      "loss": 1.4401,
+      "step": 5575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.389155847797845e-05,
+      "loss": 0.9924,
+      "step": 5576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.385591640023183e-05,
+      "loss": 1.0633,
+      "step": 5577
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.382027960327641e-05,
+      "loss": 0.4118,
+      "step": 5578
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.378464809231888e-05,
+      "loss": 0.7123,
+      "step": 5579
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.374902187256518e-05,
+      "loss": 1.6625,
+      "step": 5580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.371340094922038e-05,
+      "loss": 0.669,
+      "step": 5581
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.367778532748882e-05,
+      "loss": 1.1413,
+      "step": 5582
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.364217501257415e-05,
+      "loss": 1.235,
+      "step": 5583
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.360657000967913e-05,
+      "loss": 1.2014,
+      "step": 5584
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.357097032400581e-05,
+      "loss": 1.4414,
+      "step": 5585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.353537596075543e-05,
+      "loss": 1.2303,
+      "step": 5586
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.34997869251285e-05,
+      "loss": 0.7139,
+      "step": 5587
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.346420322232463e-05,
+      "loss": 0.8368,
+      "step": 5588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.342862485754287e-05,
+      "loss": 1.0716,
+      "step": 5589
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.339305183598123e-05,
+      "loss": 1.0223,
+      "step": 5590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.335748416283711e-05,
+      "loss": 1.8871,
+      "step": 5591
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.332192184330712e-05,
+      "loss": 1.5345,
+      "step": 5592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.328636488258703e-05,
+      "loss": 1.2122,
+      "step": 5593
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.325081328587181e-05,
+      "loss": 0.9879,
+      "step": 5594
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.321526705835576e-05,
+      "loss": 0.9552,
+      "step": 5595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.317972620523231e-05,
+      "loss": 1.3514,
+      "step": 5596
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.314419073169402e-05,
+      "loss": 1.4777,
+      "step": 5597
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.31086606429329e-05,
+      "loss": 1.0601,
+      "step": 5598
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.307313594413995e-05,
+      "loss": 0.6161,
+      "step": 5599
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.303761664050548e-05,
+      "loss": 1.4171,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.300210273721898e-05,
+      "loss": 1.3938,
+      "step": 5601
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.296659423946923e-05,
+      "loss": 1.1902,
+      "step": 5602
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.293109115244408e-05,
+      "loss": 1.783,
+      "step": 5603
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.289559348133077e-05,
+      "loss": 1.2627,
+      "step": 5604
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.286010123131556e-05,
+      "loss": 1.2644,
+      "step": 5605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.282461440758405e-05,
+      "loss": 0.5965,
+      "step": 5606
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.278913301532095e-05,
+      "loss": 1.2804,
+      "step": 5607
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.27536570597103e-05,
+      "loss": 1.4493,
+      "step": 5608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.271818654593525e-05,
+      "loss": 1.0387,
+      "step": 5609
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.268272147917813e-05,
+      "loss": 1.1291,
+      "step": 5610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.264726186462062e-05,
+      "loss": 1.5061,
+      "step": 5611
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.261180770744341e-05,
+      "loss": 1.304,
+      "step": 5612
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.257635901282655e-05,
+      "loss": 1.1409,
+      "step": 5613
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.254091578594922e-05,
+      "loss": 1.1248,
+      "step": 5614
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.250547803198982e-05,
+      "loss": 1.0522,
+      "step": 5615
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.247004575612589e-05,
+      "loss": 1.71,
+      "step": 5616
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.243461896353432e-05,
+      "loss": 0.7534,
+      "step": 5617
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.239919765939105e-05,
+      "loss": 0.976,
+      "step": 5618
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.236378184887122e-05,
+      "loss": 1.4144,
+      "step": 5619
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.232837153714932e-05,
+      "loss": 1.3055,
+      "step": 5620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.229296672939886e-05,
+      "loss": 1.2885,
+      "step": 5621
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.225756743079263e-05,
+      "loss": 0.9901,
+      "step": 5622
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.222217364650264e-05,
+      "loss": 1.445,
+      "step": 5623
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.218678538170003e-05,
+      "loss": 1.1487,
+      "step": 5624
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.215140264155514e-05,
+      "loss": 1.2271,
+      "step": 5625
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.21160254312376e-05,
+      "loss": 1.3966,
+      "step": 5626
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.208065375591612e-05,
+      "loss": 1.4681,
+      "step": 5627
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.20452876207586e-05,
+      "loss": 1.5029,
+      "step": 5628
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.200992703093221e-05,
+      "loss": 1.3117,
+      "step": 5629
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.197457199160328e-05,
+      "loss": 1.6281,
+      "step": 5630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.193922250793729e-05,
+      "loss": 1.3281,
+      "step": 5631
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.190387858509894e-05,
+      "loss": 1.3877,
+      "step": 5632
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.186854022825215e-05,
+      "loss": 1.4653,
+      "step": 5633
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.183320744255994e-05,
+      "loss": 1.1949,
+      "step": 5634
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.179788023318459e-05,
+      "loss": 1.0851,
+      "step": 5635
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.176255860528756e-05,
+      "loss": 1.1038,
+      "step": 5636
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.172724256402946e-05,
+      "loss": 0.9713,
+      "step": 5637
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.169193211457009e-05,
+      "loss": 0.5728,
+      "step": 5638
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.165662726206847e-05,
+      "loss": 1.032,
+      "step": 5639
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.162132801168278e-05,
+      "loss": 0.8522,
+      "step": 5640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.158603436857032e-05,
+      "loss": 1.5616,
+      "step": 5641
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.155074633788771e-05,
+      "loss": 0.5321,
+      "step": 5642
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.151546392479062e-05,
+      "loss": 1.1702,
+      "step": 5643
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.148018713443393e-05,
+      "loss": 1.1368,
+      "step": 5644
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.144491597197176e-05,
+      "loss": 1.2245,
+      "step": 5645
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.140965044255735e-05,
+      "loss": 1.1324,
+      "step": 5646
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.137439055134311e-05,
+      "loss": 1.0884,
+      "step": 5647
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.13391363034807e-05,
+      "loss": 0.7176,
+      "step": 5648
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.130388770412086e-05,
+      "loss": 1.2647,
+      "step": 5649
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.126864475841353e-05,
+      "loss": 1.0235,
+      "step": 5650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.123340747150787e-05,
+      "loss": 1.7297,
+      "step": 5651
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.119817584855221e-05,
+      "loss": 1.2362,
+      "step": 5652
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.116294989469396e-05,
+      "loss": 1.1182,
+      "step": 5653
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.112772961507983e-05,
+      "loss": 1.206,
+      "step": 5654
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.109251501485565e-05,
+      "loss": 1.5859,
+      "step": 5655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.105730609916634e-05,
+      "loss": 1.2185,
+      "step": 5656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.1022102873156086e-05,
+      "loss": 1.1639,
+      "step": 5657
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0986905341968245e-05,
+      "loss": 1.359,
+      "step": 5658
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.095171351074529e-05,
+      "loss": 0.7578,
+      "step": 5659
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.091652738462887e-05,
+      "loss": 0.9526,
+      "step": 5660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0881346968759864e-05,
+      "loss": 1.3092,
+      "step": 5661
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0846172268278255e-05,
+      "loss": 0.4623,
+      "step": 5662
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.081100328832312e-05,
+      "loss": 1.1614,
+      "step": 5663
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0775840034032914e-05,
+      "loss": 1.0973,
+      "step": 5664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0740682510545053e-05,
+      "loss": 1.1691,
+      "step": 5665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0705530722996165e-05,
+      "loss": 1.2449,
+      "step": 5666
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.067038467652213e-05,
+      "loss": 0.4222,
+      "step": 5667
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.063524437625787e-05,
+      "loss": 1.1844,
+      "step": 5668
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.060010982733753e-05,
+      "loss": 1.4409,
+      "step": 5669
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0564981034894405e-05,
+      "loss": 0.9907,
+      "step": 5670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0529858004061004e-05,
+      "loss": 0.6262,
+      "step": 5671
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0494740739968814e-05,
+      "loss": 1.2514,
+      "step": 5672
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.045962924774874e-05,
+      "loss": 1.3488,
+      "step": 5673
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.042452353253063e-05,
+      "loss": 0.9922,
+      "step": 5674
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.038942359944355e-05,
+      "loss": 1.2818,
+      "step": 5675
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.03543294536158e-05,
+      "loss": 1.0723,
+      "step": 5676
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.031924110017474e-05,
+      "loss": 1.3031,
+      "step": 5677
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.028415854424689e-05,
+      "loss": 0.5531,
+      "step": 5678
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0249081790958004e-05,
+      "loss": 1.4056,
+      "step": 5679
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.02140108454329e-05,
+      "loss": 0.6342,
+      "step": 5680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.017894571279556e-05,
+      "loss": 1.2279,
+      "step": 5681
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.014388639816915e-05,
+      "loss": 1.433,
+      "step": 5682
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0108832906676e-05,
+      "loss": 0.9198,
+      "step": 5683
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.007378524343754e-05,
+      "loss": 1.241,
+      "step": 5684
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0038743413574336e-05,
+      "loss": 0.7908,
+      "step": 5685
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.000370742220621e-05,
+      "loss": 1.3746,
+      "step": 5686
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9968677274452015e-05,
+      "loss": 1.0913,
+      "step": 5687
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.993365297542975e-05,
+      "loss": 0.6013,
+      "step": 5688
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.989863453025668e-05,
+      "loss": 1.5632,
+      "step": 5689
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9863621944049084e-05,
+      "loss": 1.3824,
+      "step": 5690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9828615221922455e-05,
+      "loss": 1.7697,
+      "step": 5691
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9793614368991426e-05,
+      "loss": 1.1362,
+      "step": 5692
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.975861939036975e-05,
+      "loss": 1.4172,
+      "step": 5693
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.972363029117027e-05,
+      "loss": 0.8375,
+      "step": 5694
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9688647076505146e-05,
+      "loss": 1.3049,
+      "step": 5695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.965366975148545e-05,
+      "loss": 1.7124,
+      "step": 5696
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.961869832122156e-05,
+      "loss": 1.0675,
+      "step": 5697
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9583732790822944e-05,
+      "loss": 0.5048,
+      "step": 5698
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.954877316539817e-05,
+      "loss": 0.7159,
+      "step": 5699
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.951381945005498e-05,
+      "loss": 0.7869,
+      "step": 5700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.947887164990029e-05,
+      "loss": 1.1994,
+      "step": 5701
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.94439297700401e-05,
+      "loss": 1.3581,
+      "step": 5702
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.94089938155795e-05,
+      "loss": 1.315,
+      "step": 5703
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.93740637916228e-05,
+      "loss": 1.3014,
+      "step": 5704
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.933913970327344e-05,
+      "loss": 1.2687,
+      "step": 5705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.930422155563391e-05,
+      "loss": 1.6996,
+      "step": 5706
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.926930935380594e-05,
+      "loss": 1.4169,
+      "step": 5707
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9234403102890344e-05,
+      "loss": 0.7968,
+      "step": 5708
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9199502807987005e-05,
+      "loss": 1.0133,
+      "step": 5709
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9164608474195005e-05,
+      "loss": 0.7518,
+      "step": 5710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.912972010661257e-05,
+      "loss": 0.9724,
+      "step": 5711
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.909483771033701e-05,
+      "loss": 1.3648,
+      "step": 5712
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.905996129046478e-05,
+      "loss": 0.9714,
+      "step": 5713
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9025090852091466e-05,
+      "loss": 0.9675,
+      "step": 5714
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.899022640031181e-05,
+      "loss": 1.3007,
+      "step": 5715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.895536794021954e-05,
+      "loss": 1.1159,
+      "step": 5716
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.892051547690773e-05,
+      "loss": 0.4915,
+      "step": 5717
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.888566901546839e-05,
+      "loss": 0.8367,
+      "step": 5718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.885082856099273e-05,
+      "loss": 1.8715,
+      "step": 5719
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.8815994118571105e-05,
+      "loss": 1.0047,
+      "step": 5720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.878116569329295e-05,
+      "loss": 1.2243,
+      "step": 5721
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.87463432902468e-05,
+      "loss": 0.6146,
+      "step": 5722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.87115269145204e-05,
+      "loss": 1.0638,
+      "step": 5723
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.8676716571200555e-05,
+      "loss": 1.0418,
+      "step": 5724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.864191226537312e-05,
+      "loss": 0.7044,
+      "step": 5725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.860711400212321e-05,
+      "loss": 1.1299,
+      "step": 5726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.857232178653495e-05,
+      "loss": 1.6345,
+      "step": 5727
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.853753562369164e-05,
+      "loss": 1.3428,
+      "step": 5728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.850275551867567e-05,
+      "loss": 1.0631,
+      "step": 5729
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.846798147656856e-05,
+      "loss": 0.5836,
+      "step": 5730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.8433213502450854e-05,
+      "loss": 1.2637,
+      "step": 5731
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.8398451601402425e-05,
+      "loss": 1.2813,
+      "step": 5732
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.836369577850202e-05,
+      "loss": 1.4358,
+      "step": 5733
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.832894603882766e-05,
+      "loss": 1.4166,
+      "step": 5734
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.829420238745629e-05,
+      "loss": 1.3309,
+      "step": 5735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.825946482946427e-05,
+      "loss": 1.2045,
+      "step": 5736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.822473336992682e-05,
+      "loss": 1.2859,
+      "step": 5737
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.8190008013918275e-05,
+      "loss": 1.2134,
+      "step": 5738
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.8155288766512216e-05,
+      "loss": 1.1252,
+      "step": 5739
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.812057563278128e-05,
+      "loss": 1.1549,
+      "step": 5740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.808586861779711e-05,
+      "loss": 1.749,
+      "step": 5741
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.805116772663063e-05,
+      "loss": 1.0593,
+      "step": 5742
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.8016472964351685e-05,
+      "loss": 0.8905,
+      "step": 5743
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.798178433602935e-05,
+      "loss": 1.4089,
+      "step": 5744
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.794710184673182e-05,
+      "loss": 0.5333,
+      "step": 5745
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.791242550152627e-05,
+      "loss": 1.2639,
+      "step": 5746
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.7877755305479065e-05,
+      "loss": 1.4888,
+      "step": 5747
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.78430912636557e-05,
+      "loss": 1.1533,
+      "step": 5748
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.7808433381120696e-05,
+      "loss": 1.005,
+      "step": 5749
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.7773781662937635e-05,
+      "loss": 1.2363,
+      "step": 5750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.77391361141694e-05,
+      "loss": 0.9538,
+      "step": 5751
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.7704496739877776e-05,
+      "loss": 1.6771,
+      "step": 5752
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.766986354512367e-05,
+      "loss": 0.8896,
+      "step": 5753
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.763523653496716e-05,
+      "loss": 1.0965,
+      "step": 5754
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.7600615714467446e-05,
+      "loss": 1.3191,
+      "step": 5755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.756600108868266e-05,
+      "loss": 1.2466,
+      "step": 5756
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.753139266267019e-05,
+      "loss": 1.2833,
+      "step": 5757
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.74967904414865e-05,
+      "loss": 1.5843,
+      "step": 5758
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.746219443018706e-05,
+      "loss": 1.3655,
+      "step": 5759
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.7427604633826436e-05,
+      "loss": 1.3592,
+      "step": 5760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.7393021057458394e-05,
+      "loss": 0.8624,
+      "step": 5761
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.735844370613576e-05,
+      "loss": 0.7682,
+      "step": 5762
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.732387258491033e-05,
+      "loss": 0.7929,
+      "step": 5763
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.728930769883314e-05,
+      "loss": 0.9129,
+      "step": 5764
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.725474905295426e-05,
+      "loss": 0.6025,
+      "step": 5765
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.722019665232279e-05,
+      "loss": 0.6436,
+      "step": 5766
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.7185650501987056e-05,
+      "loss": 0.9333,
+      "step": 5767
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.7151110606994276e-05,
+      "loss": 1.5159,
+      "step": 5768
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.711657697239093e-05,
+      "loss": 1.6,
+      "step": 5769
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.7082049603222544e-05,
+      "loss": 0.995,
+      "step": 5770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.7047528504533666e-05,
+      "loss": 1.2068,
+      "step": 5771
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.701301368136788e-05,
+      "loss": 0.8114,
+      "step": 5772
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6978505138768077e-05,
+      "loss": 1.1037,
+      "step": 5773
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.694400288177606e-05,
+      "loss": 1.2747,
+      "step": 5774
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6909506915432644e-05,
+      "loss": 1.2348,
+      "step": 5775
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6875017244777905e-05,
+      "loss": 0.9155,
+      "step": 5776
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6840533874850935e-05,
+      "loss": 1.266,
+      "step": 5777
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6806056810689824e-05,
+      "loss": 1.3461,
+      "step": 5778
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.677158605733184e-05,
+      "loss": 0.3762,
+      "step": 5779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.673712161981333e-05,
+      "loss": 1.2873,
+      "step": 5780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6702663503169596e-05,
+      "loss": 1.0591,
+      "step": 5781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6668211712435194e-05,
+      "loss": 0.6877,
+      "step": 5782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6633766252643596e-05,
+      "loss": 1.1003,
+      "step": 5783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.659932712882747e-05,
+      "loss": 0.5942,
+      "step": 5784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.656489434601845e-05,
+      "loss": 1.63,
+      "step": 5785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.653046790924732e-05,
+      "loss": 1.0914,
+      "step": 5786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.649604782354398e-05,
+      "loss": 0.8659,
+      "step": 5787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.646163409393723e-05,
+      "loss": 0.8244,
+      "step": 5788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.642722672545515e-05,
+      "loss": 0.9561,
+      "step": 5789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.639282572312472e-05,
+      "loss": 1.2497,
+      "step": 5790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.635843109197206e-05,
+      "loss": 1.1918,
+      "step": 5791
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.632404283702244e-05,
+      "loss": 0.6497,
+      "step": 5792
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.628966096330006e-05,
+      "loss": 1.5049,
+      "step": 5793
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.625528547582816e-05,
+      "loss": 1.3995,
+      "step": 5794
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.62209163796293e-05,
+      "loss": 1.2065,
+      "step": 5795
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.618655367972487e-05,
+      "loss": 1.3399,
+      "step": 5796
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.615219738113533e-05,
+      "loss": 0.6952,
+      "step": 5797
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.611784748888033e-05,
+      "loss": 1.2399,
+      "step": 5798
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6083504007978546e-05,
+      "loss": 1.2963,
+      "step": 5799
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.604916694344764e-05,
+      "loss": 1.1209,
+      "step": 5800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6014836300304394e-05,
+      "loss": 1.2913,
+      "step": 5801
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5980512083564716e-05,
+      "loss": 0.6459,
+      "step": 5802
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.594619429824342e-05,
+      "loss": 0.8258,
+      "step": 5803
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.591188294935454e-05,
+      "loss": 1.4564,
+      "step": 5804
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5877578041911036e-05,
+      "loss": 1.2287,
+      "step": 5805
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5843279580925015e-05,
+      "loss": 1.3316,
+      "step": 5806
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.580898757140766e-05,
+      "loss": 1.7638,
+      "step": 5807
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5774702018369074e-05,
+      "loss": 1.3954,
+      "step": 5808
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.574042292681861e-05,
+      "loss": 1.2269,
+      "step": 5809
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5706150301764495e-05,
+      "loss": 1.5207,
+      "step": 5810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.567188414821415e-05,
+      "loss": 1.14,
+      "step": 5811
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.563762447117393e-05,
+      "loss": 1.2928,
+      "step": 5812
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5603371275649354e-05,
+      "loss": 0.5838,
+      "step": 5813
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.556912456664497e-05,
+      "loss": 1.5451,
+      "step": 5814
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.553488434916429e-05,
+      "loss": 1.3848,
+      "step": 5815
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.550065062820997e-05,
+      "loss": 0.8212,
+      "step": 5816
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.546642340878374e-05,
+      "loss": 1.0243,
+      "step": 5817
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5432202695886294e-05,
+      "loss": 1.2693,
+      "step": 5818
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.539798849451736e-05,
+      "loss": 1.6164,
+      "step": 5819
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.536378080967581e-05,
+      "loss": 1.7605,
+      "step": 5820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5329579646359555e-05,
+      "loss": 1.1384,
+      "step": 5821
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.529538500956544e-05,
+      "loss": 1.327,
+      "step": 5822
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.526119690428948e-05,
+      "loss": 0.6872,
+      "step": 5823
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5227015335526745e-05,
+      "loss": 1.1777,
+      "step": 5824
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.519284030827118e-05,
+      "loss": 1.4484,
+      "step": 5825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5158671827515985e-05,
+      "loss": 1.1148,
+      "step": 5826
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.512450989825324e-05,
+      "loss": 1.3092,
+      "step": 5827
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.509035452547415e-05,
+      "loss": 1.37,
+      "step": 5828
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5056205714169016e-05,
+      "loss": 1.1541,
+      "step": 5829
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.502206346932701e-05,
+      "loss": 1.3956,
+      "step": 5830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.4987927795936514e-05,
+      "loss": 0.5506,
+      "step": 5831
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.4953798698984893e-05,
+      "loss": 0.5698,
+      "step": 5832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.4919676183458524e-05,
+      "loss": 1.7181,
+      "step": 5833
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.488556025434277e-05,
+      "loss": 0.3817,
+      "step": 5834
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.485145091662217e-05,
+      "loss": 1.1982,
+      "step": 5835
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.481734817528026e-05,
+      "loss": 0.3842,
+      "step": 5836
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.4783252035299505e-05,
+      "loss": 0.6752,
+      "step": 5837
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.474916250166151e-05,
+      "loss": 1.4277,
+      "step": 5838
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.471507957934695e-05,
+      "loss": 1.0595,
+      "step": 5839
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.468100327333543e-05,
+      "loss": 1.2367,
+      "step": 5840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.464693358860558e-05,
+      "loss": 0.9468,
+      "step": 5841
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.461287053013514e-05,
+      "loss": 1.3888,
+      "step": 5842
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.4578814102900935e-05,
+      "loss": 1.3614,
+      "step": 5843
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.4544764311878624e-05,
+      "loss": 0.8621,
+      "step": 5844
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.451072116204308e-05,
+      "loss": 1.1948,
+      "step": 5845
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.447668465836815e-05,
+      "loss": 1.3194,
+      "step": 5846
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.444265480582665e-05,
+      "loss": 0.5118,
+      "step": 5847
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.440863160939055e-05,
+      "loss": 1.1705,
+      "step": 5848
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.437461507403068e-05,
+      "loss": 1.7836,
+      "step": 5849
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.434060520471702e-05,
+      "loss": 0.8274,
+      "step": 5850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.430660200641862e-05,
+      "loss": 1.4769,
+      "step": 5851
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.427260548410336e-05,
+      "loss": 1.6465,
+      "step": 5852
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.423861564273831e-05,
+      "loss": 1.5938,
+      "step": 5853
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.420463248728958e-05,
+      "loss": 1.2364,
+      "step": 5854
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.417065602272218e-05,
+      "loss": 1.3565,
+      "step": 5855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.413668625400014e-05,
+      "loss": 1.1818,
+      "step": 5856
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.4102723186086724e-05,
+      "loss": 0.5035,
+      "step": 5857
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.4068766823943995e-05,
+      "loss": 1.0547,
+      "step": 5858
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.403481717253307e-05,
+      "loss": 1.5177,
+      "step": 5859
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.4000874236814144e-05,
+      "loss": 0.6168,
+      "step": 5860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3966938021746484e-05,
+      "loss": 1.4456,
+      "step": 5861
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.393300853228825e-05,
+      "loss": 1.2126,
+      "step": 5862
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.389908577339663e-05,
+      "loss": 1.2432,
+      "step": 5863
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.386516975002791e-05,
+      "loss": 0.6057,
+      "step": 5864
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.38312604671374e-05,
+      "loss": 0.7921,
+      "step": 5865
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3797357929679283e-05,
+      "loss": 0.8618,
+      "step": 5866
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.376346214260691e-05,
+      "loss": 1.376,
+      "step": 5867
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.372957311087262e-05,
+      "loss": 1.2276,
+      "step": 5868
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.369569083942765e-05,
+      "loss": 0.6508,
+      "step": 5869
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.366181533322241e-05,
+      "loss": 0.8899,
+      "step": 5870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.362794659720617e-05,
+      "loss": 1.3324,
+      "step": 5871
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3594084636327313e-05,
+      "loss": 1.0815,
+      "step": 5872
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.356022945553327e-05,
+      "loss": 1.4912,
+      "step": 5873
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.35263810597703e-05,
+      "loss": 1.4711,
+      "step": 5874
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3492539453983845e-05,
+      "loss": 1.4456,
+      "step": 5875
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.345870464311833e-05,
+      "loss": 1.0189,
+      "step": 5876
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3424876632117106e-05,
+      "loss": 1.4416,
+      "step": 5877
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.339105542592251e-05,
+      "loss": 0.8939,
+      "step": 5878
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.335724102947611e-05,
+      "loss": 1.6146,
+      "step": 5879
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.332343344771824e-05,
+      "loss": 1.2032,
+      "step": 5880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3289632685588264e-05,
+      "loss": 1.3509,
+      "step": 5881
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3255838748024665e-05,
+      "loss": 0.8752,
+      "step": 5882
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.32220516399649e-05,
+      "loss": 1.3084,
+      "step": 5883
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.318827136634536e-05,
+      "loss": 1.4282,
+      "step": 5884
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.315449793210139e-05,
+      "loss": 1.4371,
+      "step": 5885
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.3120731342167584e-05,
+      "loss": 1.538,
+      "step": 5886
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.3086971601477285e-05,
+      "loss": 1.313,
+      "step": 5887
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.305321871496289e-05,
+      "loss": 1.4289,
+      "step": 5888
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.3019472687555874e-05,
+      "loss": 1.2989,
+      "step": 5889
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.2985733524186685e-05,
+      "loss": 1.2465,
+      "step": 5890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.295200122978468e-05,
+      "loss": 0.6064,
+      "step": 5891
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.291827580927835e-05,
+      "loss": 1.2008,
+      "step": 5892
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.288455726759502e-05,
+      "loss": 0.5456,
+      "step": 5893
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.285084560966118e-05,
+      "loss": 0.8058,
+      "step": 5894
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.281714084040223e-05,
+      "loss": 1.3782,
+      "step": 5895
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.2783442964742516e-05,
+      "loss": 1.066,
+      "step": 5896
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.2749751987605445e-05,
+      "loss": 1.1791,
+      "step": 5897
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.271606791391346e-05,
+      "loss": 1.1949,
+      "step": 5898
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.268239074858789e-05,
+      "loss": 1.4813,
+      "step": 5899
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.264872049654901e-05,
+      "loss": 0.8086,
+      "step": 5900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.261505716271635e-05,
+      "loss": 0.6158,
+      "step": 5901
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.258140075200817e-05,
+      "loss": 0.954,
+      "step": 5902
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.2547751269341764e-05,
+      "loss": 1.2984,
+      "step": 5903
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.251410871963348e-05,
+      "loss": 1.1757,
+      "step": 5904
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.2480473107798664e-05,
+      "loss": 1.6839,
+      "step": 5905
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.244684443875156e-05,
+      "loss": 0.9747,
+      "step": 5906
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.241322271740545e-05,
+      "loss": 1.2244,
+      "step": 5907
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.2379607948672674e-05,
+      "loss": 1.2051,
+      "step": 5908
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.2346000137464404e-05,
+      "loss": 1.284,
+      "step": 5909
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.231239928869085e-05,
+      "loss": 1.3245,
+      "step": 5910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.2278805407261266e-05,
+      "loss": 0.8986,
+      "step": 5911
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.224521849808389e-05,
+      "loss": 1.4546,
+      "step": 5912
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.221163856606582e-05,
+      "loss": 1.4343,
+      "step": 5913
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.217806561611329e-05,
+      "loss": 1.0913,
+      "step": 5914
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.214449965313137e-05,
+      "loss": 0.6384,
+      "step": 5915
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.211094068202421e-05,
+      "loss": 1.5094,
+      "step": 5916
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.207738870769494e-05,
+      "loss": 1.2021,
+      "step": 5917
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.204384373504556e-05,
+      "loss": 0.7574,
+      "step": 5918
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.2010305768977167e-05,
+      "loss": 0.6428,
+      "step": 5919
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.197677481438982e-05,
+      "loss": 1.0226,
+      "step": 5920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.194325087618248e-05,
+      "loss": 0.6226,
+      "step": 5921
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.190973395925305e-05,
+      "loss": 1.4338,
+      "step": 5922
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1876224068498635e-05,
+      "loss": 0.7334,
+      "step": 5923
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1842721208815104e-05,
+      "loss": 1.3472,
+      "step": 5924
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.180922538509728e-05,
+      "loss": 1.2348,
+      "step": 5925
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.17757366022391e-05,
+      "loss": 1.3275,
+      "step": 5926
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.174225486513343e-05,
+      "loss": 0.9396,
+      "step": 5927
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1708780178672026e-05,
+      "loss": 1.0263,
+      "step": 5928
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.167531254774567e-05,
+      "loss": 1.4545,
+      "step": 5929
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.164185197724418e-05,
+      "loss": 1.5148,
+      "step": 5930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1608398472056184e-05,
+      "loss": 1.3828,
+      "step": 5931
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.157495203706947e-05,
+      "loss": 1.2562,
+      "step": 5932
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.15415126771706e-05,
+      "loss": 1.513,
+      "step": 5933
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.150808039724526e-05,
+      "loss": 1.318,
+      "step": 5934
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1474655202177976e-05,
+      "loss": 1.2453,
+      "step": 5935
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1441237096852335e-05,
+      "loss": 1.5299,
+      "step": 5936
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1407826086150864e-05,
+      "loss": 1.5137,
+      "step": 5937
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1374422174955015e-05,
+      "loss": 1.2211,
+      "step": 5938
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.134102536814527e-05,
+      "loss": 1.38,
+      "step": 5939
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.130763567060097e-05,
+      "loss": 0.77,
+      "step": 5940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1274253087200507e-05,
+      "loss": 1.1222,
+      "step": 5941
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.124087762282126e-05,
+      "loss": 0.5989,
+      "step": 5942
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.120750928233946e-05,
+      "loss": 0.545,
+      "step": 5943
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1174148070630276e-05,
+      "loss": 1.1263,
+      "step": 5944
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.114079399256808e-05,
+      "loss": 0.5444,
+      "step": 5945
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.110744705302592e-05,
+      "loss": 1.0109,
+      "step": 5946
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.107410725687593e-05,
+      "loss": 1.1459,
+      "step": 5947
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.104077460898918e-05,
+      "loss": 1.0783,
+      "step": 5948
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.100744911423576e-05,
+      "loss": 1.0285,
+      "step": 5949
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.097413077748456e-05,
+      "loss": 1.2615,
+      "step": 5950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.094081960360356e-05,
+      "loss": 0.6139,
+      "step": 5951
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.090751559745971e-05,
+      "loss": 1.0772,
+      "step": 5952
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.087421876391876e-05,
+      "loss": 1.3671,
+      "step": 5953
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.084092910784557e-05,
+      "loss": 1.7261,
+      "step": 5954
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0807646634103844e-05,
+      "loss": 0.8968,
+      "step": 5955
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0774371347556314e-05,
+      "loss": 0.8303,
+      "step": 5956
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.074110325306465e-05,
+      "loss": 0.9755,
+      "step": 5957
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0707842355489364e-05,
+      "loss": 1.0314,
+      "step": 5958
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0674588659690115e-05,
+      "loss": 1.2125,
+      "step": 5959
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.06413421705253e-05,
+      "loss": 0.9711,
+      "step": 5960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.060810289285243e-05,
+      "loss": 0.3216,
+      "step": 5961
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0574870831527833e-05,
+      "loss": 1.0337,
+      "step": 5962
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.054164599140686e-05,
+      "loss": 1.2799,
+      "step": 5963
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.050842837734385e-05,
+      "loss": 1.1889,
+      "step": 5964
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.047521799419194e-05,
+      "loss": 1.0462,
+      "step": 5965
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0442014846803334e-05,
+      "loss": 0.9786,
+      "step": 5966
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0408818940029176e-05,
+      "loss": 1.1301,
+      "step": 5967
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.037563027871948e-05,
+      "loss": 0.5898,
+      "step": 5968
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.03424488677232e-05,
+      "loss": 1.064,
+      "step": 5969
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.030927471188831e-05,
+      "loss": 1.0188,
+      "step": 5970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0276107816061733e-05,
+      "loss": 0.9561,
+      "step": 5971
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.024294818508919e-05,
+      "loss": 1.453,
+      "step": 5972
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0209795823815464e-05,
+      "loss": 0.8416,
+      "step": 5973
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.017665073708428e-05,
+      "loss": 1.262,
+      "step": 5974
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0143512929738225e-05,
+      "loss": 1.204,
+      "step": 5975
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.01103824066189e-05,
+      "loss": 1.2411,
+      "step": 5976
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.007725917256675e-05,
+      "loss": 1.4429,
+      "step": 5977
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.004414323242123e-05,
+      "loss": 1.2594,
+      "step": 5978
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.001103459102074e-05,
+      "loss": 1.2325,
+      "step": 5979
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.997793325320253e-05,
+      "loss": 1.6082,
+      "step": 5980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.994483922380285e-05,
+      "loss": 1.3493,
+      "step": 5981
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.991175250765692e-05,
+      "loss": 1.125,
+      "step": 5982
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9878673109598786e-05,
+      "loss": 0.5485,
+      "step": 5983
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.984560103446145e-05,
+      "loss": 0.7308,
+      "step": 5984
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9812536287076904e-05,
+      "loss": 0.8484,
+      "step": 5985
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9779478872276074e-05,
+      "loss": 1.226,
+      "step": 5986
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.97464287948887e-05,
+      "loss": 0.9928,
+      "step": 5987
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.971338605974357e-05,
+      "loss": 1.2379,
+      "step": 5988
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.96803506716684e-05,
+      "loss": 1.2912,
+      "step": 5989
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.964732263548972e-05,
+      "loss": 0.7288,
+      "step": 5990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.961430195603306e-05,
+      "loss": 0.9495,
+      "step": 5991
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.958128863812288e-05,
+      "loss": 0.8255,
+      "step": 5992
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.954828268658259e-05,
+      "loss": 0.6742,
+      "step": 5993
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9515284106234425e-05,
+      "loss": 1.3356,
+      "step": 5994
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.948229290189964e-05,
+      "loss": 1.1502,
+      "step": 5995
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.944930907839841e-05,
+      "loss": 1.1388,
+      "step": 5996
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.941633264054973e-05,
+      "loss": 1.2989,
+      "step": 5997
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9383363593171675e-05,
+      "loss": 1.4247,
+      "step": 5998
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.935040194108104e-05,
+      "loss": 0.5956,
+      "step": 5999
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.931744768909372e-05,
+      "loss": 1.0994,
+      "step": 6000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.928450084202448e-05,
+      "loss": 1.0104,
+      "step": 6001
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.925156140468693e-05,
+      "loss": 1.4484,
+      "step": 6002
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.921862938189366e-05,
+      "loss": 0.438,
+      "step": 6003
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.918570477845622e-05,
+      "loss": 1.0828,
+      "step": 6004
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.915278759918498e-05,
+      "loss": 1.1672,
+      "step": 6005
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9119877848889184e-05,
+      "loss": 0.5956,
+      "step": 6006
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.908697553237725e-05,
+      "loss": 1.1901,
+      "step": 6007
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9054080654456226e-05,
+      "loss": 0.9272,
+      "step": 6008
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.902119321993217e-05,
+      "loss": 1.3127,
+      "step": 6009
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8988313233610097e-05,
+      "loss": 1.1769,
+      "step": 6010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8955440700293934e-05,
+      "loss": 1.3038,
+      "step": 6011
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.892257562478645e-05,
+      "loss": 0.9815,
+      "step": 6012
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.888971801188932e-05,
+      "loss": 1.489,
+      "step": 6013
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.885686786640321e-05,
+      "loss": 1.1175,
+      "step": 6014
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.882402519312769e-05,
+      "loss": 1.0136,
+      "step": 6015
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.879118999686112e-05,
+      "loss": 1.452,
+      "step": 6016
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8758362282400907e-05,
+      "loss": 1.3995,
+      "step": 6017
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8725542054543326e-05,
+      "loss": 1.4272,
+      "step": 6018
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.869272931808347e-05,
+      "loss": 0.4898,
+      "step": 6019
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.865992407781549e-05,
+      "loss": 1.4238,
+      "step": 6020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.862712633853228e-05,
+      "loss": 1.2326,
+      "step": 6021
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.859433610502575e-05,
+      "loss": 1.1382,
+      "step": 6022
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.856155338208672e-05,
+      "loss": 0.4323,
+      "step": 6023
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.852877817450481e-05,
+      "loss": 0.8439,
+      "step": 6024
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.849601048706862e-05,
+      "loss": 1.1551,
+      "step": 6025
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.84632503245657e-05,
+      "loss": 1.2876,
+      "step": 6026
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8430497691782374e-05,
+      "loss": 1.4876,
+      "step": 6027
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.839775259350388e-05,
+      "loss": 1.2619,
+      "step": 6028
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.836501503451453e-05,
+      "loss": 0.7616,
+      "step": 6029
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8332285019597344e-05,
+      "loss": 1.2665,
+      "step": 6030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8299562553534286e-05,
+      "loss": 1.1311,
+      "step": 6031
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.826684764110624e-05,
+      "loss": 1.3175,
+      "step": 6032
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.823414028709303e-05,
+      "loss": 1.036,
+      "step": 6033
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.820144049627331e-05,
+      "loss": 1.427,
+      "step": 6034
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8168748273424536e-05,
+      "loss": 1.0552,
+      "step": 6035
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.813606362332336e-05,
+      "loss": 1.3342,
+      "step": 6036
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.810338655074502e-05,
+      "loss": 1.3935,
+      "step": 6037
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.807071706046377e-05,
+      "loss": 0.786,
+      "step": 6038
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.803805515725275e-05,
+      "loss": 1.5686,
+      "step": 6039
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.800540084588404e-05,
+      "loss": 0.8458,
+      "step": 6040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7972754131128496e-05,
+      "loss": 0.9035,
+      "step": 6041
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.794011501775599e-05,
+      "loss": 0.5243,
+      "step": 6042
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7907483510535165e-05,
+      "loss": 0.7676,
+      "step": 6043
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.787485961423364e-05,
+      "loss": 0.8508,
+      "step": 6044
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.784224333361792e-05,
+      "loss": 0.7905,
+      "step": 6045
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.780963467345332e-05,
+      "loss": 0.4014,
+      "step": 6046
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.777703363850411e-05,
+      "loss": 1.7289,
+      "step": 6047
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7744440233533474e-05,
+      "loss": 0.9551,
+      "step": 6048
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.771185446330339e-05,
+      "loss": 1.2306,
+      "step": 6049
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.76792763325747e-05,
+      "loss": 1.2036,
+      "step": 6050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.764670584610734e-05,
+      "loss": 0.9166,
+      "step": 6051
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7614143008659915e-05,
+      "loss": 1.2978,
+      "step": 6052
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7581587824989934e-05,
+      "loss": 0.7285,
+      "step": 6053
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.754904029985389e-05,
+      "loss": 1.439,
+      "step": 6054
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7516500438007125e-05,
+      "loss": 1.5364,
+      "step": 6055
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.748396824420377e-05,
+      "loss": 1.9458,
+      "step": 6056
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7451443723196956e-05,
+      "loss": 1.069,
+      "step": 6057
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.741892687973864e-05,
+      "loss": 1.615,
+      "step": 6058
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.738641771857963e-05,
+      "loss": 1.5333,
+      "step": 6059
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.73539162444697e-05,
+      "loss": 0.5268,
+      "step": 6060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.732142246215736e-05,
+      "loss": 0.8865,
+      "step": 6061
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.728893637639016e-05,
+      "loss": 1.5637,
+      "step": 6062
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.725645799191436e-05,
+      "loss": 0.7946,
+      "step": 6063
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.722398731347525e-05,
+      "loss": 1.2705,
+      "step": 6064
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.719152434581687e-05,
+      "loss": 1.2424,
+      "step": 6065
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7159069093682206e-05,
+      "loss": 1.2654,
+      "step": 6066
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7126621561813137e-05,
+      "loss": 1.1263,
+      "step": 6067
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7094181754950297e-05,
+      "loss": 1.1924,
+      "step": 6068
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.70617496778333e-05,
+      "loss": 1.4098,
+      "step": 6069
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.702932533520066e-05,
+      "loss": 1.1443,
+      "step": 6070
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6996908731789635e-05,
+      "loss": 0.6607,
+      "step": 6071
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6964499872336354e-05,
+      "loss": 1.6985,
+      "step": 6072
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.693209876157603e-05,
+      "loss": 0.7141,
+      "step": 6073
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.689970540424251e-05,
+      "loss": 1.382,
+      "step": 6074
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.686731980506855e-05,
+      "loss": 1.0366,
+      "step": 6075
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.683494196878586e-05,
+      "loss": 1.1661,
+      "step": 6076
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6802571900125005e-05,
+      "loss": 1.0964,
+      "step": 6077
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.677020960381528e-05,
+      "loss": 1.033,
+      "step": 6078
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.673785508458502e-05,
+      "loss": 1.5296,
+      "step": 6079
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.670550834716134e-05,
+      "loss": 1.0642,
+      "step": 6080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6673169396270176e-05,
+      "loss": 1.3055,
+      "step": 6081
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.664083823663644e-05,
+      "loss": 1.4055,
+      "step": 6082
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.660851487298378e-05,
+      "loss": 1.3415,
+      "step": 6083
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.657619931003479e-05,
+      "loss": 1.2767,
+      "step": 6084
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6543891552510934e-05,
+      "loss": 0.7047,
+      "step": 6085
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.651159160513242e-05,
+      "loss": 0.7062,
+      "step": 6086
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.64792994726185e-05,
+      "loss": 1.1717,
+      "step": 6087
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.644701515968707e-05,
+      "loss": 0.9217,
+      "step": 6088
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.641473867105509e-05,
+      "loss": 1.6574,
+      "step": 6089
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.638247001143821e-05,
+      "loss": 0.7475,
+      "step": 6090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.635020918555102e-05,
+      "loss": 1.05,
+      "step": 6091
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.631795619810701e-05,
+      "loss": 0.7319,
+      "step": 6092
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6285711053818434e-05,
+      "loss": 0.4804,
+      "step": 6093
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6253473757396336e-05,
+      "loss": 0.6835,
+      "step": 6094
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.622124431355088e-05,
+      "loss": 1.4044,
+      "step": 6095
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.618902272699083e-05,
+      "loss": 1.4418,
+      "step": 6096
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.615680900242384e-05,
+      "loss": 1.2439,
+      "step": 6097
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.612460314455651e-05,
+      "loss": 1.6216,
+      "step": 6098
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6092405158094274e-05,
+      "loss": 1.1308,
+      "step": 6099
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6060215047741316e-05,
+      "loss": 0.8806,
+      "step": 6100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.602803281820076e-05,
+      "loss": 0.4071,
+      "step": 6101
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5995858474174603e-05,
+      "loss": 0.6429,
+      "step": 6102
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.596369202036356e-05,
+      "loss": 0.9484,
+      "step": 6103
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.593153346146735e-05,
+      "loss": 1.3031,
+      "step": 6104
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5899382802184396e-05,
+      "loss": 1.0099,
+      "step": 6105
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.586724004721207e-05,
+      "loss": 0.7596,
+      "step": 6106
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.583510520124659e-05,
+      "loss": 0.8173,
+      "step": 6107
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.580297826898289e-05,
+      "loss": 1.0307,
+      "step": 6108
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5770859255114906e-05,
+      "loss": 1.4277,
+      "step": 6109
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5738748164335364e-05,
+      "loss": 1.4818,
+      "step": 6110
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.57066450013358e-05,
+      "loss": 1.2781,
+      "step": 6111
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5674549770806574e-05,
+      "loss": 1.2097,
+      "step": 6112
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.564246247743694e-05,
+      "loss": 1.4006,
+      "step": 6113
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.561038312591503e-05,
+      "loss": 0.8418,
+      "step": 6114
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.557831172092773e-05,
+      "loss": 1.0725,
+      "step": 6115
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5546248267160716e-05,
+      "loss": 1.1509,
+      "step": 6116
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5514192769298725e-05,
+      "loss": 0.8703,
+      "step": 6117
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.548214523202513e-05,
+      "loss": 1.3926,
+      "step": 6118
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.545010566002215e-05,
+      "loss": 0.9218,
+      "step": 6119
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.541807405797093e-05,
+      "loss": 1.2928,
+      "step": 6120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5386050430551454e-05,
+      "loss": 0.5932,
+      "step": 6121
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.535403478244242e-05,
+      "loss": 1.0639,
+      "step": 6122
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5322027118321475e-05,
+      "loss": 0.8658,
+      "step": 6123
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5290027442865114e-05,
+      "loss": 0.5185,
+      "step": 6124
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.525803576074853e-05,
+      "loss": 0.7004,
+      "step": 6125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5226052076645896e-05,
+      "loss": 1.2034,
+      "step": 6126
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.51940763952301e-05,
+      "loss": 1.1131,
+      "step": 6127
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.516210872117293e-05,
+      "loss": 1.0029,
+      "step": 6128
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5130149059145035e-05,
+      "loss": 1.3136,
+      "step": 6129
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.509819741381578e-05,
+      "loss": 1.0019,
+      "step": 6130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.506625378985344e-05,
+      "loss": 1.3609,
+      "step": 6131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.503431819192515e-05,
+      "loss": 1.4705,
+      "step": 6132
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.50023906246968e-05,
+      "loss": 1.0838,
+      "step": 6133
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.497047109283309e-05,
+      "loss": 0.7605,
+      "step": 6134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.493855960099761e-05,
+      "loss": 1.225,
+      "step": 6135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.49066561538528e-05,
+      "loss": 0.6384,
+      "step": 6136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4874760756059796e-05,
+      "loss": 1.0618,
+      "step": 6137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4842873412278694e-05,
+      "loss": 0.6383,
+      "step": 6138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.481099412716838e-05,
+      "loss": 1.6668,
+      "step": 6139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.477912290538653e-05,
+      "loss": 1.1338,
+      "step": 6140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.474725975158958e-05,
+      "loss": 0.4984,
+      "step": 6141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.471540467043291e-05,
+      "loss": 1.2056,
+      "step": 6142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.468355766657072e-05,
+      "loss": 1.0718,
+      "step": 6143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4651718744655904e-05,
+      "loss": 1.5571,
+      "step": 6144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4619887909340285e-05,
+      "loss": 1.4097,
+      "step": 6145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4588065165274504e-05,
+      "loss": 1.1914,
+      "step": 6146
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4556250517107924e-05,
+      "loss": 0.9089,
+      "step": 6147
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.452444396948885e-05,
+      "loss": 1.405,
+      "step": 6148
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.449264552706428e-05,
+      "loss": 1.3901,
+      "step": 6149
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.446085519448013e-05,
+      "loss": 1.4254,
+      "step": 6150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4429072976381114e-05,
+      "loss": 1.3776,
+      "step": 6151
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4397298877410676e-05,
+      "loss": 0.698,
+      "step": 6152
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.436553290221117e-05,
+      "loss": 1.0384,
+      "step": 6153
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4333775055423755e-05,
+      "loss": 0.6627,
+      "step": 6154
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.430202534168836e-05,
+      "loss": 1.5058,
+      "step": 6155
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.427028376564364e-05,
+      "loss": 1.4841,
+      "step": 6156
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.423855033192733e-05,
+      "loss": 1.108,
+      "step": 6157
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4206825045175737e-05,
+      "loss": 1.198,
+      "step": 6158
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4175107910024005e-05,
+      "loss": 0.8543,
+      "step": 6159
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.414339893110616e-05,
+      "loss": 1.1448,
+      "step": 6160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4111698113055054e-05,
+      "loss": 1.0223,
+      "step": 6161
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.408000546050226e-05,
+      "loss": 1.3205,
+      "step": 6162
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.404832097807817e-05,
+      "loss": 1.1978,
+      "step": 6163
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.401664467041203e-05,
+      "loss": 1.6039,
+      "step": 6164
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.398497654213192e-05,
+      "loss": 1.1224,
+      "step": 6165
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.39533165978646e-05,
+      "loss": 1.4433,
+      "step": 6166
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3921664842235744e-05,
+      "loss": 1.3681,
+      "step": 6167
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.389002127986984e-05,
+      "loss": 0.818,
+      "step": 6168
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3858385915390064e-05,
+      "loss": 0.7222,
+      "step": 6169
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.382675875341853e-05,
+      "loss": 1.1189,
+      "step": 6170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3795139798576026e-05,
+      "loss": 1.1714,
+      "step": 6171
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.376352905548223e-05,
+      "loss": 1.2929,
+      "step": 6172
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.373192652875564e-05,
+      "loss": 1.2099,
+      "step": 6173
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.370033222301343e-05,
+      "loss": 1.3564,
+      "step": 6174
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3668746142871687e-05,
+      "loss": 0.8129,
+      "step": 6175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.363716829294531e-05,
+      "loss": 1.4575,
+      "step": 6176
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.360559867784788e-05,
+      "loss": 1.4247,
+      "step": 6177
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.357403730219179e-05,
+      "loss": 1.0933,
+      "step": 6178
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.354248417058844e-05,
+      "loss": 0.6986,
+      "step": 6179
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.351093928764776e-05,
+      "loss": 1.2736,
+      "step": 6180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.347940265797855e-05,
+      "loss": 0.5643,
+      "step": 6181
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3447874286188496e-05,
+      "loss": 1.6639,
+      "step": 6182
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.341635417688402e-05,
+      "loss": 0.6212,
+      "step": 6183
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3384842334670287e-05,
+      "loss": 1.2966,
+      "step": 6184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.335333876415131e-05,
+      "loss": 1.0577,
+      "step": 6185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3321843469929935e-05,
+      "loss": 1.0847,
+      "step": 6186
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3290356456607714e-05,
+      "loss": 1.0466,
+      "step": 6187
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.325887772878498e-05,
+      "loss": 1.4185,
+      "step": 6188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.322740729106093e-05,
+      "loss": 0.6486,
+      "step": 6189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.319594514803357e-05,
+      "loss": 1.3749,
+      "step": 6190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.316449130429954e-05,
+      "loss": 0.8541,
+      "step": 6191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.313304576445447e-05,
+      "loss": 1.1981,
+      "step": 6192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.310160853309259e-05,
+      "loss": 1.4392,
+      "step": 6193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.307017961480703e-05,
+      "loss": 1.2853,
+      "step": 6194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.303875901418973e-05,
+      "loss": 1.3683,
+      "step": 6195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.300734673583128e-05,
+      "loss": 1.3394,
+      "step": 6196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.297594278432118e-05,
+      "loss": 1.4245,
+      "step": 6197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2944547164247695e-05,
+      "loss": 0.6947,
+      "step": 6198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.291315988019782e-05,
+      "loss": 1.1493,
+      "step": 6199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2881780936757276e-05,
+      "loss": 0.5668,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2850410338510815e-05,
+      "loss": 0.8928,
+      "step": 6201
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2819048090041705e-05,
+      "loss": 0.4739,
+      "step": 6202
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.278769419593207e-05,
+      "loss": 0.4847,
+      "step": 6203
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.275634866076287e-05,
+      "loss": 1.2872,
+      "step": 6204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2725011489113854e-05,
+      "loss": 1.4777,
+      "step": 6205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.269368268556342e-05,
+      "loss": 1.6685,
+      "step": 6206
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.266236225468886e-05,
+      "loss": 1.2669,
+      "step": 6207
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.263105020106627e-05,
+      "loss": 1.0223,
+      "step": 6208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.259974652927037e-05,
+      "loss": 0.5899,
+      "step": 6209
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2568451243874816e-05,
+      "loss": 0.8652,
+      "step": 6210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.253716434945192e-05,
+      "loss": 0.6401,
+      "step": 6211
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.250588585057288e-05,
+      "loss": 1.4956,
+      "step": 6212
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2474615751807534e-05,
+      "loss": 1.2758,
+      "step": 6213
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.244335405772464e-05,
+      "loss": 1.5784,
+      "step": 6214
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.241210077289157e-05,
+      "loss": 1.3398,
+      "step": 6215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.238085590187461e-05,
+      "loss": 0.9029,
+      "step": 6216
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2349619449238774e-05,
+      "loss": 0.9615,
+      "step": 6217
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.231839141954775e-05,
+      "loss": 1.3585,
+      "step": 6218
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.228717181736413e-05,
+      "loss": 1.1286,
+      "step": 6219
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2255960647249256e-05,
+      "loss": 1.1798,
+      "step": 6220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.222475791376315e-05,
+      "loss": 0.8991,
+      "step": 6221
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.219356362146459e-05,
+      "loss": 1.4376,
+      "step": 6222
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.216237777491132e-05,
+      "loss": 1.3285,
+      "step": 6223
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.213120037865965e-05,
+      "loss": 0.5036,
+      "step": 6224
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2100031437264684e-05,
+      "loss": 1.2991,
+      "step": 6225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.206887095528035e-05,
+      "loss": 1.0821,
+      "step": 6226
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2037718937259366e-05,
+      "loss": 1.3558,
+      "step": 6227
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.200657538775308e-05,
+      "loss": 0.9284,
+      "step": 6228
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1975440311311734e-05,
+      "loss": 1.2491,
+      "step": 6229
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.19443137124843e-05,
+      "loss": 1.3066,
+      "step": 6230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1913195595818435e-05,
+      "loss": 1.0173,
+      "step": 6231
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1882085965860695e-05,
+      "loss": 0.8195,
+      "step": 6232
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1850984827156234e-05,
+      "loss": 1.5025,
+      "step": 6233
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.18198921842491e-05,
+      "loss": 0.7723,
+      "step": 6234
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.178880804168206e-05,
+      "loss": 0.4209,
+      "step": 6235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.175773240399659e-05,
+      "loss": 1.2762,
+      "step": 6236
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1726665275733e-05,
+      "loss": 1.408,
+      "step": 6237
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.169560666143027e-05,
+      "loss": 0.9245,
+      "step": 6238
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.166455656562625e-05,
+      "loss": 1.339,
+      "step": 6239
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1633514992857406e-05,
+      "loss": 0.7208,
+      "step": 6240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.160248194765907e-05,
+      "loss": 1.455,
+      "step": 6241
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.157145743456532e-05,
+      "loss": 1.4358,
+      "step": 6242
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1540441458108925e-05,
+      "loss": 1.2309,
+      "step": 6243
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.150943402282137e-05,
+      "loss": 1.2037,
+      "step": 6244
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.147843513323312e-05,
+      "loss": 0.9124,
+      "step": 6245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.144744479387314e-05,
+      "loss": 0.5696,
+      "step": 6246
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.14164630092692e-05,
+      "loss": 1.4451,
+      "step": 6247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1385489783947914e-05,
+      "loss": 1.4485,
+      "step": 6248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1354525122434615e-05,
+      "loss": 0.7907,
+      "step": 6249
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.132356902925329e-05,
+      "loss": 1.1042,
+      "step": 6250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.12926215089268e-05,
+      "loss": 1.5366,
+      "step": 6251
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.12616825659767e-05,
+      "loss": 1.2251,
+      "step": 6252
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.123075220492324e-05,
+      "loss": 0.9092,
+      "step": 6253
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1199830430285534e-05,
+      "loss": 0.7308,
+      "step": 6254
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.11689172465813e-05,
+      "loss": 0.5749,
+      "step": 6255
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.113801265832711e-05,
+      "loss": 1.2038,
+      "step": 6256
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.110711667003827e-05,
+      "loss": 0.5606,
+      "step": 6257
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.107622928622876e-05,
+      "loss": 0.5044,
+      "step": 6258
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.104535051141136e-05,
+      "loss": 1.5606,
+      "step": 6259
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.101448035009762e-05,
+      "loss": 1.3098,
+      "step": 6260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.098361880679775e-05,
+      "loss": 1.3955,
+      "step": 6261
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.0952765886020725e-05,
+      "loss": 1.0346,
+      "step": 6262
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.092192159227429e-05,
+      "loss": 1.1715,
+      "step": 6263
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.089108593006496e-05,
+      "loss": 1.1536,
+      "step": 6264
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.086025890389792e-05,
+      "loss": 1.2906,
+      "step": 6265
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.0829440518277026e-05,
+      "loss": 1.0127,
+      "step": 6266
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.079863077770514e-05,
+      "loss": 1.285,
+      "step": 6267
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.0767829686683565e-05,
+      "loss": 1.3112,
+      "step": 6268
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.0737037249712476e-05,
+      "loss": 0.9891,
+      "step": 6269
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.0706253471290756e-05,
+      "loss": 1.1605,
+      "step": 6270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.067547835591611e-05,
+      "loss": 0.5842,
+      "step": 6271
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.064471190808481e-05,
+      "loss": 1.6011,
+      "step": 6272
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.0613954132292e-05,
+      "loss": 1.1495,
+      "step": 6273
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.058320503303152e-05,
+      "loss": 1.2459,
+      "step": 6274
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.05524646147959e-05,
+      "loss": 1.1744,
+      "step": 6275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.052173288207649e-05,
+      "loss": 0.8395,
+      "step": 6276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.0491009839363234e-05,
+      "loss": 1.0461,
+      "step": 6277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.046029549114493e-05,
+      "loss": 0.4445,
+      "step": 6278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.042958984190911e-05,
+      "loss": 1.0345,
+      "step": 6279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.039889289614189e-05,
+      "loss": 0.6952,
+      "step": 6280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.036820465832829e-05,
+      "loss": 1.1899,
+      "step": 6281
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.033752513295198e-05,
+      "loss": 0.9142,
+      "step": 6282
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.030685432449535e-05,
+      "loss": 1.0029,
+      "step": 6283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.0276192237439467e-05,
+      "loss": 1.2626,
+      "step": 6284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.024553887626422e-05,
+      "loss": 0.8357,
+      "step": 6285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.021489424544823e-05,
+      "loss": 1.2786,
+      "step": 6286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.018425834946873e-05,
+      "loss": 1.3327,
+      "step": 6287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.0153631192801754e-05,
+      "loss": 0.8517,
+      "step": 6288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.012301277992211e-05,
+      "loss": 1.4771,
+      "step": 6289
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.0092403115303225e-05,
+      "loss": 1.3136,
+      "step": 6290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.006180220341724e-05,
+      "loss": 1.295,
+      "step": 6291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.003121004873512e-05,
+      "loss": 0.3584,
+      "step": 6292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.0000626655726527e-05,
+      "loss": 0.5535,
+      "step": 6293
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.997005202885975e-05,
+      "loss": 1.5447,
+      "step": 6294
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9939486172601883e-05,
+      "loss": 1.3466,
+      "step": 6295
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9908929091418754e-05,
+      "loss": 1.4255,
+      "step": 6296
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.987838078977481e-05,
+      "loss": 0.5306,
+      "step": 6297
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.984784127213334e-05,
+      "loss": 1.1022,
+      "step": 6298
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.981731054295622e-05,
+      "loss": 0.978,
+      "step": 6299
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.978678860670415e-05,
+      "loss": 0.6847,
+      "step": 6300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9756275467836526e-05,
+      "loss": 0.9959,
+      "step": 6301
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.972577113081136e-05,
+      "loss": 0.9294,
+      "step": 6302
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.969527560008551e-05,
+      "loss": 1.1779,
+      "step": 6303
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.966478888011451e-05,
+      "loss": 1.4234,
+      "step": 6304
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.963431097535257e-05,
+      "loss": 1.2356,
+      "step": 6305
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9603841890252547e-05,
+      "loss": 1.3952,
+      "step": 6306
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.957338162926623e-05,
+      "loss": 1.1593,
+      "step": 6307
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.954293019684392e-05,
+      "loss": 0.4022,
+      "step": 6308
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9512487597434655e-05,
+      "loss": 1.1837,
+      "step": 6309
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.948205383548625e-05,
+      "loss": 1.6373,
+      "step": 6310
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.945162891544524e-05,
+      "loss": 1.2449,
+      "step": 6311
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.942121284175677e-05,
+      "loss": 0.5415,
+      "step": 6312
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9390805618864724e-05,
+      "loss": 1.1456,
+      "step": 6313
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9360407251211753e-05,
+      "loss": 1.366,
+      "step": 6314
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.933001774323919e-05,
+      "loss": 1.3969,
+      "step": 6315
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9299637099387024e-05,
+      "loss": 1.0826,
+      "step": 6316
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9269265324093994e-05,
+      "loss": 0.9558,
+      "step": 6317
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9238902421797586e-05,
+      "loss": 0.6046,
+      "step": 6318
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.920854839693386e-05,
+      "loss": 1.2435,
+      "step": 6319
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.917820325393773e-05,
+      "loss": 1.3919,
+      "step": 6320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9147866997242665e-05,
+      "loss": 1.1792,
+      "step": 6321
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.911753963128095e-05,
+      "loss": 0.5495,
+      "step": 6322
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.908722116048357e-05,
+      "loss": 1.4705,
+      "step": 6323
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.905691158928009e-05,
+      "loss": 1.1724,
+      "step": 6324
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9026610922098895e-05,
+      "loss": 0.7383,
+      "step": 6325
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8996319163367065e-05,
+      "loss": 0.8595,
+      "step": 6326
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.896603631751031e-05,
+      "loss": 0.9653,
+      "step": 6327
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.893576238895301e-05,
+      "loss": 1.199,
+      "step": 6328
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.890549738211843e-05,
+      "loss": 1.3974,
+      "step": 6329
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8875241301428345e-05,
+      "loss": 0.9938,
+      "step": 6330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.884499415130325e-05,
+      "loss": 1.4261,
+      "step": 6331
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8814755936162406e-05,
+      "loss": 1.1717,
+      "step": 6332
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.878452666042376e-05,
+      "loss": 1.1893,
+      "step": 6333
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8754306328503865e-05,
+      "loss": 1.1901,
+      "step": 6334
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8724094944818104e-05,
+      "loss": 1.1151,
+      "step": 6335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8693892513780394e-05,
+      "loss": 0.6712,
+      "step": 6336
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8663699039803506e-05,
+      "loss": 1.4634,
+      "step": 6337
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.863351452729875e-05,
+      "loss": 1.07,
+      "step": 6338
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.860333898067624e-05,
+      "loss": 1.2603,
+      "step": 6339
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.857317240434478e-05,
+      "loss": 1.1657,
+      "step": 6340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.854301480271175e-05,
+      "loss": 1.24,
+      "step": 6341
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.851286618018336e-05,
+      "loss": 1.4809,
+      "step": 6342
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.848272654116437e-05,
+      "loss": 1.2715,
+      "step": 6343
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.845259589005833e-05,
+      "loss": 0.9356,
+      "step": 6344
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.842247423126749e-05,
+      "loss": 1.2841,
+      "step": 6345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.839236156919268e-05,
+      "loss": 1.5169,
+      "step": 6346
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.836225790823349e-05,
+      "loss": 1.1684,
+      "step": 6347
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8332163252788226e-05,
+      "loss": 1.2422,
+      "step": 6348
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8302077607253794e-05,
+      "loss": 1.492,
+      "step": 6349
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.827200097602577e-05,
+      "loss": 1.3103,
+      "step": 6350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.824193336349859e-05,
+      "loss": 1.2211,
+      "step": 6351
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.82118747740652e-05,
+      "loss": 1.5492,
+      "step": 6352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.818182521211722e-05,
+      "loss": 0.591,
+      "step": 6353
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.815178468204504e-05,
+      "loss": 0.62,
+      "step": 6354
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.812175318823776e-05,
+      "loss": 0.4369,
+      "step": 6355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.809173073508301e-05,
+      "loss": 1.2079,
+      "step": 6356
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8061717326967215e-05,
+      "loss": 0.765,
+      "step": 6357
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.80317129682755e-05,
+      "loss": 1.6017,
+      "step": 6358
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.800171766339155e-05,
+      "loss": 0.6041,
+      "step": 6359
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7971731416697844e-05,
+      "loss": 1.3136,
+      "step": 6360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7941754232575444e-05,
+      "loss": 0.935,
+      "step": 6361
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7911786115404194e-05,
+      "loss": 0.7391,
+      "step": 6362
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.788182706956248e-05,
+      "loss": 0.5232,
+      "step": 6363
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.785187709942751e-05,
+      "loss": 0.8939,
+      "step": 6364
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.782193620937502e-05,
+      "loss": 1.3925,
+      "step": 6365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7792004403779526e-05,
+      "loss": 0.7721,
+      "step": 6366
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7762081687014204e-05,
+      "loss": 0.6565,
+      "step": 6367
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.773216806345082e-05,
+      "loss": 1.1252,
+      "step": 6368
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7702263537459914e-05,
+      "loss": 1.3348,
+      "step": 6369
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.767236811341067e-05,
+      "loss": 1.1673,
+      "step": 6370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7642481795670915e-05,
+      "loss": 1.627,
+      "step": 6371
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.761260458860706e-05,
+      "loss": 1.3144,
+      "step": 6372
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7582736496584436e-05,
+      "loss": 0.9202,
+      "step": 6373
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7552877523966824e-05,
+      "loss": 1.5984,
+      "step": 6374
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7523027675116674e-05,
+      "loss": 1.4552,
+      "step": 6375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.749318695439523e-05,
+      "loss": 1.3675,
+      "step": 6376
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7463355366162356e-05,
+      "loss": 1.0682,
+      "step": 6377
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.74335329147765e-05,
+      "loss": 1.3837,
+      "step": 6378
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7403719604594866e-05,
+      "loss": 0.4627,
+      "step": 6379
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.737391543997334e-05,
+      "loss": 1.1201,
+      "step": 6380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.734412042526635e-05,
+      "loss": 1.5794,
+      "step": 6381
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7314334564827125e-05,
+      "loss": 1.456,
+      "step": 6382
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.728455786300744e-05,
+      "loss": 1.171,
+      "step": 6383
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7254790324157815e-05,
+      "loss": 0.9117,
+      "step": 6384
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7225031952627444e-05,
+      "loss": 1.5434,
+      "step": 6385
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.719528275276406e-05,
+      "loss": 0.5603,
+      "step": 6386
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.716554272891421e-05,
+      "loss": 0.9491,
+      "step": 6387
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.713581188542297e-05,
+      "loss": 0.6195,
+      "step": 6388
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.710609022663418e-05,
+      "loss": 1.198,
+      "step": 6389
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.707637775689024e-05,
+      "loss": 1.2044,
+      "step": 6390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.704667448053228e-05,
+      "loss": 0.935,
+      "step": 6391
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7016980401900105e-05,
+      "loss": 1.5286,
+      "step": 6392
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.69872955253321e-05,
+      "loss": 0.9918,
+      "step": 6393
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.695761985516527e-05,
+      "loss": 0.5569,
+      "step": 6394
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.692795339573548e-05,
+      "loss": 1.5794,
+      "step": 6395
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6898296151377054e-05,
+      "loss": 1.2923,
+      "step": 6396
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.686864812642299e-05,
+      "loss": 1.5654,
+      "step": 6397
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.683900932520502e-05,
+      "loss": 0.6087,
+      "step": 6398
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.680937975205353e-05,
+      "loss": 1.2595,
+      "step": 6399
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.677975941129742e-05,
+      "loss": 0.6565,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.67501483072644e-05,
+      "loss": 1.3367,
+      "step": 6401
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.672054644428079e-05,
+      "loss": 1.2608,
+      "step": 6402
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.669095382667146e-05,
+      "loss": 1.2179,
+      "step": 6403
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.666137045876009e-05,
+      "loss": 1.2023,
+      "step": 6404
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6631796344868854e-05,
+      "loss": 0.5075,
+      "step": 6405
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.660223148931866e-05,
+      "loss": 1.1078,
+      "step": 6406
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.657267589642912e-05,
+      "loss": 0.9036,
+      "step": 6407
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.654312957051832e-05,
+      "loss": 1.0332,
+      "step": 6408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.651359251590314e-05,
+      "loss": 1.0708,
+      "step": 6409
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6484064736899094e-05,
+      "loss": 1.1799,
+      "step": 6410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.645454623782026e-05,
+      "loss": 1.0228,
+      "step": 6411
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.642503702297938e-05,
+      "loss": 1.3063,
+      "step": 6412
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.639553709668791e-05,
+      "loss": 0.8935,
+      "step": 6413
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.636604646325592e-05,
+      "loss": 1.0336,
+      "step": 6414
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.633656512699208e-05,
+      "loss": 0.9509,
+      "step": 6415
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.630709309220366e-05,
+      "loss": 1.2886,
+      "step": 6416
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.627763036319678e-05,
+      "loss": 0.5524,
+      "step": 6417
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.624817694427598e-05,
+      "loss": 1.3576,
+      "step": 6418
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6218732839744486e-05,
+      "loss": 0.5629,
+      "step": 6419
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.618929805390424e-05,
+      "loss": 1.2701,
+      "step": 6420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.615987259105581e-05,
+      "loss": 1.2502,
+      "step": 6421
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.61304564554983e-05,
+      "loss": 1.2334,
+      "step": 6422
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.610104965152955e-05,
+      "loss": 1.5647,
+      "step": 6423
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6071652183446056e-05,
+      "loss": 0.8999,
+      "step": 6424
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.604226405554282e-05,
+      "loss": 1.3969,
+      "step": 6425
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.601288527211364e-05,
+      "loss": 1.3245,
+      "step": 6426
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5983515837450795e-05,
+      "loss": 0.8645,
+      "step": 6427
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.595415575584532e-05,
+      "loss": 0.919,
+      "step": 6428
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.592480503158686e-05,
+      "loss": 1.3299,
+      "step": 6429
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5895463668963604e-05,
+      "loss": 1.2009,
+      "step": 6430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.586613167226247e-05,
+      "loss": 0.8,
+      "step": 6431
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.583680904576901e-05,
+      "loss": 0.6682,
+      "step": 6432
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.580749579376737e-05,
+      "loss": 1.3023,
+      "step": 6433
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5778191920540246e-05,
+      "loss": 1.2602,
+      "step": 6434
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5748897430369124e-05,
+      "loss": 1.3546,
+      "step": 6435
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.571961232753405e-05,
+      "loss": 1.4493,
+      "step": 6436
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.569033661631364e-05,
+      "loss": 1.2184,
+      "step": 6437
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.56610703009852e-05,
+      "loss": 0.8439,
+      "step": 6438
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.563181338582472e-05,
+      "loss": 1.0526,
+      "step": 6439
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.560256587510668e-05,
+      "loss": 1.5011,
+      "step": 6440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.557332777310426e-05,
+      "loss": 0.5651,
+      "step": 6441
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5544099084089265e-05,
+      "loss": 1.0354,
+      "step": 6442
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5514879812332166e-05,
+      "loss": 1.1596,
+      "step": 6443
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.548566996210194e-05,
+      "loss": 1.3737,
+      "step": 6444
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.545646953766629e-05,
+      "loss": 0.6627,
+      "step": 6445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.542727854329155e-05,
+      "loss": 1.156,
+      "step": 6446
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.539809698324258e-05,
+      "loss": 1.144,
+      "step": 6447
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.536892486178297e-05,
+      "loss": 0.6615,
+      "step": 6448
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.533976218317482e-05,
+      "loss": 1.5406,
+      "step": 6449
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.531060895167895e-05,
+      "loss": 0.9887,
+      "step": 6450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.528146517155478e-05,
+      "loss": 0.8118,
+      "step": 6451
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5252330847060286e-05,
+      "loss": 1.3083,
+      "step": 6452
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5223205982452115e-05,
+      "loss": 1.2636,
+      "step": 6453
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.519409058198557e-05,
+      "loss": 1.2563,
+      "step": 6454
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.516498464991449e-05,
+      "loss": 1.4543,
+      "step": 6455
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.513588819049128e-05,
+      "loss": 1.5773,
+      "step": 6456
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.510680120796722e-05,
+      "loss": 1.2903,
+      "step": 6457
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5077723706591924e-05,
+      "loss": 0.4199,
+      "step": 6458
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.504865569061371e-05,
+      "loss": 1.0075,
+      "step": 6459
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.501959716427957e-05,
+      "loss": 0.9851,
+      "step": 6460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.499054813183509e-05,
+      "loss": 0.9187,
+      "step": 6461
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4961508597524386e-05,
+      "loss": 1.1889,
+      "step": 6462
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.49324785655903e-05,
+      "loss": 1.1064,
+      "step": 6463
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.490345804027418e-05,
+      "loss": 1.128,
+      "step": 6464
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.48744470258161e-05,
+      "loss": 0.696,
+      "step": 6465
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4845445526454626e-05,
+      "loss": 1.3789,
+      "step": 6466
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4816453546427005e-05,
+      "loss": 1.0703,
+      "step": 6467
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.478747108996913e-05,
+      "loss": 1.0021,
+      "step": 6468
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4758498161315355e-05,
+      "loss": 0.8341,
+      "step": 6469
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.472953476469884e-05,
+      "loss": 0.763,
+      "step": 6470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4700580904351155e-05,
+      "loss": 1.2142,
+      "step": 6471
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.467163658450262e-05,
+      "loss": 1.4179,
+      "step": 6472
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.464270180938215e-05,
+      "loss": 1.0872,
+      "step": 6473
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.461377658321715e-05,
+      "loss": 1.5972,
+      "step": 6474
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.458486091023374e-05,
+      "loss": 1.5951,
+      "step": 6475
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.455595479465665e-05,
+      "loss": 1.3059,
+      "step": 6476
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4527058240709154e-05,
+      "loss": 1.1264,
+      "step": 6477
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4498171252613064e-05,
+      "loss": 1.296,
+      "step": 6478
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.446929383458902e-05,
+      "loss": 1.5828,
+      "step": 6479
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.444042599085607e-05,
+      "loss": 0.5146,
+      "step": 6480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4411567725631886e-05,
+      "loss": 1.0877,
+      "step": 6481
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.438271904313279e-05,
+      "loss": 1.0744,
+      "step": 6482
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.435387994757373e-05,
+      "loss": 0.7139,
+      "step": 6483
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.432505044316814e-05,
+      "loss": 1.9693,
+      "step": 6484
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.429623053412818e-05,
+      "loss": 1.1548,
+      "step": 6485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4267420224664516e-05,
+      "loss": 1.3599,
+      "step": 6486
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4238619518986445e-05,
+      "loss": 1.0497,
+      "step": 6487
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.420982842130191e-05,
+      "loss": 1.0343,
+      "step": 6488
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.418104693581734e-05,
+      "loss": 1.466,
+      "step": 6489
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.415227506673788e-05,
+      "loss": 1.1689,
+      "step": 6490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4123512818267145e-05,
+      "loss": 1.146,
+      "step": 6491
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.409476019460749e-05,
+      "loss": 1.4496,
+      "step": 6492
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.40660171999597e-05,
+      "loss": 1.4352,
+      "step": 6493
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4037283838523295e-05,
+      "loss": 0.5849,
+      "step": 6494
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.400856011449633e-05,
+      "loss": 1.329,
+      "step": 6495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3979846032075425e-05,
+      "loss": 1.1046,
+      "step": 6496
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.395114159545584e-05,
+      "loss": 1.1432,
+      "step": 6497
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.392244680883142e-05,
+      "loss": 1.4613,
+      "step": 6498
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3893761676394566e-05,
+      "loss": 1.5731,
+      "step": 6499
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3865086202336226e-05,
+      "loss": 1.6619,
+      "step": 6500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3836420390846116e-05,
+      "loss": 0.9875,
+      "step": 6501
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3807764246112375e-05,
+      "loss": 1.3814,
+      "step": 6502
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3779117772321724e-05,
+      "loss": 0.6512,
+      "step": 6503
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.375048097365957e-05,
+      "loss": 1.4633,
+      "step": 6504
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.372185385430989e-05,
+      "loss": 0.567,
+      "step": 6505
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.369323641845516e-05,
+      "loss": 0.4904,
+      "step": 6506
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3664628670276524e-05,
+      "loss": 1.6888,
+      "step": 6507
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.363603061395371e-05,
+      "loss": 1.7469,
+      "step": 6508
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3607442253664956e-05,
+      "loss": 0.5755,
+      "step": 6509
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.357886359358718e-05,
+      "loss": 1.0167,
+      "step": 6510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.355029463789578e-05,
+      "loss": 1.1044,
+      "step": 6511
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3521735390764844e-05,
+      "loss": 0.8401,
+      "step": 6512
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3493185856366985e-05,
+      "loss": 1.3042,
+      "step": 6513
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3464646038873395e-05,
+      "loss": 1.3263,
+      "step": 6514
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3436115942453795e-05,
+      "loss": 0.9008,
+      "step": 6515
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.340759557127661e-05,
+      "loss": 0.8795,
+      "step": 6516
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.337908492950877e-05,
+      "loss": 1.2568,
+      "step": 6517
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.335058402131577e-05,
+      "loss": 1.2118,
+      "step": 6518
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3322092850861694e-05,
+      "loss": 0.8736,
+      "step": 6519
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.329361142230927e-05,
+      "loss": 0.9346,
+      "step": 6520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3265139739819706e-05,
+      "loss": 1.585,
+      "step": 6521
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3236677807552765e-05,
+      "loss": 1.4059,
+      "step": 6522
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3208225629666966e-05,
+      "loss": 0.8947,
+      "step": 6523
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.317978321031923e-05,
+      "loss": 1.4259,
+      "step": 6524
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3151350553665064e-05,
+      "loss": 1.1829,
+      "step": 6525
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3122927663858625e-05,
+      "loss": 1.0286,
+      "step": 6526
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3094514545052635e-05,
+      "loss": 1.3704,
+      "step": 6527
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.306611120139831e-05,
+      "loss": 1.5594,
+      "step": 6528
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.303771763704551e-05,
+      "loss": 0.8238,
+      "step": 6529
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.300933385614268e-05,
+      "loss": 1.147,
+      "step": 6530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2980959862836735e-05,
+      "loss": 1.2719,
+      "step": 6531
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.29525956612733e-05,
+      "loss": 1.4078,
+      "step": 6532
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.292424125559641e-05,
+      "loss": 1.1649,
+      "step": 6533
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.289589664994881e-05,
+      "loss": 1.0421,
+      "step": 6534
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2867561848471775e-05,
+      "loss": 1.1335,
+      "step": 6535
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.28392368553051e-05,
+      "loss": 1.3905,
+      "step": 6536
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2810921674587114e-05,
+      "loss": 1.4193,
+      "step": 6537
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2782616310454906e-05,
+      "loss": 1.1486,
+      "step": 6538
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.275432076704392e-05,
+      "loss": 1.5027,
+      "step": 6539
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.272603504848823e-05,
+      "loss": 1.647,
+      "step": 6540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.269775915892053e-05,
+      "loss": 1.6233,
+      "step": 6541
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.266949310247205e-05,
+      "loss": 1.219,
+      "step": 6542
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.264123688327254e-05,
+      "loss": 1.1104,
+      "step": 6543
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.261299050545028e-05,
+      "loss": 1.4802,
+      "step": 6544
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.258475397313232e-05,
+      "loss": 0.5309,
+      "step": 6545
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.255652729044405e-05,
+      "loss": 1.2006,
+      "step": 6546
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.252831046150946e-05,
+      "loss": 1.1684,
+      "step": 6547
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2500103490451184e-05,
+      "loss": 1.4279,
+      "step": 6548
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2471906381390394e-05,
+      "loss": 1.3887,
+      "step": 6549
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.244371913844674e-05,
+      "loss": 0.5406,
+      "step": 6550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.24155417657385e-05,
+      "loss": 1.5852,
+      "step": 6551
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.238737426738256e-05,
+      "loss": 0.6394,
+      "step": 6552
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2359216647494215e-05,
+      "loss": 1.635,
+      "step": 6553
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.233106891018749e-05,
+      "loss": 1.7136,
+      "step": 6554
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.230293105957478e-05,
+      "loss": 0.7039,
+      "step": 6555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.227480309976719e-05,
+      "loss": 1.0778,
+      "step": 6556
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2246685034874355e-05,
+      "loss": 1.3162,
+      "step": 6557
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.221857686900437e-05,
+      "loss": 1.4167,
+      "step": 6558
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.219047860626398e-05,
+      "loss": 1.0503,
+      "step": 6559
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.216239025075847e-05,
+      "loss": 1.2916,
+      "step": 6560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2134311806591635e-05,
+      "loss": 0.8916,
+      "step": 6561
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.210624327786581e-05,
+      "loss": 1.0527,
+      "step": 6562
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.207818466868195e-05,
+      "loss": 0.8938,
+      "step": 6563
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2050135983139564e-05,
+      "loss": 1.4933,
+      "step": 6564
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2022097225336634e-05,
+      "loss": 1.2399,
+      "step": 6565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.199406839936967e-05,
+      "loss": 1.4129,
+      "step": 6566
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.196604950933393e-05,
+      "loss": 1.0695,
+      "step": 6567
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1938040559323e-05,
+      "loss": 1.3746,
+      "step": 6568
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1910041553429085e-05,
+      "loss": 0.6576,
+      "step": 6569
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1882052495742966e-05,
+      "loss": 0.9628,
+      "step": 6570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.185407339035399e-05,
+      "loss": 1.4085,
+      "step": 6571
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1826104241349944e-05,
+      "loss": 1.0398,
+      "step": 6572
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1798145052817274e-05,
+      "loss": 0.689,
+      "step": 6573
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1770195828840945e-05,
+      "loss": 1.1319,
+      "step": 6574
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1742256573504395e-05,
+      "loss": 0.8837,
+      "step": 6575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.171432729088971e-05,
+      "loss": 1.0719,
+      "step": 6576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.168640798507742e-05,
+      "loss": 1.1049,
+      "step": 6577
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.165849866014665e-05,
+      "loss": 1.1666,
+      "step": 6578
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.163059932017511e-05,
+      "loss": 0.6106,
+      "step": 6579
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1602709969238944e-05,
+      "loss": 1.1805,
+      "step": 6580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.157483061141292e-05,
+      "loss": 1.465,
+      "step": 6581
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1546961250770346e-05,
+      "loss": 1.307,
+      "step": 6582
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.151910189138302e-05,
+      "loss": 1.3798,
+      "step": 6583
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1491252537321255e-05,
+      "loss": 1.3008,
+      "step": 6584
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1463413192654015e-05,
+      "loss": 1.2984,
+      "step": 6585
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.143558386144874e-05,
+      "loss": 1.2332,
+      "step": 6586
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1407764547771344e-05,
+      "loss": 0.8905,
+      "step": 6587
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.137995525568638e-05,
+      "loss": 1.2916,
+      "step": 6588
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.135215598925691e-05,
+      "loss": 0.8984,
+      "step": 6589
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.13243667525445e-05,
+      "loss": 1.5529,
+      "step": 6590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1296587549609226e-05,
+      "loss": 0.6559,
+      "step": 6591
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1268818384509756e-05,
+      "loss": 1.2998,
+      "step": 6592
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.124105926130333e-05,
+      "loss": 1.1696,
+      "step": 6593
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1213310184045584e-05,
+      "loss": 1.1546,
+      "step": 6594
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.11855711567908e-05,
+      "loss": 1.1245,
+      "step": 6595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.11578421835918e-05,
+      "loss": 1.3256,
+      "step": 6596
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.113012326849982e-05,
+      "loss": 1.4086,
+      "step": 6597
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.110241441556477e-05,
+      "loss": 1.2608,
+      "step": 6598
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1074715628834946e-05,
+      "loss": 1.3253,
+      "step": 6599
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.10470269123573e-05,
+      "loss": 0.8257,
+      "step": 6600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1019348270177273e-05,
+      "loss": 0.5682,
+      "step": 6601
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.099167970633878e-05,
+      "loss": 0.4699,
+      "step": 6602
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.096402122488432e-05,
+      "loss": 0.6973,
+      "step": 6603
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0936372829854944e-05,
+      "loss": 1.1733,
+      "step": 6604
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.090873452529015e-05,
+      "loss": 1.0512,
+      "step": 6605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.088110631522798e-05,
+      "loss": 1.1873,
+      "step": 6606
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.085348820370506e-05,
+      "loss": 0.724,
+      "step": 6607
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0825880194756515e-05,
+      "loss": 0.6745,
+      "step": 6608
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0798282292415936e-05,
+      "loss": 0.6621,
+      "step": 6609
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0770694500715514e-05,
+      "loss": 1.1366,
+      "step": 6610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.074311682368596e-05,
+      "loss": 0.7554,
+      "step": 6611
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.071554926535642e-05,
+      "loss": 0.7037,
+      "step": 6612
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.068799182975467e-05,
+      "loss": 1.7392,
+      "step": 6613
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.066044452090691e-05,
+      "loss": 1.4087,
+      "step": 6614
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0632907342837976e-05,
+      "loss": 1.3093,
+      "step": 6615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0605380299571086e-05,
+      "loss": 0.5644,
+      "step": 6616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.057786339512808e-05,
+      "loss": 0.9406,
+      "step": 6617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.055035663352932e-05,
+      "loss": 1.3273,
+      "step": 6618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.052286001879358e-05,
+      "loss": 0.612,
+      "step": 6619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.049537355493829e-05,
+      "loss": 1.3142,
+      "step": 6620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.046789724597927e-05,
+      "loss": 0.8988,
+      "step": 6621
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0440431095930942e-05,
+      "loss": 1.0623,
+      "step": 6622
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0412975108806264e-05,
+      "loss": 0.9166,
+      "step": 6623
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0385529288616575e-05,
+      "loss": 0.9493,
+      "step": 6624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0358093639371854e-05,
+      "loss": 1.0154,
+      "step": 6625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.03306681650806e-05,
+      "loss": 1.0228,
+      "step": 6626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0303252869749745e-05,
+      "loss": 1.1259,
+      "step": 6627
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0275847757384702e-05,
+      "loss": 1.12,
+      "step": 6628
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0248452831989583e-05,
+      "loss": 1.0493,
+      "step": 6629
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0221068097566852e-05,
+      "loss": 1.4572,
+      "step": 6630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0193693558117486e-05,
+      "loss": 1.0729,
+      "step": 6631
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0166329217641032e-05,
+      "loss": 0.5742,
+      "step": 6632
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0138975080135567e-05,
+      "loss": 1.2586,
+      "step": 6633
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0111631149597575e-05,
+      "loss": 0.5545,
+      "step": 6634
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.008429743002217e-05,
+      "loss": 1.2017,
+      "step": 6635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.005697392540284e-05,
+      "loss": 0.5718,
+      "step": 6636
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0029660639731715e-05,
+      "loss": 1.1918,
+      "step": 6637
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0002357576999384e-05,
+      "loss": 0.7448,
+      "step": 6638
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9975064741194868e-05,
+      "loss": 1.3651,
+      "step": 6639
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9947782136305835e-05,
+      "loss": 1.3333,
+      "step": 6640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9920509766318294e-05,
+      "loss": 1.4102,
+      "step": 6641
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9893247635216925e-05,
+      "loss": 1.1221,
+      "step": 6642
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9865995746984754e-05,
+      "loss": 1.5412,
+      "step": 6643
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.983875410560344e-05,
+      "loss": 1.6862,
+      "step": 6644
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9811522715053096e-05,
+      "loss": 0.7462,
+      "step": 6645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9784301579312302e-05,
+      "loss": 0.5769,
+      "step": 6646
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.975709070235818e-05,
+      "loss": 1.2447,
+      "step": 6647
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.972989008816639e-05,
+      "loss": 1.0797,
+      "step": 6648
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9702699740711003e-05,
+      "loss": 1.3995,
+      "step": 6649
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.967551966396458e-05,
+      "loss": 0.6384,
+      "step": 6650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.964834986189837e-05,
+      "loss": 1.2087,
+      "step": 6651
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9621190338481907e-05,
+      "loss": 1.1404,
+      "step": 6652
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.959404109768328e-05,
+      "loss": 0.6876,
+      "step": 6653
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.956690214346913e-05,
+      "loss": 0.9831,
+      "step": 6654
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.95397734798046e-05,
+      "loss": 1.4235,
+      "step": 6655
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.951265511065322e-05,
+      "loss": 0.6041,
+      "step": 6656
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9485547039977125e-05,
+      "loss": 1.3435,
+      "step": 6657
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9458449271736944e-05,
+      "loss": 1.1399,
+      "step": 6658
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.94313618098917e-05,
+      "loss": 0.717,
+      "step": 6659
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.940428465839904e-05,
+      "loss": 1.4741,
+      "step": 6660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9377217821214987e-05,
+      "loss": 0.7681,
+      "step": 6661
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.935016130229412e-05,
+      "loss": 1.4177,
+      "step": 6662
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.932311510558956e-05,
+      "loss": 1.4569,
+      "step": 6663
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.929607923505282e-05,
+      "loss": 1.0314,
+      "step": 6664
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9269053694633897e-05,
+      "loss": 0.5042,
+      "step": 6665
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9242038488281388e-05,
+      "loss": 1.0869,
+      "step": 6666
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9215033619942334e-05,
+      "loss": 1.25,
+      "step": 6667
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.918803909356219e-05,
+      "loss": 1.0389,
+      "step": 6668
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9161054913084995e-05,
+      "loss": 1.1215,
+      "step": 6669
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9134081082453278e-05,
+      "loss": 1.3367,
+      "step": 6670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9107117605607982e-05,
+      "loss": 0.9563,
+      "step": 6671
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9080164486488516e-05,
+      "loss": 1.3694,
+      "step": 6672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9053221729032966e-05,
+      "loss": 1.0094,
+      "step": 6673
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9026289337177704e-05,
+      "loss": 1.1651,
+      "step": 6674
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.899936731485762e-05,
+      "loss": 0.8232,
+      "step": 6675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.897245566600617e-05,
+      "loss": 1.1422,
+      "step": 6676
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8945554394555276e-05,
+      "loss": 0.8872,
+      "step": 6677
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8918663504435263e-05,
+      "loss": 1.1883,
+      "step": 6678
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8891782999575013e-05,
+      "loss": 0.9647,
+      "step": 6679
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8864912883901918e-05,
+      "loss": 1.3831,
+      "step": 6680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8838053161341737e-05,
+      "loss": 0.8406,
+      "step": 6681
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8811203835818835e-05,
+      "loss": 0.83,
+      "step": 6682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8784364911255958e-05,
+      "loss": 1.1178,
+      "step": 6683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8757536391574392e-05,
+      "loss": 1.0693,
+      "step": 6684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8730718280693924e-05,
+      "loss": 0.8831,
+      "step": 6685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8703910582532756e-05,
+      "loss": 0.8792,
+      "step": 6686
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.867711330100753e-05,
+      "loss": 1.2534,
+      "step": 6687
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.865032644003356e-05,
+      "loss": 1.2558,
+      "step": 6688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8623550003524458e-05,
+      "loss": 0.9322,
+      "step": 6689
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8596783995392317e-05,
+      "loss": 0.8287,
+      "step": 6690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8570028419547788e-05,
+      "loss": 1.238,
+      "step": 6691
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8543283279900023e-05,
+      "loss": 1.2441,
+      "step": 6692
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8516548580356528e-05,
+      "loss": 1.2769,
+      "step": 6693
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.848982432482329e-05,
+      "loss": 1.29,
+      "step": 6694
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.846311051720496e-05,
+      "loss": 1.2138,
+      "step": 6695
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8436407161404466e-05,
+      "loss": 1.2485,
+      "step": 6696
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.840971426132324e-05,
+      "loss": 1.6751,
+      "step": 6697
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.838303182086125e-05,
+      "loss": 0.5406,
+      "step": 6698
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8356359843916936e-05,
+      "loss": 1.3656,
+      "step": 6699
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8329698334387102e-05,
+      "loss": 1.4285,
+      "step": 6700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8303047296167152e-05,
+      "loss": 1.2809,
+      "step": 6701
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8276406733150918e-05,
+      "loss": 1.2413,
+      "step": 6702
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.824977664923064e-05,
+      "loss": 1.2779,
+      "step": 6703
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8223157048297134e-05,
+      "loss": 1.3445,
+      "step": 6704
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.819654793423957e-05,
+      "loss": 1.1053,
+      "step": 6705
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8169949310945666e-05,
+      "loss": 1.1317,
+      "step": 6706
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8143361182301608e-05,
+      "loss": 1.2327,
+      "step": 6707
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8116783552191995e-05,
+      "loss": 1.1499,
+      "step": 6708
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8090216424499914e-05,
+      "loss": 1.4084,
+      "step": 6709
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.806365980310699e-05,
+      "loss": 1.2931,
+      "step": 6710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.803711369189319e-05,
+      "loss": 0.5074,
+      "step": 6711
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8010578094736995e-05,
+      "loss": 1.0151,
+      "step": 6712
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7984053015515387e-05,
+      "loss": 1.0711,
+      "step": 6713
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7957538458103792e-05,
+      "loss": 1.3841,
+      "step": 6714
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.793103442637609e-05,
+      "loss": 1.2864,
+      "step": 6715
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7904540924204536e-05,
+      "loss": 1.067,
+      "step": 6716
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.787805795546008e-05,
+      "loss": 1.3646,
+      "step": 6717
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7851585524011903e-05,
+      "loss": 0.5874,
+      "step": 6718
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.782512363372771e-05,
+      "loss": 1.4781,
+      "step": 6719
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7798672288473715e-05,
+      "loss": 0.8455,
+      "step": 6720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.77722314921146e-05,
+      "loss": 1.5072,
+      "step": 6721
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7745801248513392e-05,
+      "loss": 0.9655,
+      "step": 6722
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.77193815615317e-05,
+      "loss": 1.3282,
+      "step": 6723
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7692972435029563e-05,
+      "loss": 0.5873,
+      "step": 6724
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.766657387286541e-05,
+      "loss": 1.0712,
+      "step": 6725
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7640185878896208e-05,
+      "loss": 0.6094,
+      "step": 6726
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.761380845697731e-05,
+      "loss": 1.1875,
+      "step": 6727
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7587441610962573e-05,
+      "loss": 1.4165,
+      "step": 6728
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.756108534470434e-05,
+      "loss": 0.6576,
+      "step": 6729
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7534739662053287e-05,
+      "loss": 0.6597,
+      "step": 6730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.750840456685867e-05,
+      "loss": 1.4781,
+      "step": 6731
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7482080062968163e-05,
+      "loss": 1.325,
+      "step": 6732
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.745576615422787e-05,
+      "loss": 1.2703,
+      "step": 6733
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7429462844482302e-05,
+      "loss": 0.8266,
+      "step": 6734
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.740317013757451e-05,
+      "loss": 1.2639,
+      "step": 6735
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7376888037345994e-05,
+      "loss": 0.8341,
+      "step": 6736
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7350616547636622e-05,
+      "loss": 1.2302,
+      "step": 6737
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7324355672284764e-05,
+      "loss": 1.3477,
+      "step": 6738
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7298105415127296e-05,
+      "loss": 1.5893,
+      "step": 6739
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7271865779999427e-05,
+      "loss": 1.4922,
+      "step": 6740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7245636770734863e-05,
+      "loss": 1.4596,
+      "step": 6741
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7219418391165787e-05,
+      "loss": 1.001,
+      "step": 6742
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7193210645122813e-05,
+      "loss": 1.3581,
+      "step": 6743
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.716701353643497e-05,
+      "loss": 0.8308,
+      "step": 6744
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7140827068929762e-05,
+      "loss": 1.4332,
+      "step": 6745
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7114651246433186e-05,
+      "loss": 0.9267,
+      "step": 6746
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7088486072769547e-05,
+      "loss": 1.0199,
+      "step": 6747
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7062331551761756e-05,
+      "loss": 1.4196,
+      "step": 6748
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.703618768723102e-05,
+      "loss": 0.9123,
+      "step": 6749
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7010054482997104e-05,
+      "loss": 1.5016,
+      "step": 6750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6983931942878184e-05,
+      "loss": 1.1463,
+      "step": 6751
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6957820070690808e-05,
+      "loss": 1.1158,
+      "step": 6752
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6931718870250057e-05,
+      "loss": 1.4676,
+      "step": 6753
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.690562834536945e-05,
+      "loss": 1.2308,
+      "step": 6754
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6879548499860875e-05,
+      "loss": 0.812,
+      "step": 6755
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6853479337534683e-05,
+      "loss": 1.2433,
+      "step": 6756
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6827420862199694e-05,
+      "loss": 0.6563,
+      "step": 6757
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6801373077663194e-05,
+      "loss": 1.3962,
+      "step": 6758
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.67753359877308e-05,
+      "loss": 0.5717,
+      "step": 6759
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.674930959620666e-05,
+      "loss": 1.3869,
+      "step": 6760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6723293906893375e-05,
+      "loss": 1.2273,
+      "step": 6761
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6697288923591868e-05,
+      "loss": 1.4289,
+      "step": 6762
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.667129465010162e-05,
+      "loss": 1.4235,
+      "step": 6763
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6645311090220448e-05,
+      "loss": 1.295,
+      "step": 6764
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6619338247744674e-05,
+      "loss": 0.8002,
+      "step": 6765
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6593376126469073e-05,
+      "loss": 1.3147,
+      "step": 6766
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.656742473018674e-05,
+      "loss": 1.2745,
+      "step": 6767
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6541484062689338e-05,
+      "loss": 0.7571,
+      "step": 6768
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.651555412776684e-05,
+      "loss": 0.9675,
+      "step": 6769
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.648963492920776e-05,
+      "loss": 1.083,
+      "step": 6770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6463726470798967e-05,
+      "loss": 1.0615,
+      "step": 6771
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6437828756325777e-05,
+      "loss": 1.3336,
+      "step": 6772
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6411941789571993e-05,
+      "loss": 1.073,
+      "step": 6773
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6386065574319752e-05,
+      "loss": 1.0438,
+      "step": 6774
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6360200114349687e-05,
+      "loss": 0.5506,
+      "step": 6775
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6334345413440874e-05,
+      "loss": 1.0165,
+      "step": 6776
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.630850147537076e-05,
+      "loss": 1.0666,
+      "step": 6777
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.628266830391518e-05,
+      "loss": 1.1545,
+      "step": 6778
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6256845902848592e-05,
+      "loss": 1.2476,
+      "step": 6779
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6231034275943666e-05,
+      "loss": 0.8089,
+      "step": 6780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6205233426971577e-05,
+      "loss": 0.7455,
+      "step": 6781
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.617944335970194e-05,
+      "loss": 0.999,
+      "step": 6782
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6153664077902828e-05,
+      "loss": 0.9444,
+      "step": 6783
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.612789558534062e-05,
+      "loss": 0.9897,
+      "step": 6784
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6102137885780276e-05,
+      "loss": 1.5301,
+      "step": 6785
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.607639098298501e-05,
+      "loss": 0.8621,
+      "step": 6786
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6050654880716587e-05,
+      "loss": 0.8315,
+      "step": 6787
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.602492958273518e-05,
+      "loss": 1.2414,
+      "step": 6788
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.59992150927993e-05,
+      "loss": 1.3351,
+      "step": 6789
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5973511414665953e-05,
+      "loss": 0.5719,
+      "step": 6790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5947818552090585e-05,
+      "loss": 1.0865,
+      "step": 6791
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5922136508827e-05,
+      "loss": 1.3327,
+      "step": 6792
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5896465288627405e-05,
+      "loss": 0.8648,
+      "step": 6793
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5870804895242485e-05,
+      "loss": 1.1638,
+      "step": 6794
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5845155332421368e-05,
+      "loss": 0.7875,
+      "step": 6795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.58195166039115e-05,
+      "loss": 0.6517,
+      "step": 6796
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5793888713458814e-05,
+      "loss": 1.1411,
+      "step": 6797
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5768271664807676e-05,
+      "loss": 0.7025,
+      "step": 6798
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.574266546170081e-05,
+      "loss": 1.0639,
+      "step": 6799
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.571707010787934e-05,
+      "loss": 1.1145,
+      "step": 6800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5691485607082932e-05,
+      "loss": 1.2095,
+      "step": 6801
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.566591196304955e-05,
+      "loss": 1.3118,
+      "step": 6802
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5640349179515555e-05,
+      "loss": 1.3859,
+      "step": 6803
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5614797260215805e-05,
+      "loss": 1.0226,
+      "step": 6804
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.558925620888356e-05,
+      "loss": 1.4529,
+      "step": 6805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5563726029250424e-05,
+      "loss": 1.1171,
+      "step": 6806
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.55382067250465e-05,
+      "loss": 1.1417,
+      "step": 6807
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5512698300000203e-05,
+      "loss": 0.8439,
+      "step": 6808
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5487200757838435e-05,
+      "loss": 1.4736,
+      "step": 6809
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.546171410228654e-05,
+      "loss": 0.8973,
+      "step": 6810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5436238337068118e-05,
+      "loss": 1.1694,
+      "step": 6811
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5410773465905347e-05,
+      "loss": 1.5261,
+      "step": 6812
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5385319492518757e-05,
+      "loss": 0.6832,
+      "step": 6813
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5359876420627226e-05,
+      "loss": 1.1955,
+      "step": 6814
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5334444253948053e-05,
+      "loss": 0.886,
+      "step": 6815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5309022996197084e-05,
+      "loss": 0.3982,
+      "step": 6816
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5283612651088405e-05,
+      "loss": 0.4781,
+      "step": 6817
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5258213222334538e-05,
+      "loss": 1.0647,
+      "step": 6818
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5232824713646463e-05,
+      "loss": 0.6419,
+      "step": 6819
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5207447128733576e-05,
+      "loss": 1.1859,
+      "step": 6820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5182080471303615e-05,
+      "loss": 1.4257,
+      "step": 6821
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5156724745062664e-05,
+      "loss": 1.4861,
+      "step": 6822
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.513137995371545e-05,
+      "loss": 0.8877,
+      "step": 6823
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5106046100964852e-05,
+      "loss": 0.9685,
+      "step": 6824
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5080723190512233e-05,
+      "loss": 0.9954,
+      "step": 6825
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.505541122605739e-05,
+      "loss": 0.5842,
+      "step": 6826
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.503011021129853e-05,
+      "loss": 1.2783,
+      "step": 6827
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5004820149932183e-05,
+      "loss": 0.9815,
+      "step": 6828
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4979541045653344e-05,
+      "loss": 0.7,
+      "step": 6829
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4954272902155408e-05,
+      "loss": 1.1304,
+      "step": 6830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4929015723130112e-05,
+      "loss": 1.0349,
+      "step": 6831
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4903769512267673e-05,
+      "loss": 1.3707,
+      "step": 6832
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4878534273256605e-05,
+      "loss": 1.3805,
+      "step": 6833
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4853310009783903e-05,
+      "loss": 1.2965,
+      "step": 6834
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4828096725534968e-05,
+      "loss": 1.1561,
+      "step": 6835
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4802894424193513e-05,
+      "loss": 1.4075,
+      "step": 6836
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.477770310944164e-05,
+      "loss": 1.0599,
+      "step": 6837
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.475252278496003e-05,
+      "loss": 1.0034,
+      "step": 6838
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.472735345442755e-05,
+      "loss": 0.9542,
+      "step": 6839
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.470219512152151e-05,
+      "loss": 0.9598,
+      "step": 6840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4677047789917674e-05,
+      "loss": 0.747,
+      "step": 6841
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.46519114632902e-05,
+      "loss": 0.6705,
+      "step": 6842
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4626786145311566e-05,
+      "loss": 1.2393,
+      "step": 6843
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4601671839652606e-05,
+      "loss": 1.29,
+      "step": 6844
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.457656854998277e-05,
+      "loss": 1.2816,
+      "step": 6845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4551476279969656e-05,
+      "loss": 1.2144,
+      "step": 6846
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4526395033279327e-05,
+      "loss": 1.2118,
+      "step": 6847
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4501324813576275e-05,
+      "loss": 1.2714,
+      "step": 6848
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4476265624523398e-05,
+      "loss": 1.198,
+      "step": 6849
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4451217469781874e-05,
+      "loss": 0.6903,
+      "step": 6850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.442618035301135e-05,
+      "loss": 0.9133,
+      "step": 6851
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4401154277869897e-05,
+      "loss": 0.7567,
+      "step": 6852
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.437613924801386e-05,
+      "loss": 0.8055,
+      "step": 6853
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.435113526709809e-05,
+      "loss": 1.1187,
+      "step": 6854
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.432614233877569e-05,
+      "loss": 0.9452,
+      "step": 6855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4301160466698282e-05,
+      "loss": 1.2978,
+      "step": 6856
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4276189654515822e-05,
+      "loss": 0.8142,
+      "step": 6857
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4251229905876592e-05,
+      "loss": 1.4487,
+      "step": 6858
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.422628122442734e-05,
+      "loss": 0.5283,
+      "step": 6859
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.420134361381319e-05,
+      "loss": 0.521,
+      "step": 6860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4176417077677605e-05,
+      "loss": 1.4681,
+      "step": 6861
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.41515016196624e-05,
+      "loss": 0.9285,
+      "step": 6862
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4126597243407866e-05,
+      "loss": 1.0179,
+      "step": 6863
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4101703952552644e-05,
+      "loss": 0.9274,
+      "step": 6864
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.40768217507337e-05,
+      "loss": 1.5933,
+      "step": 6865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.405195064158645e-05,
+      "loss": 1.2591,
+      "step": 6866
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4027090628744664e-05,
+      "loss": 1.5567,
+      "step": 6867
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.400224171584048e-05,
+      "loss": 0.4816,
+      "step": 6868
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.397740390650439e-05,
+      "loss": 1.2506,
+      "step": 6869
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.395257720436531e-05,
+      "loss": 1.3951,
+      "step": 6870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.392776161305057e-05,
+      "loss": 1.2201,
+      "step": 6871
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3902957136185744e-05,
+      "loss": 1.1989,
+      "step": 6872
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3878163777394892e-05,
+      "loss": 1.2082,
+      "step": 6873
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.385338154030047e-05,
+      "loss": 0.5461,
+      "step": 6874
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.382861042852318e-05,
+      "loss": 1.3009,
+      "step": 6875
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3803850445682253e-05,
+      "loss": 0.6481,
+      "step": 6876
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3779101595395136e-05,
+      "loss": 1.2171,
+      "step": 6877
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3754363881277784e-05,
+      "loss": 0.9786,
+      "step": 6878
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3729637306944508e-05,
+      "loss": 1.0661,
+      "step": 6879
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.370492187600787e-05,
+      "loss": 0.9566,
+      "step": 6880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3680217592078944e-05,
+      "loss": 0.9747,
+      "step": 6881
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3655524458767142e-05,
+      "loss": 1.2531,
+      "step": 6882
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.363084247968019e-05,
+      "loss": 0.7645,
+      "step": 6883
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3606171658424214e-05,
+      "loss": 0.9436,
+      "step": 6884
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.358151199860372e-05,
+      "loss": 0.4714,
+      "step": 6885
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3556863503821635e-05,
+      "loss": 1.2065,
+      "step": 6886
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.353222617767912e-05,
+      "loss": 0.5329,
+      "step": 6887
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3507600023775824e-05,
+      "loss": 1.193,
+      "step": 6888
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3482985045709748e-05,
+      "loss": 0.7455,
+      "step": 6889
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3458381247077187e-05,
+      "loss": 1.2542,
+      "step": 6890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3433788631472898e-05,
+      "loss": 1.0084,
+      "step": 6891
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3409207202489903e-05,
+      "loss": 0.885,
+      "step": 6892
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.338463696371971e-05,
+      "loss": 0.9856,
+      "step": 6893
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3360077918752064e-05,
+      "loss": 1.2815,
+      "step": 6894
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.333553007117516e-05,
+      "loss": 1.1479,
+      "step": 6895
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.331099342457558e-05,
+      "loss": 1.1446,
+      "step": 6896
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3286467982538142e-05,
+      "loss": 1.2961,
+      "step": 6897
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.326195374864618e-05,
+      "loss": 1.2878,
+      "step": 6898
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.323745072648127e-05,
+      "loss": 1.0851,
+      "step": 6899
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3212958919623408e-05,
+      "loss": 1.7515,
+      "step": 6900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3188478331650976e-05,
+      "loss": 1.2827,
+      "step": 6901
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.316400896614064e-05,
+      "loss": 1.1829,
+      "step": 6902
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3139550826667478e-05,
+      "loss": 1.1945,
+      "step": 6903
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.311510391680496e-05,
+      "loss": 1.2055,
+      "step": 6904
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3090668240124847e-05,
+      "loss": 1.3551,
+      "step": 6905
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.306624380019725e-05,
+      "loss": 1.3367,
+      "step": 6906
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.304183060059072e-05,
+      "loss": 0.9359,
+      "step": 6907
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.301742864487213e-05,
+      "loss": 0.7859,
+      "step": 6908
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2993037936606644e-05,
+      "loss": 0.711,
+      "step": 6909
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2968658479357884e-05,
+      "loss": 0.7783,
+      "step": 6910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2944290276687797e-05,
+      "loss": 0.3407,
+      "step": 6911
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.291993333215663e-05,
+      "loss": 0.6231,
+      "step": 6912
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.289558764932306e-05,
+      "loss": 1.6859,
+      "step": 6913
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.287125323174405e-05,
+      "loss": 1.0097,
+      "step": 6914
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2846930082974983e-05,
+      "loss": 1.4975,
+      "step": 6915
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.282261820656958e-05,
+      "loss": 0.8347,
+      "step": 6916
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2798317606079854e-05,
+      "loss": 0.819,
+      "step": 6917
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.277402828505626e-05,
+      "loss": 1.2081,
+      "step": 6918
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.274975024704753e-05,
+      "loss": 1.1341,
+      "step": 6919
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.272548349560082e-05,
+      "loss": 1.2658,
+      "step": 6920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2701228034261536e-05,
+      "loss": 1.3872,
+      "step": 6921
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.267698386657353e-05,
+      "loss": 0.6668,
+      "step": 6922
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2652750996079007e-05,
+      "loss": 1.4304,
+      "step": 6923
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2628529426318412e-05,
+      "loss": 0.6406,
+      "step": 6924
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2604319160830643e-05,
+      "loss": 1.331,
+      "step": 6925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2580120203152943e-05,
+      "loss": 1.2539,
+      "step": 6926
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2555932556820858e-05,
+      "loss": 0.7865,
+      "step": 6927
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2531756225368216e-05,
+      "loss": 1.3353,
+      "step": 6928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2507591212327406e-05,
+      "loss": 1.2606,
+      "step": 6929
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2483437521228968e-05,
+      "loss": 1.5636,
+      "step": 6930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2459295155601823e-05,
+      "loss": 0.9909,
+      "step": 6931
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2435164118973285e-05,
+      "loss": 0.8455,
+      "step": 6932
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2411044414869033e-05,
+      "loss": 0.702,
+      "step": 6933
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.238693604681298e-05,
+      "loss": 1.1764,
+      "step": 6934
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2362839018327518e-05,
+      "loss": 0.9464,
+      "step": 6935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.233875333293326e-05,
+      "loss": 1.1321,
+      "step": 6936
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.231467899414924e-05,
+      "loss": 1.176,
+      "step": 6937
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.229061600549285e-05,
+      "loss": 1.1004,
+      "step": 6938
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2266564370479725e-05,
+      "loss": 1.1041,
+      "step": 6939
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.224252409262393e-05,
+      "loss": 1.2685,
+      "step": 6940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2218495175437858e-05,
+      "loss": 0.6716,
+      "step": 6941
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.219447762243223e-05,
+      "loss": 1.3045,
+      "step": 6942
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.217047143711606e-05,
+      "loss": 0.6431,
+      "step": 6943
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2146476622996758e-05,
+      "loss": 1.0901,
+      "step": 6944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2122493183580105e-05,
+      "loss": 1.179,
+      "step": 6945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.209852112237012e-05,
+      "loss": 0.9506,
+      "step": 6946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.207456044286924e-05,
+      "loss": 1.1221,
+      "step": 6947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2050611148578238e-05,
+      "loss": 1.2316,
+      "step": 6948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2026673242996165e-05,
+      "loss": 1.0925,
+      "step": 6949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2002746729620406e-05,
+      "loss": 0.6954,
+      "step": 6950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1978831611946805e-05,
+      "loss": 1.214,
+      "step": 6951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1954927893469414e-05,
+      "loss": 1.2205,
+      "step": 6952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1931035577680627e-05,
+      "loss": 1.1421,
+      "step": 6953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.190715466807124e-05,
+      "loss": 0.9828,
+      "step": 6954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.188328516813035e-05,
+      "loss": 1.066,
+      "step": 6955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.185942708134536e-05,
+      "loss": 1.3536,
+      "step": 6956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.183558041120206e-05,
+      "loss": 0.5235,
+      "step": 6957
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1811745161184505e-05,
+      "loss": 1.0939,
+      "step": 6958
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1787921334775128e-05,
+      "loss": 1.0091,
+      "step": 6959
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.176410893545473e-05,
+      "loss": 1.1241,
+      "step": 6960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1740307966702313e-05,
+      "loss": 0.9225,
+      "step": 6961
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.171651843199535e-05,
+      "loss": 1.1206,
+      "step": 6962
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.169274033480958e-05,
+      "loss": 1.7649,
+      "step": 6963
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1668973678619075e-05,
+      "loss": 1.1731,
+      "step": 6964
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1645218466896168e-05,
+      "loss": 0.577,
+      "step": 6965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.16214747031117e-05,
+      "loss": 1.2165,
+      "step": 6966
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1597742390734675e-05,
+      "loss": 0.974,
+      "step": 6967
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1574021533232446e-05,
+      "loss": 1.1136,
+      "step": 6968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.155031213407075e-05,
+      "loss": 1.4233,
+      "step": 6969
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1526614196713647e-05,
+      "loss": 0.8921,
+      "step": 6970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1502927724623467e-05,
+      "loss": 1.0548,
+      "step": 6971
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1479252721260855e-05,
+      "loss": 1.0528,
+      "step": 6972
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1455589190084925e-05,
+      "loss": 1.1385,
+      "step": 6973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1431937134552947e-05,
+      "loss": 1.0024,
+      "step": 6974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1408296558120565e-05,
+      "loss": 1.4213,
+      "step": 6975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.138466746424177e-05,
+      "loss": 0.7495,
+      "step": 6976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.136104985636891e-05,
+      "loss": 0.8327,
+      "step": 6977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1337443737952543e-05,
+      "loss": 1.5747,
+      "step": 6978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.131384911244165e-05,
+      "loss": 1.1327,
+      "step": 6979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.129026598328352e-05,
+      "loss": 1.4094,
+      "step": 6980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.126669435392368e-05,
+      "loss": 1.151,
+      "step": 6981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1243134227806096e-05,
+      "loss": 1.3816,
+      "step": 6982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.121958560837294e-05,
+      "loss": 1.1104,
+      "step": 6983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1196048499064792e-05,
+      "loss": 0.7266,
+      "step": 6984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1172522903320524e-05,
+      "loss": 0.8839,
+      "step": 6985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.114900882457731e-05,
+      "loss": 0.9816,
+      "step": 6986
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.112550626627058e-05,
+      "loss": 1.1423,
+      "step": 6987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.110201523183426e-05,
+      "loss": 0.5635,
+      "step": 6988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.107853572470043e-05,
+      "loss": 0.8685,
+      "step": 6989
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1055067748299518e-05,
+      "loss": 1.3224,
+      "step": 6990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1031611306060306e-05,
+      "loss": 0.9924,
+      "step": 6991
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.100816640140989e-05,
+      "loss": 1.3394,
+      "step": 6992
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.098473303777365e-05,
+      "loss": 1.2005,
+      "step": 6993
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.096131121857523e-05,
+      "loss": 0.7784,
+      "step": 6994
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0937900947236753e-05,
+      "loss": 0.9768,
+      "step": 6995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0914502227178502e-05,
+      "loss": 0.704,
+      "step": 6996
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.089111506181909e-05,
+      "loss": 1.0941,
+      "step": 6997
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.086773945457551e-05,
+      "loss": 0.6223,
+      "step": 6998
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.084437540886305e-05,
+      "loss": 1.3342,
+      "step": 6999
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0821022928095223e-05,
+      "loss": 0.5901,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.079768201568396e-05,
+      "loss": 1.2712,
+      "step": 7001
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0774352675039476e-05,
+      "loss": 0.5133,
+      "step": 7002
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0751034909570245e-05,
+      "loss": 0.9812,
+      "step": 7003
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0727728722683105e-05,
+      "loss": 1.3886,
+      "step": 7004
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0704434117783144e-05,
+      "loss": 1.1582,
+      "step": 7005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0681151098273834e-05,
+      "loss": 0.7926,
+      "step": 7006
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.065787966755691e-05,
+      "loss": 1.058,
+      "step": 7007
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.063461982903242e-05,
+      "loss": 0.5835,
+      "step": 7008
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.061137158609865e-05,
+      "loss": 0.9321,
+      "step": 7009
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0588134942152372e-05,
+      "loss": 1.4668,
+      "step": 7010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.056490990058848e-05,
+      "loss": 0.543,
+      "step": 7011
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0541696464800242e-05,
+      "loss": 1.489,
+      "step": 7012
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.051849463817924e-05,
+      "loss": 1.0503,
+      "step": 7013
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.049530442411539e-05,
+      "loss": 1.4798,
+      "step": 7014
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.04721258259968e-05,
+      "loss": 1.3594,
+      "step": 7015
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.044895884721001e-05,
+      "loss": 1.3106,
+      "step": 7016
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0425803491139805e-05,
+      "loss": 0.985,
+      "step": 7017
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0402659761169253e-05,
+      "loss": 0.9248,
+      "step": 7018
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.037952766067972e-05,
+      "loss": 1.2356,
+      "step": 7019
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.035640719305092e-05,
+      "loss": 1.2586,
+      "step": 7020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0333298361660868e-05,
+      "loss": 1.2986,
+      "step": 7021
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.031020116988581e-05,
+      "loss": 0.9091,
+      "step": 7022
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.028711562110036e-05,
+      "loss": 0.6002,
+      "step": 7023
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.026404171867743e-05,
+      "loss": 0.7489,
+      "step": 7024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0240979465988143e-05,
+      "loss": 1.414,
+      "step": 7025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0217928866402058e-05,
+      "loss": 1.052,
+      "step": 7026
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0194889923286887e-05,
+      "loss": 1.2336,
+      "step": 7027
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0171862640008742e-05,
+      "loss": 0.9574,
+      "step": 7028
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0148847019932028e-05,
+      "loss": 1.2126,
+      "step": 7029
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0125843066419357e-05,
+      "loss": 1.5886,
+      "step": 7030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.010285078283173e-05,
+      "loss": 1.3769,
+      "step": 7031
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0079870172528436e-05,
+      "loss": 1.1429,
+      "step": 7032
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0056901238866988e-05,
+      "loss": 1.1437,
+      "step": 7033
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.003394398520324e-05,
+      "loss": 1.686,
+      "step": 7034
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0010998414891337e-05,
+      "loss": 0.9801,
+      "step": 7035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9988064531283744e-05,
+      "loss": 1.5692,
+      "step": 7036
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9965142337731157e-05,
+      "loss": 1.0982,
+      "step": 7037
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9942231837582602e-05,
+      "loss": 1.0477,
+      "step": 7038
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9919333034185438e-05,
+      "loss": 1.2118,
+      "step": 7039
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.98964459308852e-05,
+      "loss": 0.8494,
+      "step": 7040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9873570531025853e-05,
+      "loss": 1.1325,
+      "step": 7041
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9850706837949507e-05,
+      "loss": 1.1988,
+      "step": 7042
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9827854854996708e-05,
+      "loss": 0.5773,
+      "step": 7043
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9805014585506154e-05,
+      "loss": 1.4972,
+      "step": 7044
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.978218603281493e-05,
+      "loss": 1.2524,
+      "step": 7045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9759369200258414e-05,
+      "loss": 1.2709,
+      "step": 7046
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9736564091170162e-05,
+      "loss": 1.1361,
+      "step": 7047
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.971377070888214e-05,
+      "loss": 1.0617,
+      "step": 7048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9690989056724517e-05,
+      "loss": 0.6482,
+      "step": 7049
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9668219138025802e-05,
+      "loss": 0.733,
+      "step": 7050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9645460956112784e-05,
+      "loss": 1.2887,
+      "step": 7051
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.962271451431047e-05,
+      "loss": 1.1805,
+      "step": 7052
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9599979815942238e-05,
+      "loss": 0.9937,
+      "step": 7053
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.957725686432974e-05,
+      "loss": 0.4895,
+      "step": 7054
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9554545662792866e-05,
+      "loss": 0.7029,
+      "step": 7055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9531846214649774e-05,
+      "loss": 1.1259,
+      "step": 7056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9509158523216976e-05,
+      "loss": 1.3691,
+      "step": 7057
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9486482591809253e-05,
+      "loss": 0.9103,
+      "step": 7058
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.946381842373959e-05,
+      "loss": 1.5827,
+      "step": 7059
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9441166022319345e-05,
+      "loss": 0.8469,
+      "step": 7060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9418525390858155e-05,
+      "loss": 1.1395,
+      "step": 7061
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.939589653266384e-05,
+      "loss": 1.1597,
+      "step": 7062
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9373279451042614e-05,
+      "loss": 0.737,
+      "step": 7063
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9350674149298888e-05,
+      "loss": 0.4623,
+      "step": 7064
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.932808063073538e-05,
+      "loss": 0.8725,
+      "step": 7065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9305498898653142e-05,
+      "loss": 0.8601,
+      "step": 7066
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9282928956351397e-05,
+      "loss": 1.3841,
+      "step": 7067
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9260370807127735e-05,
+      "loss": 0.9238,
+      "step": 7068
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9237824454277966e-05,
+      "loss": 0.8423,
+      "step": 7069
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9215289901096233e-05,
+      "loss": 1.2008,
+      "step": 7070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9192767150874867e-05,
+      "loss": 0.6489,
+      "step": 7071
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.917025620690457e-05,
+      "loss": 0.6385,
+      "step": 7072
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9147757072474293e-05,
+      "loss": 1.1242,
+      "step": 7073
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.91252697508712e-05,
+      "loss": 0.6784,
+      "step": 7074
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9102794245380806e-05,
+      "loss": 1.0252,
+      "step": 7075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9080330559286886e-05,
+      "loss": 0.397,
+      "step": 7076
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9057878695871447e-05,
+      "loss": 1.5412,
+      "step": 7077
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9035438658414785e-05,
+      "loss": 1.1103,
+      "step": 7078
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9013010450195467e-05,
+      "loss": 1.2242,
+      "step": 7079
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.899059407449041e-05,
+      "loss": 0.6774,
+      "step": 7080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8968189534574655e-05,
+      "loss": 1.2688,
+      "step": 7081
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8945796833721618e-05,
+      "loss": 0.7142,
+      "step": 7082
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8923415975202995e-05,
+      "loss": 0.9233,
+      "step": 7083
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.890104696228867e-05,
+      "loss": 1.0857,
+      "step": 7084
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8878689798246885e-05,
+      "loss": 1.5288,
+      "step": 7085
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8856344486344056e-05,
+      "loss": 1.4647,
+      "step": 7086
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8834011029844945e-05,
+      "loss": 1.1256,
+      "step": 7087
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.881168943201259e-05,
+      "loss": 1.1819,
+      "step": 7088
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.87893796961082e-05,
+      "loss": 1.2658,
+      "step": 7089
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8767081825391364e-05,
+      "loss": 0.5854,
+      "step": 7090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8744795823119886e-05,
+      "loss": 1.4076,
+      "step": 7091
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.872252169254982e-05,
+      "loss": 0.6789,
+      "step": 7092
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8700259436935487e-05,
+      "loss": 0.8848,
+      "step": 7093
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.86780090595295e-05,
+      "loss": 0.6132,
+      "step": 7094
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8655770563582777e-05,
+      "loss": 0.9598,
+      "step": 7095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.863354395234437e-05,
+      "loss": 0.6153,
+      "step": 7096
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8611329229061702e-05,
+      "loss": 1.4206,
+      "step": 7097
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8589126396980472e-05,
+      "loss": 1.1929,
+      "step": 7098
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.856693545934456e-05,
+      "loss": 1.3552,
+      "step": 7099
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8544756419396104e-05,
+      "loss": 1.0502,
+      "step": 7100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.852258928037566e-05,
+      "loss": 1.4145,
+      "step": 7101
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8500434045521854e-05,
+      "loss": 0.7494,
+      "step": 7102
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8478290718071657e-05,
+      "loss": 1.1617,
+      "step": 7103
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.84561593012603e-05,
+      "loss": 1.1567,
+      "step": 7104
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8434039798321302e-05,
+      "loss": 1.1488,
+      "step": 7105
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.841193221248636e-05,
+      "loss": 1.1446,
+      "step": 7106
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.838983654698553e-05,
+      "loss": 1.271,
+      "step": 7107
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8367752805047024e-05,
+      "loss": 1.0037,
+      "step": 7108
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8345680989897375e-05,
+      "loss": 1.3566,
+      "step": 7109
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.83236211047614e-05,
+      "loss": 1.3342,
+      "step": 7110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.830157315286207e-05,
+      "loss": 1.3315,
+      "step": 7111
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8279537137420723e-05,
+      "loss": 0.6201,
+      "step": 7112
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8257513061656916e-05,
+      "loss": 1.4813,
+      "step": 7113
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.823550092878844e-05,
+      "loss": 1.5541,
+      "step": 7114
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8213500742031285e-05,
+      "loss": 0.7664,
+      "step": 7115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.819151250459987e-05,
+      "loss": 1.2261,
+      "step": 7116
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8169536219706727e-05,
+      "loss": 1.0652,
+      "step": 7117
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.814757189056263e-05,
+      "loss": 1.2133,
+      "step": 7118
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.812561952037669e-05,
+      "loss": 1.2359,
+      "step": 7119
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8103679112356264e-05,
+      "loss": 1.4271,
+      "step": 7120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8081750669706888e-05,
+      "loss": 1.451,
+      "step": 7121
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.805983419563235e-05,
+      "loss": 0.5348,
+      "step": 7122
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8037929693334843e-05,
+      "loss": 1.1391,
+      "step": 7123
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8016037166014642e-05,
+      "loss": 1.107,
+      "step": 7124
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7994156616870305e-05,
+      "loss": 0.4952,
+      "step": 7125
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7972288049098696e-05,
+      "loss": 1.0425,
+      "step": 7126
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7950431465894913e-05,
+      "loss": 1.41,
+      "step": 7127
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7928586870452248e-05,
+      "loss": 0.9055,
+      "step": 7128
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7906754265962288e-05,
+      "loss": 1.5305,
+      "step": 7129
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.788493365561491e-05,
+      "loss": 0.9093,
+      "step": 7130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7863125042598127e-05,
+      "loss": 1.049,
+      "step": 7131
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7841328430098314e-05,
+      "loss": 1.3307,
+      "step": 7132
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.781954382129999e-05,
+      "loss": 1.1678,
+      "step": 7133
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7797771219386007e-05,
+      "loss": 0.8757,
+      "step": 7134
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.777601062753743e-05,
+      "loss": 0.5084,
+      "step": 7135
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7754262048933567e-05,
+      "loss": 1.2807,
+      "step": 7136
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7732525486751893e-05,
+      "loss": 1.4406,
+      "step": 7137
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7710800944168327e-05,
+      "loss": 0.5693,
+      "step": 7138
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.768908842435685e-05,
+      "loss": 1.2223,
+      "step": 7139
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.766738793048971e-05,
+      "loss": 1.3108,
+      "step": 7140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7645699465737475e-05,
+      "loss": 1.3054,
+      "step": 7141
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7624023033268942e-05,
+      "loss": 1.0579,
+      "step": 7142
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.760235863625106e-05,
+      "loss": 1.1281,
+      "step": 7143
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7580706277849103e-05,
+      "loss": 1.0701,
+      "step": 7144
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7559065961226606e-05,
+      "loss": 0.5382,
+      "step": 7145
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7537437689545254e-05,
+      "loss": 1.7189,
+      "step": 7146
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7515821465965032e-05,
+      "loss": 0.7622,
+      "step": 7147
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7494217293644143e-05,
+      "loss": 1.5364,
+      "step": 7148
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7472625175739087e-05,
+      "loss": 0.9231,
+      "step": 7149
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7451045115404508e-05,
+      "loss": 1.048,
+      "step": 7150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7429477115793348e-05,
+      "loss": 1.0464,
+      "step": 7151
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7407921180056806e-05,
+      "loss": 1.4188,
+      "step": 7152
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.738637731134425e-05,
+      "loss": 1.318,
+      "step": 7153
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.736484551280335e-05,
+      "loss": 1.1262,
+      "step": 7154
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7343325787579968e-05,
+      "loss": 0.679,
+      "step": 7155
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7321818138818204e-05,
+      "loss": 1.4295,
+      "step": 7156
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7300322569660467e-05,
+      "loss": 1.0292,
+      "step": 7157
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.72788390832473e-05,
+      "loss": 1.2255,
+      "step": 7158
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7257367682717485e-05,
+      "loss": 1.2141,
+      "step": 7159
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7235908371208165e-05,
+      "loss": 1.2153,
+      "step": 7160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.72144611518546e-05,
+      "loss": 0.613,
+      "step": 7161
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.719302602779026e-05,
+      "loss": 1.1679,
+      "step": 7162
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7171603002146942e-05,
+      "loss": 1.0117,
+      "step": 7163
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.715019207805466e-05,
+      "loss": 1.7054,
+      "step": 7164
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7128793258641573e-05,
+      "loss": 0.6674,
+      "step": 7165
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7107406547034167e-05,
+      "loss": 1.2607,
+      "step": 7166
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708603194635714e-05,
+      "loss": 1.1867,
+      "step": 7167
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706466945973336e-05,
+      "loss": 1.2943,
+      "step": 7168
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7043319090284017e-05,
+      "loss": 1.022,
+      "step": 7169
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7021980841128426e-05,
+      "loss": 0.907,
+      "step": 7170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7000654715384256e-05,
+      "loss": 1.4036,
+      "step": 7171
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.697934071616728e-05,
+      "loss": 0.6982,
+      "step": 7172
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.695803884659156e-05,
+      "loss": 0.6602,
+      "step": 7173
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6936749109769447e-05,
+      "loss": 1.6438,
+      "step": 7174
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6915471508811365e-05,
+      "loss": 1.5025,
+      "step": 7175
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.689420604682612e-05,
+      "loss": 1.1516,
+      "step": 7176
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6872952726920644e-05,
+      "loss": 1.6023,
+      "step": 7177
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6851711552200135e-05,
+      "loss": 0.5032,
+      "step": 7178
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6830482525768043e-05,
+      "loss": 0.7687,
+      "step": 7179
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.680926565072596e-05,
+      "loss": 1.1535,
+      "step": 7180
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6788060930173778e-05,
+      "loss": 0.9617,
+      "step": 7181
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.676686836720962e-05,
+      "loss": 1.2747,
+      "step": 7182
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6745687964929768e-05,
+      "loss": 1.1447,
+      "step": 7183
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6724519726428735e-05,
+      "loss": 1.1702,
+      "step": 7184
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.670336365479932e-05,
+      "loss": 1.019,
+      "step": 7185
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6682219753132535e-05,
+      "loss": 1.3386,
+      "step": 7186
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6661088024517514e-05,
+      "loss": 1.2937,
+      "step": 7187
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6639968472041723e-05,
+      "loss": 0.6453,
+      "step": 7188
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.661886109879084e-05,
+      "loss": 1.1976,
+      "step": 7189
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6597765907848683e-05,
+      "loss": 0.9337,
+      "step": 7190
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6576682902297392e-05,
+      "loss": 1.4129,
+      "step": 7191
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.655561208521721e-05,
+      "loss": 0.7135,
+      "step": 7192
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.653455345968671e-05,
+      "loss": 1.279,
+      "step": 7193
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.651350702878266e-05,
+      "loss": 1.4149,
+      "step": 7194
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6492472795579973e-05,
+      "loss": 0.6168,
+      "step": 7195
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6471450763151874e-05,
+      "loss": 0.72,
+      "step": 7196
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6450440934569734e-05,
+      "loss": 1.0953,
+      "step": 7197
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6429443312903193e-05,
+      "loss": 0.9091,
+      "step": 7198
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6408457901220064e-05,
+      "loss": 1.0652,
+      "step": 7199
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6387484702586398e-05,
+      "loss": 1.5717,
+      "step": 7200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6366523720066506e-05,
+      "loss": 1.0381,
+      "step": 7201
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.634557495672281e-05,
+      "loss": 1.1255,
+      "step": 7202
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.632463841561602e-05,
+      "loss": 0.6066,
+      "step": 7203
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.630371409980509e-05,
+      "loss": 0.5536,
+      "step": 7204
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.628280201234711e-05,
+      "loss": 0.7571,
+      "step": 7205
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6261902156297392e-05,
+      "loss": 1.2583,
+      "step": 7206
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6241014534709508e-05,
+      "loss": 1.2451,
+      "step": 7207
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.622013915063526e-05,
+      "loss": 1.4448,
+      "step": 7208
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6199276007124554e-05,
+      "loss": 0.9524,
+      "step": 7209
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6178425107225615e-05,
+      "loss": 1.3336,
+      "step": 7210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6157586453984853e-05,
+      "loss": 1.1666,
+      "step": 7211
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.613676005044684e-05,
+      "loss": 0.7832,
+      "step": 7212
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6115945899654438e-05,
+      "loss": 1.2072,
+      "step": 7213
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6095144004648634e-05,
+      "loss": 0.988,
+      "step": 7214
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.607435436846867e-05,
+      "loss": 1.3836,
+      "step": 7215
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6053576994152043e-05,
+      "loss": 0.5841,
+      "step": 7216
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6032811884734333e-05,
+      "loss": 1.2143,
+      "step": 7217
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.601205904324945e-05,
+      "loss": 0.7905,
+      "step": 7218
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5991318472729477e-05,
+      "loss": 1.3251,
+      "step": 7219
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.597059017620468e-05,
+      "loss": 0.8752,
+      "step": 7220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5949874156703504e-05,
+      "loss": 0.9804,
+      "step": 7221
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.592917041725267e-05,
+      "loss": 1.252,
+      "step": 7222
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5908478960877094e-05,
+      "loss": 1.4775,
+      "step": 7223
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5887799790599846e-05,
+      "loss": 0.6028,
+      "step": 7224
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.586713290944224e-05,
+      "loss": 0.6348,
+      "step": 7225
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5846478320423817e-05,
+      "loss": 0.8499,
+      "step": 7226
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.582583602656228e-05,
+      "loss": 1.6009,
+      "step": 7227
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.580520603087351e-05,
+      "loss": 1.1437,
+      "step": 7228
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5784588336371652e-05,
+      "loss": 0.6122,
+      "step": 7229
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5763982946069077e-05,
+      "loss": 1.0222,
+      "step": 7230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5743389862976244e-05,
+      "loss": 0.5907,
+      "step": 7231
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.572280909010191e-05,
+      "loss": 0.9265,
+      "step": 7232
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5702240630453026e-05,
+      "loss": 0.9608,
+      "step": 7233
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5681684487034687e-05,
+      "loss": 1.2083,
+      "step": 7234
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5661140662850272e-05,
+      "loss": 1.3858,
+      "step": 7235
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5640609160901266e-05,
+      "loss": 0.9668,
+      "step": 7236
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5620089984187415e-05,
+      "loss": 0.6217,
+      "step": 7237
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5599583135706685e-05,
+      "loss": 0.8454,
+      "step": 7238
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5579088618455152e-05,
+      "loss": 1.3998,
+      "step": 7239
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.555860643542717e-05,
+      "loss": 1.5664,
+      "step": 7240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5538136589615292e-05,
+      "loss": 1.2028,
+      "step": 7241
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5517679084010217e-05,
+      "loss": 0.6381,
+      "step": 7242
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5497233921600817e-05,
+      "loss": 0.6767,
+      "step": 7243
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5476801105374317e-05,
+      "loss": 0.609,
+      "step": 7244
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.545638063831596e-05,
+      "loss": 1.209,
+      "step": 7245
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5435972523409247e-05,
+      "loss": 1.103,
+      "step": 7246
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.541557676363591e-05,
+      "loss": 1.3522,
+      "step": 7247
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.539519336197587e-05,
+      "loss": 1.2996,
+      "step": 7248
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5374822321407188e-05,
+      "loss": 1.0529,
+      "step": 7249
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5354463644906114e-05,
+      "loss": 1.4034,
+      "step": 7250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.533411733544722e-05,
+      "loss": 1.0851,
+      "step": 7251
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5313783396003135e-05,
+      "loss": 0.6773,
+      "step": 7252
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5293461829544697e-05,
+      "loss": 0.9246,
+      "step": 7253
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5273152639041012e-05,
+      "loss": 1.0946,
+      "step": 7254
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5252855827459333e-05,
+      "loss": 0.5946,
+      "step": 7255
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.523257139776506e-05,
+      "loss": 1.2453,
+      "step": 7256
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5212299352921877e-05,
+      "loss": 0.8757,
+      "step": 7257
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5192039695891557e-05,
+      "loss": 0.9834,
+      "step": 7258
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5171792429634146e-05,
+      "loss": 1.2228,
+      "step": 7259
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5151557557107876e-05,
+      "loss": 1.3154,
+      "step": 7260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5131335081269082e-05,
+      "loss": 0.9787,
+      "step": 7261
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5111125005072369e-05,
+      "loss": 1.0964,
+      "step": 7262
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5090927331470539e-05,
+      "loss": 1.2277,
+      "step": 7263
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5070742063414522e-05,
+      "loss": 0.7722,
+      "step": 7264
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5050569203853427e-05,
+      "loss": 0.7126,
+      "step": 7265
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.503040875573467e-05,
+      "loss": 1.0132,
+      "step": 7266
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5010260722003722e-05,
+      "loss": 1.1874,
+      "step": 7267
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.499012510560428e-05,
+      "loss": 0.6725,
+      "step": 7268
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4970001909478249e-05,
+      "loss": 1.1036,
+      "step": 7269
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4949891136565719e-05,
+      "loss": 0.5519,
+      "step": 7270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4929792789804952e-05,
+      "loss": 1.1314,
+      "step": 7271
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4909706872132335e-05,
+      "loss": 1.2738,
+      "step": 7272
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4889633386482593e-05,
+      "loss": 1.0038,
+      "step": 7273
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4869572335788495e-05,
+      "loss": 0.7204,
+      "step": 7274
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4849523722981018e-05,
+      "loss": 0.9044,
+      "step": 7275
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4829487550989362e-05,
+      "loss": 1.5627,
+      "step": 7276
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.480946382274091e-05,
+      "loss": 0.7289,
+      "step": 7277
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4789452541161174e-05,
+      "loss": 0.9604,
+      "step": 7278
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4769453709173886e-05,
+      "loss": 0.6221,
+      "step": 7279
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4749467329700973e-05,
+      "loss": 1.158,
+      "step": 7280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4729493405662497e-05,
+      "loss": 1.0323,
+      "step": 7281
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4709531939976761e-05,
+      "loss": 1.2067,
+      "step": 7282
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4689582935560164e-05,
+      "loss": 1.2827,
+      "step": 7283
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4669646395327363e-05,
+      "loss": 1.2625,
+      "step": 7284
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4649722322191184e-05,
+      "loss": 1.4447,
+      "step": 7285
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4629810719062597e-05,
+      "loss": 1.3298,
+      "step": 7286
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.46099115888507e-05,
+      "loss": 1.4058,
+      "step": 7287
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.459002493446293e-05,
+      "loss": 0.9244,
+      "step": 7288
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4570150758804779e-05,
+      "loss": 0.5476,
+      "step": 7289
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4550289064779887e-05,
+      "loss": 1.0978,
+      "step": 7290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.453043985529018e-05,
+      "loss": 0.8429,
+      "step": 7291
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.45106031332357e-05,
+      "loss": 1.1368,
+      "step": 7292
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4490778901514645e-05,
+      "loss": 0.8208,
+      "step": 7293
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.447096716302342e-05,
+      "loss": 1.3263,
+      "step": 7294
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4451167920656617e-05,
+      "loss": 1.0909,
+      "step": 7295
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.443138117730698e-05,
+      "loss": 1.0955,
+      "step": 7296
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4411606935865384e-05,
+      "loss": 1.2564,
+      "step": 7297
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4391845199220966e-05,
+      "loss": 0.8803,
+      "step": 7298
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4372095970261002e-05,
+      "loss": 0.9942,
+      "step": 7299
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4352359251870894e-05,
+      "loss": 1.3775,
+      "step": 7300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4332635046934261e-05,
+      "loss": 1.3852,
+      "step": 7301
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4312923358332919e-05,
+      "loss": 0.788,
+      "step": 7302
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4293224188946785e-05,
+      "loss": 0.6547,
+      "step": 7303
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4273537541654015e-05,
+      "loss": 0.8514,
+      "step": 7304
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.425386341933086e-05,
+      "loss": 1.2957,
+      "step": 7305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4234201824851822e-05,
+      "loss": 1.0227,
+      "step": 7306
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.421455276108955e-05,
+      "loss": 1.2429,
+      "step": 7307
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4194916230914835e-05,
+      "loss": 1.4248,
+      "step": 7308
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.417529223719658e-05,
+      "loss": 0.7172,
+      "step": 7309
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.415568078280205e-05,
+      "loss": 0.973,
+      "step": 7310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4136081870596485e-05,
+      "loss": 1.1185,
+      "step": 7311
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.411649550344335e-05,
+      "loss": 0.5599,
+      "step": 7312
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4096921684204312e-05,
+      "loss": 1.2212,
+      "step": 7313
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4077360415739204e-05,
+      "loss": 1.678,
+      "step": 7314
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4057811700905965e-05,
+      "loss": 1.0208,
+      "step": 7315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4038275542560741e-05,
+      "loss": 1.5697,
+      "step": 7316
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4018751943557884e-05,
+      "loss": 1.3721,
+      "step": 7317
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3999240906749811e-05,
+      "loss": 0.7661,
+      "step": 7318
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.397974243498722e-05,
+      "loss": 1.144,
+      "step": 7319
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3960256531118842e-05,
+      "loss": 1.3496,
+      "step": 7320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3940783197991714e-05,
+      "loss": 1.1085,
+      "step": 7321
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3921322438450912e-05,
+      "loss": 1.3187,
+      "step": 7322
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3901874255339743e-05,
+      "loss": 0.3874,
+      "step": 7323
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3882438651499696e-05,
+      "loss": 1.1293,
+      "step": 7324
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3863015629770337e-05,
+      "loss": 1.1384,
+      "step": 7325
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3843605192989484e-05,
+      "loss": 1.1487,
+      "step": 7326
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3824207343993056e-05,
+      "loss": 1.3423,
+      "step": 7327
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3804822085615155e-05,
+      "loss": 1.1152,
+      "step": 7328
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3785449420688067e-05,
+      "loss": 1.3781,
+      "step": 7329
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3766089352042177e-05,
+      "loss": 0.71,
+      "step": 7330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3746741882506076e-05,
+      "loss": 0.9001,
+      "step": 7331
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3727407014906546e-05,
+      "loss": 1.1957,
+      "step": 7332
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3708084752068451e-05,
+      "loss": 1.4314,
+      "step": 7333
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3688775096814833e-05,
+      "loss": 1.3242,
+      "step": 7334
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3669478051966921e-05,
+      "loss": 1.5911,
+      "step": 7335
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3650193620344121e-05,
+      "loss": 0.6738,
+      "step": 7336
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3630921804763907e-05,
+      "loss": 1.3413,
+      "step": 7337
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3611662608041998e-05,
+      "loss": 1.1508,
+      "step": 7338
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.359241603299225e-05,
+      "loss": 0.4562,
+      "step": 7339
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.357318208242664e-05,
+      "loss": 0.8345,
+      "step": 7340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3553960759155337e-05,
+      "loss": 1.2468,
+      "step": 7341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3534752065986633e-05,
+      "loss": 0.5627,
+      "step": 7342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3515556005727004e-05,
+      "loss": 0.8327,
+      "step": 7343
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3496372581181083e-05,
+      "loss": 0.8798,
+      "step": 7344
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3477201795151617e-05,
+      "loss": 0.6884,
+      "step": 7345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3458043650439556e-05,
+      "loss": 1.2226,
+      "step": 7346
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.343889814984396e-05,
+      "loss": 1.2987,
+      "step": 7347
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3419765296162079e-05,
+      "loss": 1.5733,
+      "step": 7348
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3400645092189268e-05,
+      "loss": 1.4221,
+      "step": 7349
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3381537540719091e-05,
+      "loss": 1.6909,
+      "step": 7350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3362442644543239e-05,
+      "loss": 1.3616,
+      "step": 7351
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3343360406451522e-05,
+      "loss": 0.8079,
+      "step": 7352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3324290829231945e-05,
+      "loss": 1.3804,
+      "step": 7353
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3305233915670668e-05,
+      "loss": 1.2926,
+      "step": 7354
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3286189668551963e-05,
+      "loss": 0.8976,
+      "step": 7355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3267158090658239e-05,
+      "loss": 1.2837,
+      "step": 7356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3248139184770115e-05,
+      "loss": 1.3315,
+      "step": 7357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3229132953666346e-05,
+      "loss": 1.4813,
+      "step": 7358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3210139400123777e-05,
+      "loss": 0.5608,
+      "step": 7359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3191158526917446e-05,
+      "loss": 1.0161,
+      "step": 7360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3172190336820578e-05,
+      "loss": 0.9494,
+      "step": 7361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3153234832604433e-05,
+      "loss": 1.2929,
+      "step": 7362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3134292017038529e-05,
+      "loss": 0.5876,
+      "step": 7363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3115361892890465e-05,
+      "loss": 1.2981,
+      "step": 7364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3096444462926005e-05,
+      "loss": 0.9203,
+      "step": 7365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3077539729909083e-05,
+      "loss": 0.9714,
+      "step": 7366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3058647696601723e-05,
+      "loss": 1.0015,
+      "step": 7367
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3039768365764138e-05,
+      "loss": 1.3867,
+      "step": 7368
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.302090174015469e-05,
+      "loss": 0.7016,
+      "step": 7369
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.300204782252985e-05,
+      "loss": 0.7121,
+      "step": 7370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2983206615644228e-05,
+      "loss": 1.216,
+      "step": 7371
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2964378122250609e-05,
+      "loss": 0.763,
+      "step": 7372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2945562345099948e-05,
+      "loss": 1.2989,
+      "step": 7373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2926759286941248e-05,
+      "loss": 0.9707,
+      "step": 7374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2907968950521732e-05,
+      "loss": 0.5882,
+      "step": 7375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2889191338586759e-05,
+      "loss": 1.5556,
+      "step": 7376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2870426453879802e-05,
+      "loss": 0.9211,
+      "step": 7377
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2851674299142458e-05,
+      "loss": 0.8254,
+      "step": 7378
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2832934877114499e-05,
+      "loss": 1.6786,
+      "step": 7379
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2814208190533861e-05,
+      "loss": 1.0766,
+      "step": 7380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2795494242136542e-05,
+      "loss": 1.2366,
+      "step": 7381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2776793034656743e-05,
+      "loss": 1.251,
+      "step": 7382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2758104570826801e-05,
+      "loss": 1.3573,
+      "step": 7383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2739428853377133e-05,
+      "loss": 1.5641,
+      "step": 7384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2720765885036367e-05,
+      "loss": 0.8379,
+      "step": 7385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2702115668531222e-05,
+      "loss": 1.3837,
+      "step": 7386
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2683478206586551e-05,
+      "loss": 1.4189,
+      "step": 7387
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2664853501925412e-05,
+      "loss": 1.121,
+      "step": 7388
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.264624155726889e-05,
+      "loss": 0.7496,
+      "step": 7389
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.262764237533628e-05,
+      "loss": 0.5955,
+      "step": 7390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2609055958845018e-05,
+      "loss": 0.6552,
+      "step": 7391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2590482310510632e-05,
+      "loss": 1.2219,
+      "step": 7392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2571921433046773e-05,
+      "loss": 0.8752,
+      "step": 7393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.255337332916533e-05,
+      "loss": 1.382,
+      "step": 7394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.253483800157621e-05,
+      "loss": 1.4543,
+      "step": 7395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.251631545298747e-05,
+      "loss": 1.4856,
+      "step": 7396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2497805686105369e-05,
+      "loss": 1.2022,
+      "step": 7397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2479308703634263e-05,
+      "loss": 1.0004,
+      "step": 7398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2460824508276614e-05,
+      "loss": 1.4559,
+      "step": 7399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2442353102732985e-05,
+      "loss": 0.841,
+      "step": 7400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2423894489702215e-05,
+      "loss": 1.8111,
+      "step": 7401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2405448671881126e-05,
+      "loss": 0.9532,
+      "step": 7402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2387015651964717e-05,
+      "loss": 0.5918,
+      "step": 7403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2368595432646146e-05,
+      "loss": 1.1859,
+      "step": 7404
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2350188016616682e-05,
+      "loss": 1.3132,
+      "step": 7405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2331793406565684e-05,
+      "loss": 1.6558,
+      "step": 7406
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2313411605180724e-05,
+      "loss": 0.8769,
+      "step": 7407
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2295042615147424e-05,
+      "loss": 1.2241,
+      "step": 7408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2276686439149553e-05,
+      "loss": 1.1685,
+      "step": 7409
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2258343079869073e-05,
+      "loss": 1.1455,
+      "step": 7410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2240012539985956e-05,
+      "loss": 1.0312,
+      "step": 7411
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2221694822178409e-05,
+      "loss": 1.6205,
+      "step": 7412
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.220338992912272e-05,
+      "loss": 1.1861,
+      "step": 7413
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2185097863493289e-05,
+      "loss": 1.4193,
+      "step": 7414
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2166818627962628e-05,
+      "loss": 0.7997,
+      "step": 7415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2148552225201481e-05,
+      "loss": 0.6811,
+      "step": 7416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2130298657878592e-05,
+      "loss": 0.627,
+      "step": 7417
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.211205792866087e-05,
+      "loss": 1.4649,
+      "step": 7418
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2093830040213362e-05,
+      "loss": 0.9114,
+      "step": 7419
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2075614995199269e-05,
+      "loss": 0.7272,
+      "step": 7420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2057412796279855e-05,
+      "loss": 0.8303,
+      "step": 7421
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2039223446114489e-05,
+      "loss": 0.4961,
+      "step": 7422
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2021046947360793e-05,
+      "loss": 0.7045,
+      "step": 7423
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.200288330267436e-05,
+      "loss": 1.3623,
+      "step": 7424
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1984732514708986e-05,
+      "loss": 1.0984,
+      "step": 7425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1966594586116564e-05,
+      "loss": 1.3065,
+      "step": 7426
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1948469519547144e-05,
+      "loss": 0.7861,
+      "step": 7427
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1930357317648844e-05,
+      "loss": 1.2679,
+      "step": 7428
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1912257983067943e-05,
+      "loss": 0.6786,
+      "step": 7429
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1894171518448793e-05,
+      "loss": 0.9211,
+      "step": 7430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1876097926433915e-05,
+      "loss": 0.7162,
+      "step": 7431
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1858037209663953e-05,
+      "loss": 0.4126,
+      "step": 7432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1839989370777616e-05,
+      "loss": 1.4451,
+      "step": 7433
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1821954412411774e-05,
+      "loss": 1.2528,
+      "step": 7434
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.180393233720143e-05,
+      "loss": 0.9983,
+      "step": 7435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1785923147779654e-05,
+      "loss": 1.4153,
+      "step": 7436
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1767926846777622e-05,
+      "loss": 1.1614,
+      "step": 7437
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.174994343682474e-05,
+      "loss": 1.3139,
+      "step": 7438
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1731972920548418e-05,
+      "loss": 0.5958,
+      "step": 7439
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.17140153005742e-05,
+      "loss": 1.0383,
+      "step": 7440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1696070579525776e-05,
+      "loss": 1.3159,
+      "step": 7441
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.167813876002497e-05,
+      "loss": 1.0233,
+      "step": 7442
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1660219844691644e-05,
+      "loss": 1.1216,
+      "step": 7443
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1642313836143836e-05,
+      "loss": 1.161,
+      "step": 7444
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.162442073699772e-05,
+      "loss": 1.5599,
+      "step": 7445
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1606540549867517e-05,
+      "loss": 1.3439,
+      "step": 7446
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.158867327736557e-05,
+      "loss": 0.8719,
+      "step": 7447
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1570818922102378e-05,
+      "loss": 1.2136,
+      "step": 7448
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1552977486686556e-05,
+      "loss": 1.118,
+      "step": 7449
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1535148973724763e-05,
+      "loss": 1.178,
+      "step": 7450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1517333385821826e-05,
+      "loss": 0.7987,
+      "step": 7451
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1499530725580709e-05,
+      "loss": 0.957,
+      "step": 7452
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1481740995602408e-05,
+      "loss": 1.6054,
+      "step": 7453
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1463964198486098e-05,
+      "loss": 1.3442,
+      "step": 7454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1446200336829005e-05,
+      "loss": 0.839,
+      "step": 7455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1428449413226527e-05,
+      "loss": 1.5532,
+      "step": 7456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1410711430272147e-05,
+      "loss": 1.368,
+      "step": 7457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1392986390557447e-05,
+      "loss": 0.6881,
+      "step": 7458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1375274296672078e-05,
+      "loss": 0.7397,
+      "step": 7459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1357575151203925e-05,
+      "loss": 0.9306,
+      "step": 7460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1339888956738876e-05,
+      "loss": 1.3823,
+      "step": 7461
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1322215715860906e-05,
+      "loss": 1.2286,
+      "step": 7462
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1304555431152174e-05,
+      "loss": 1.0137,
+      "step": 7463
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1286908105192951e-05,
+      "loss": 1.0575,
+      "step": 7464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1269273740561536e-05,
+      "loss": 1.1931,
+      "step": 7465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1251652339834384e-05,
+      "loss": 1.5069,
+      "step": 7466
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1234043905586077e-05,
+      "loss": 0.9596,
+      "step": 7467
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1216448440389237e-05,
+      "loss": 1.6471,
+      "step": 7468
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1198865946814673e-05,
+      "loss": 1.6517,
+      "step": 7469
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1181296427431209e-05,
+      "loss": 1.0026,
+      "step": 7470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1163739884805846e-05,
+      "loss": 1.0322,
+      "step": 7471
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1146196321503689e-05,
+      "loss": 1.2744,
+      "step": 7472
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1128665740087863e-05,
+      "loss": 0.7066,
+      "step": 7473
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1111148143119709e-05,
+      "loss": 1.0988,
+      "step": 7474
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.109364353315857e-05,
+      "loss": 1.1465,
+      "step": 7475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1076151912761979e-05,
+      "loss": 1.1563,
+      "step": 7476
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1058673284485488e-05,
+      "loss": 1.2478,
+      "step": 7477
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1041207650882811e-05,
+      "loss": 0.5982,
+      "step": 7478
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1023755014505766e-05,
+      "loss": 1.0848,
+      "step": 7479
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.100631537790422e-05,
+      "loss": 1.3064,
+      "step": 7480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.098888874362619e-05,
+      "loss": 1.5655,
+      "step": 7481
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0971475114217778e-05,
+      "loss": 1.396,
+      "step": 7482
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0954074492223188e-05,
+      "loss": 1.7321,
+      "step": 7483
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0936686880184677e-05,
+      "loss": 1.2458,
+      "step": 7484
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0919312280642668e-05,
+      "loss": 1.2128,
+      "step": 7485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0901950696135688e-05,
+      "loss": 0.7442,
+      "step": 7486
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0884602129200283e-05,
+      "loss": 1.1661,
+      "step": 7487
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0867266582371171e-05,
+      "loss": 1.048,
+      "step": 7488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.084994405818115e-05,
+      "loss": 1.3522,
+      "step": 7489
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0832634559161092e-05,
+      "loss": 0.8223,
+      "step": 7490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0815338087839999e-05,
+      "loss": 1.1354,
+      "step": 7491
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0798054646744926e-05,
+      "loss": 0.6913,
+      "step": 7492
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0780784238401065e-05,
+      "loss": 1.3485,
+      "step": 7493
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0763526865331707e-05,
+      "loss": 1.155,
+      "step": 7494
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0746282530058194e-05,
+      "loss": 0.7155,
+      "step": 7495
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0729051235099996e-05,
+      "loss": 0.9462,
+      "step": 7496
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.07118329829747e-05,
+      "loss": 1.2005,
+      "step": 7497
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0694627776197941e-05,
+      "loss": 1.3145,
+      "step": 7498
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0677435617283449e-05,
+      "loss": 1.199,
+      "step": 7499
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0660256508743072e-05,
+      "loss": 0.869,
+      "step": 7500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0643090453086767e-05,
+      "loss": 1.2318,
+      "step": 7501
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0625937452822544e-05,
+      "loss": 0.7142,
+      "step": 7502
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0608797510456514e-05,
+      "loss": 0.9221,
+      "step": 7503
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0591670628492933e-05,
+      "loss": 1.5264,
+      "step": 7504
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0574556809434066e-05,
+      "loss": 1.3041,
+      "step": 7505
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.055745605578029e-05,
+      "loss": 0.5534,
+      "step": 7506
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.054036837003013e-05,
+      "loss": 0.8354,
+      "step": 7507
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0523293754680163e-05,
+      "loss": 0.5756,
+      "step": 7508
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0506232212225032e-05,
+      "loss": 1.2542,
+      "step": 7509
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0489183745157504e-05,
+      "loss": 1.1173,
+      "step": 7510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0472148355968448e-05,
+      "loss": 1.3943,
+      "step": 7511
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0455126047146769e-05,
+      "loss": 1.4617,
+      "step": 7512
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0438116821179523e-05,
+      "loss": 0.936,
+      "step": 7513
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0421120680551788e-05,
+      "loss": 0.6201,
+      "step": 7514
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.040413762774679e-05,
+      "loss": 1.1326,
+      "step": 7515
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0387167665245824e-05,
+      "loss": 1.1898,
+      "step": 7516
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0370210795528245e-05,
+      "loss": 0.9896,
+      "step": 7517
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0353267021071533e-05,
+      "loss": 0.9949,
+      "step": 7518
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.033633634435125e-05,
+      "loss": 1.2422,
+      "step": 7519
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0319418767841027e-05,
+      "loss": 0.8119,
+      "step": 7520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0302514294012544e-05,
+      "loss": 0.7964,
+      "step": 7521
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.028562292533568e-05,
+      "loss": 1.2715,
+      "step": 7522
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0268744664278295e-05,
+      "loss": 1.3373,
+      "step": 7523
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0251879513306362e-05,
+      "loss": 0.9978,
+      "step": 7524
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0235027474883951e-05,
+      "loss": 0.4235,
+      "step": 7525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.021818855147324e-05,
+      "loss": 1.1644,
+      "step": 7526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0201362745534427e-05,
+      "loss": 1.0846,
+      "step": 7527
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.018455005952581e-05,
+      "loss": 0.8456,
+      "step": 7528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0167750495903827e-05,
+      "loss": 0.9978,
+      "step": 7529
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0150964057122958e-05,
+      "loss": 1.3005,
+      "step": 7530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.013419074563574e-05,
+      "loss": 1.5779,
+      "step": 7531
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0117430563892826e-05,
+      "loss": 1.2748,
+      "step": 7532
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.010068351434298e-05,
+      "loss": 0.3996,
+      "step": 7533
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0083949599432963e-05,
+      "loss": 1.205,
+      "step": 7534
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.006722882160771e-05,
+      "loss": 0.5462,
+      "step": 7535
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0050521183310136e-05,
+      "loss": 0.9756,
+      "step": 7536
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0033826686981328e-05,
+      "loss": 1.2545,
+      "step": 7537
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0017145335060429e-05,
+      "loss": 1.2813,
+      "step": 7538
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.000047712998462e-05,
+      "loss": 1.063,
+      "step": 7539
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.983822074189198e-06,
+      "loss": 0.7402,
+      "step": 7540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.967180170107549e-06,
+      "loss": 1.2598,
+      "step": 7541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.950551420171117e-06,
+      "loss": 1.2893,
+      "step": 7542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.93393582680937e-06,
+      "loss": 1.0113,
+      "step": 7543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.917333392450001e-06,
+      "loss": 1.146,
+      "step": 7544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.900744119518646e-06,
+      "loss": 1.0132,
+      "step": 7545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.884168010439043e-06,
+      "loss": 0.8357,
+      "step": 7546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.867605067633045e-06,
+      "loss": 0.6643,
+      "step": 7547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.851055293520594e-06,
+      "loss": 1.0046,
+      "step": 7548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.834518690519634e-06,
+      "loss": 1.1902,
+      "step": 7549
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.81799526104621e-06,
+      "loss": 1.2641,
+      "step": 7550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.801485007514522e-06,
+      "loss": 1.3844,
+      "step": 7551
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.784987932336753e-06,
+      "loss": 1.1559,
+      "step": 7552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.768504037923165e-06,
+      "loss": 1.3587,
+      "step": 7553
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.75203332668213e-06,
+      "loss": 1.1471,
+      "step": 7554
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.735575801020124e-06,
+      "loss": 0.9572,
+      "step": 7555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.719131463341603e-06,
+      "loss": 0.8534,
+      "step": 7556
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.702700316049185e-06,
+      "loss": 0.6111,
+      "step": 7557
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.686282361543498e-06,
+      "loss": 0.8244,
+      "step": 7558
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.669877602223276e-06,
+      "loss": 0.9278,
+      "step": 7559
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.653486040485337e-06,
+      "loss": 1.0949,
+      "step": 7560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.637107678724521e-06,
+      "loss": 1.2326,
+      "step": 7561
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.620742519333792e-06,
+      "loss": 1.2572,
+      "step": 7562
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.604390564704169e-06,
+      "loss": 1.0379,
+      "step": 7563
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.588051817224741e-06,
+      "loss": 0.9615,
+      "step": 7564
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.5717262792826e-06,
+      "loss": 1.0926,
+      "step": 7565
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.555413953263048e-06,
+      "loss": 0.5279,
+      "step": 7566
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.539114841549357e-06,
+      "loss": 1.3042,
+      "step": 7567
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.522828946522865e-06,
+      "loss": 1.5569,
+      "step": 7568
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.506556270563028e-06,
+      "loss": 0.7202,
+      "step": 7569
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.49029681604735e-06,
+      "loss": 1.2939,
+      "step": 7570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.47405058535138e-06,
+      "loss": 1.3984,
+      "step": 7571
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.45781758084876e-06,
+      "loss": 0.7615,
+      "step": 7572
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.441597804911228e-06,
+      "loss": 1.4819,
+      "step": 7573
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.425391259908523e-06,
+      "loss": 1.4359,
+      "step": 7574
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.409197948208481e-06,
+      "loss": 0.8155,
+      "step": 7575
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.393017872177023e-06,
+      "loss": 0.7923,
+      "step": 7576
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.376851034178124e-06,
+      "loss": 1.2509,
+      "step": 7577
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.360697436573807e-06,
+      "loss": 1.1581,
+      "step": 7578
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.344557081724203e-06,
+      "loss": 0.5923,
+      "step": 7579
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.32842997198744e-06,
+      "loss": 1.2603,
+      "step": 7580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.312316109719765e-06,
+      "loss": 0.6788,
+      "step": 7581
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.296215497275507e-06,
+      "loss": 1.1069,
+      "step": 7582
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.280128137006994e-06,
+      "loss": 0.7641,
+      "step": 7583
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.264054031264658e-06,
+      "loss": 1.1583,
+      "step": 7584
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.247993182397008e-06,
+      "loss": 1.3237,
+      "step": 7585
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.231945592750579e-06,
+      "loss": 1.321,
+      "step": 7586
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.215911264669964e-06,
+      "loss": 1.1562,
+      "step": 7587
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.199890200497897e-06,
+      "loss": 0.4575,
+      "step": 7588
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.183882402575084e-06,
+      "loss": 0.6251,
+      "step": 7589
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.167887873240311e-06,
+      "loss": 1.1407,
+      "step": 7590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.151906614830451e-06,
+      "loss": 1.0063,
+      "step": 7591
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.135938629680463e-06,
+      "loss": 1.3219,
+      "step": 7592
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.119983920123287e-06,
+      "loss": 1.3291,
+      "step": 7593
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.104042488489984e-06,
+      "loss": 1.0902,
+      "step": 7594
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.088114337109687e-06,
+      "loss": 1.3586,
+      "step": 7595
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.072199468309505e-06,
+      "loss": 1.1076,
+      "step": 7596
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.056297884414722e-06,
+      "loss": 0.5029,
+      "step": 7597
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.040409587748566e-06,
+      "loss": 1.1905,
+      "step": 7598
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.024534580632416e-06,
+      "loss": 1.7879,
+      "step": 7599
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.008672865385648e-06,
+      "loss": 0.9384,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.992824444325742e-06,
+      "loss": 1.6502,
+      "step": 7601
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.976989319768215e-06,
+      "loss": 1.2065,
+      "step": 7602
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.961167494026613e-06,
+      "loss": 1.2179,
+      "step": 7603
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.945358969412599e-06,
+      "loss": 1.1789,
+      "step": 7604
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.929563748235824e-06,
+      "loss": 1.4015,
+      "step": 7605
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.913781832804058e-06,
+      "loss": 0.7112,
+      "step": 7606
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.898013225423107e-06,
+      "loss": 1.223,
+      "step": 7607
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.88225792839682e-06,
+      "loss": 1.017,
+      "step": 7608
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.866515944027054e-06,
+      "loss": 1.3188,
+      "step": 7609
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.85078727461387e-06,
+      "loss": 1.5641,
+      "step": 7610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.83507192245523e-06,
+      "loss": 1.0898,
+      "step": 7611
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.819369889847196e-06,
+      "loss": 0.642,
+      "step": 7612
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.803681179083912e-06,
+      "loss": 1.218,
+      "step": 7613
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.788005792457588e-06,
+      "loss": 1.882,
+      "step": 7614
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.772343732258414e-06,
+      "loss": 1.365,
+      "step": 7615
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.756695000774707e-06,
+      "loss": 1.5074,
+      "step": 7616
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.741059600292812e-06,
+      "loss": 1.2548,
+      "step": 7617
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.725437533097081e-06,
+      "loss": 1.1586,
+      "step": 7618
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.709828801470021e-06,
+      "loss": 0.5897,
+      "step": 7619
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.694233407692064e-06,
+      "loss": 1.1092,
+      "step": 7620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.67865135404179e-06,
+      "loss": 1.3338,
+      "step": 7621
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.663082642795817e-06,
+      "loss": 1.4277,
+      "step": 7622
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.64752727622875e-06,
+      "loss": 1.1012,
+      "step": 7623
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.63198525661334e-06,
+      "loss": 1.201,
+      "step": 7624
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.616456586220279e-06,
+      "loss": 0.4872,
+      "step": 7625
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.600941267318419e-06,
+      "loss": 0.9951,
+      "step": 7626
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.58543930217457e-06,
+      "loss": 1.3272,
+      "step": 7627
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.56995069305363e-06,
+      "loss": 0.6881,
+      "step": 7628
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.554475442218591e-06,
+      "loss": 1.2431,
+      "step": 7629
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.539013551930409e-06,
+      "loss": 1.0181,
+      "step": 7630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.523565024448098e-06,
+      "loss": 1.3374,
+      "step": 7631
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.508129862028814e-06,
+      "loss": 0.5757,
+      "step": 7632
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.492708066927667e-06,
+      "loss": 1.4218,
+      "step": 7633
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.477299641397818e-06,
+      "loss": 1.3877,
+      "step": 7634
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.461904587690517e-06,
+      "loss": 1.123,
+      "step": 7635
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.446522908055066e-06,
+      "loss": 1.2392,
+      "step": 7636
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.431154604738733e-06,
+      "loss": 0.6345,
+      "step": 7637
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.415799679986924e-06,
+      "loss": 1.4988,
+      "step": 7638
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.400458136043065e-06,
+      "loss": 0.8851,
+      "step": 7639
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.385129975148576e-06,
+      "loss": 1.3259,
+      "step": 7640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.369815199543007e-06,
+      "loss": 0.9931,
+      "step": 7641
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.354513811463848e-06,
+      "loss": 0.9216,
+      "step": 7642
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.339225813146745e-06,
+      "loss": 1.2663,
+      "step": 7643
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.32395120682532e-06,
+      "loss": 0.8147,
+      "step": 7644
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.308689994731223e-06,
+      "loss": 1.1423,
+      "step": 7645
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.293442179094203e-06,
+      "loss": 1.6622,
+      "step": 7646
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.278207762142042e-06,
+      "loss": 0.9931,
+      "step": 7647
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.262986746100532e-06,
+      "loss": 1.3496,
+      "step": 7648
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.24777913319349e-06,
+      "loss": 1.0645,
+      "step": 7649
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.232584925642839e-06,
+      "loss": 0.843,
+      "step": 7650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.217404125668527e-06,
+      "loss": 1.419,
+      "step": 7651
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.202236735488489e-06,
+      "loss": 0.5208,
+      "step": 7652
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.187082757318766e-06,
+      "loss": 0.7436,
+      "step": 7653
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.171942193373416e-06,
+      "loss": 0.7552,
+      "step": 7654
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.156815045864529e-06,
+      "loss": 0.6488,
+      "step": 7655
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.141701317002216e-06,
+      "loss": 0.8978,
+      "step": 7656
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.126601008994672e-06,
+      "loss": 0.9702,
+      "step": 7657
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.111514124048124e-06,
+      "loss": 1.2595,
+      "step": 7658
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.0964406643668e-06,
+      "loss": 1.2696,
+      "step": 7659
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.081380632153002e-06,
+      "loss": 1.6902,
+      "step": 7660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.06633402960707e-06,
+      "loss": 1.1146,
+      "step": 7661
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.051300858927357e-06,
+      "loss": 1.3466,
+      "step": 7662
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.036281122310274e-06,
+      "loss": 0.9393,
+      "step": 7663
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.021274821950253e-06,
+      "loss": 1.2587,
+      "step": 7664
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.006281960039775e-06,
+      "loss": 0.8789,
+      "step": 7665
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.991302538769373e-06,
+      "loss": 1.6181,
+      "step": 7666
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.97633656032758e-06,
+      "loss": 1.1259,
+      "step": 7667
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.961384026900986e-06,
+      "loss": 0.9684,
+      "step": 7668
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.946444940674236e-06,
+      "loss": 0.5394,
+      "step": 7669
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.93151930382997e-06,
+      "loss": 1.2517,
+      "step": 7670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.916607118548846e-06,
+      "loss": 0.5863,
+      "step": 7671
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.901708387009665e-06,
+      "loss": 0.7839,
+      "step": 7672
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.886823111389152e-06,
+      "loss": 0.8587,
+      "step": 7673
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.871951293862078e-06,
+      "loss": 0.832,
+      "step": 7674
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.857092936601317e-06,
+      "loss": 0.7829,
+      "step": 7675
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.842248041777722e-06,
+      "loss": 0.8929,
+      "step": 7676
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.82741661156019e-06,
+      "loss": 0.9541,
+      "step": 7677
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.812598648115632e-06,
+      "loss": 0.4065,
+      "step": 7678
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.797794153609017e-06,
+      "loss": 1.5638,
+      "step": 7679
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.783003130203358e-06,
+      "loss": 1.2484,
+      "step": 7680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.768225580059663e-06,
+      "loss": 0.5709,
+      "step": 7681
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.75346150533698e-06,
+      "loss": 1.214,
+      "step": 7682
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.738710908192438e-06,
+      "loss": 1.1546,
+      "step": 7683
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.723973790781114e-06,
+      "loss": 1.1596,
+      "step": 7684
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.709250155256187e-06,
+      "loss": 1.3029,
+      "step": 7685
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.694540003768814e-06,
+      "loss": 1.4233,
+      "step": 7686
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.679843338468218e-06,
+      "loss": 1.2931,
+      "step": 7687
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.66516016150165e-06,
+      "loss": 0.9246,
+      "step": 7688
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.650490475014349e-06,
+      "loss": 1.1037,
+      "step": 7689
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.635834281149646e-06,
+      "loss": 1.0948,
+      "step": 7690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.621191582048859e-06,
+      "loss": 0.9407,
+      "step": 7691
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.606562379851334e-06,
+      "loss": 0.8745,
+      "step": 7692
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.591946676694428e-06,
+      "loss": 1.2316,
+      "step": 7693
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5773444747136215e-06,
+      "loss": 1.215,
+      "step": 7694
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.562755776042307e-06,
+      "loss": 0.8677,
+      "step": 7695
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.548180582811937e-06,
+      "loss": 0.52,
+      "step": 7696
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5336188971520284e-06,
+      "loss": 0.9144,
+      "step": 7697
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.519070721190114e-06,
+      "loss": 1.4214,
+      "step": 7698
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5045360570517145e-06,
+      "loss": 0.5499,
+      "step": 7699
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.490014906860399e-06,
+      "loss": 1.3981,
+      "step": 7700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.47550727273777e-06,
+      "loss": 1.0388,
+      "step": 7701
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.461013156803465e-06,
+      "loss": 1.1895,
+      "step": 7702
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.446532561175102e-06,
+      "loss": 1.4106,
+      "step": 7703
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.432065487968354e-06,
+      "loss": 0.9728,
+      "step": 7704
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.417611939296965e-06,
+      "loss": 1.3395,
+      "step": 7705
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.403171917272589e-06,
+      "loss": 1.1874,
+      "step": 7706
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.388745424005028e-06,
+      "loss": 0.9366,
+      "step": 7707
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.374332461601996e-06,
+      "loss": 1.2284,
+      "step": 7708
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.359933032169308e-06,
+      "loss": 1.0775,
+      "step": 7709
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.345547137810804e-06,
+      "loss": 0.508,
+      "step": 7710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3311747806282585e-06,
+      "loss": 1.5016,
+      "step": 7711
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.316815962721568e-06,
+      "loss": 1.0751,
+      "step": 7712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.302470686188623e-06,
+      "loss": 0.447,
+      "step": 7713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.288138953125312e-06,
+      "loss": 1.0382,
+      "step": 7714
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2738207656255054e-06,
+      "loss": 1.1542,
+      "step": 7715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.25951612578124e-06,
+      "loss": 1.1949,
+      "step": 7716
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.245225035682424e-06,
+      "loss": 1.2882,
+      "step": 7717
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.2309474974170285e-06,
+      "loss": 0.7613,
+      "step": 7718
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.216683513071077e-06,
+      "loss": 0.4906,
+      "step": 7719
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.202433084728621e-06,
+      "loss": 1.5706,
+      "step": 7720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.188196214471654e-06,
+      "loss": 1.0104,
+      "step": 7721
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.173972904380255e-06,
+      "loss": 1.1825,
+      "step": 7722
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.159763156532528e-06,
+      "loss": 0.8139,
+      "step": 7723
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.145566973004558e-06,
+      "loss": 0.8692,
+      "step": 7724
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.131384355870441e-06,
+      "loss": 1.277,
+      "step": 7725
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.117215307202341e-06,
+      "loss": 0.9529,
+      "step": 7726
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.103059829070402e-06,
+      "loss": 1.0086,
+      "step": 7727
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.088917923542782e-06,
+      "loss": 1.193,
+      "step": 7728
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.0747895926857045e-06,
+      "loss": 0.9981,
+      "step": 7729
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.060674838563319e-06,
+      "loss": 1.1872,
+      "step": 7730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.046573663237888e-06,
+      "loss": 1.3239,
+      "step": 7731
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.0324860687696504e-06,
+      "loss": 1.1939,
+      "step": 7732
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.018412057216816e-06,
+      "loss": 0.8523,
+      "step": 7733
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.004351630635697e-06,
+      "loss": 1.7309,
+      "step": 7734
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.9903047910805705e-06,
+      "loss": 1.2857,
+      "step": 7735
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.976271540603719e-06,
+      "loss": 1.0985,
+      "step": 7736
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.962251881255433e-06,
+      "loss": 0.903,
+      "step": 7737
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.948245815084098e-06,
+      "loss": 1.0145,
+      "step": 7738
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.934253344136022e-06,
+      "loss": 0.6456,
+      "step": 7739
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.920274470455535e-06,
+      "loss": 1.2816,
+      "step": 7740
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.906309196085048e-06,
+      "loss": 1.6306,
+      "step": 7741
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.89235752306493e-06,
+      "loss": 1.2384,
+      "step": 7742
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.878419453433549e-06,
+      "loss": 0.7047,
+      "step": 7743
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.864494989227343e-06,
+      "loss": 1.3026,
+      "step": 7744
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.850584132480731e-06,
+      "loss": 1.2956,
+      "step": 7745
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.836686885226107e-06,
+      "loss": 0.7056,
+      "step": 7746
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.822803249493959e-06,
+      "loss": 1.3288,
+      "step": 7747
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.808933227312697e-06,
+      "loss": 1.4045,
+      "step": 7748
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.795076820708823e-06,
+      "loss": 1.3487,
+      "step": 7749
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.7812340317067845e-06,
+      "loss": 1.1375,
+      "step": 7750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.767404862329074e-06,
+      "loss": 1.071,
+      "step": 7751
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.7535893145962096e-06,
+      "loss": 0.7412,
+      "step": 7752
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.739787390526664e-06,
+      "loss": 1.054,
+      "step": 7753
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.72599909213697e-06,
+      "loss": 1.257,
+      "step": 7754
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.712224421441637e-06,
+      "loss": 1.0625,
+      "step": 7755
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.698463380453213e-06,
+      "loss": 1.3906,
+      "step": 7756
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.684715971182243e-06,
+      "loss": 1.1288,
+      "step": 7757
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.670982195637265e-06,
+      "loss": 1.362,
+      "step": 7758
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.65726205582482e-06,
+      "loss": 1.1891,
+      "step": 7759
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.643555553749514e-06,
+      "loss": 1.8036,
+      "step": 7760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.629862691413913e-06,
+      "loss": 1.5891,
+      "step": 7761
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.61618347081856e-06,
+      "loss": 0.4877,
+      "step": 7762
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.602517893962068e-06,
+      "loss": 1.0045,
+      "step": 7763
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.588865962841051e-06,
+      "loss": 0.5032,
+      "step": 7764
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.575227679450069e-06,
+      "loss": 0.4785,
+      "step": 7765
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.561603045781739e-06,
+      "loss": 0.9786,
+      "step": 7766
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.547992063826713e-06,
+      "loss": 0.6249,
+      "step": 7767
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.534394735573557e-06,
+      "loss": 0.4849,
+      "step": 7768
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.520811063008925e-06,
+      "loss": 0.8047,
+      "step": 7769
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.507241048117418e-06,
+      "loss": 0.7616,
+      "step": 7770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.493684692881685e-06,
+      "loss": 1.1171,
+      "step": 7771
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.480141999282374e-06,
+      "loss": 1.5156,
+      "step": 7772
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.466612969298102e-06,
+      "loss": 1.4058,
+      "step": 7773
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.453097604905534e-06,
+      "loss": 1.3202,
+      "step": 7774
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4395959080792875e-06,
+      "loss": 1.1731,
+      "step": 7775
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.426107880792054e-06,
+      "loss": 0.9052,
+      "step": 7776
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.412633525014444e-06,
+      "loss": 0.7252,
+      "step": 7777
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.399172842715129e-06,
+      "loss": 1.3098,
+      "step": 7778
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.38572583586079e-06,
+      "loss": 1.0612,
+      "step": 7779
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.3722925064160665e-06,
+      "loss": 1.5169,
+      "step": 7780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.35887285634359e-06,
+      "loss": 0.9669,
+      "step": 7781
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.34546688760409e-06,
+      "loss": 1.2575,
+      "step": 7782
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.332074602156201e-06,
+      "loss": 1.1185,
+      "step": 7783
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.318696001956559e-06,
+      "loss": 0.6105,
+      "step": 7784
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.305331088959854e-06,
+      "loss": 1.3843,
+      "step": 7785
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.29197986511878e-06,
+      "loss": 1.2032,
+      "step": 7786
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.278642332383944e-06,
+      "loss": 1.3658,
+      "step": 7787
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.265318492704042e-06,
+      "loss": 0.9981,
+      "step": 7788
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.2520083480257615e-06,
+      "loss": 1.2537,
+      "step": 7789
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.238711900293725e-06,
+      "loss": 1.0398,
+      "step": 7790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.225429151450634e-06,
+      "loss": 0.8906,
+      "step": 7791
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.212160103437104e-06,
+      "loss": 0.5663,
+      "step": 7792
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.198904758191826e-06,
+      "loss": 0.5116,
+      "step": 7793
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.185663117651463e-06,
+      "loss": 1.6222,
+      "step": 7794
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.172435183750646e-06,
+      "loss": 1.439,
+      "step": 7795
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.15922095842204e-06,
+      "loss": 1.4008,
+      "step": 7796
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.146020443596312e-06,
+      "loss": 0.7394,
+      "step": 7797
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.132833641202096e-06,
+      "loss": 1.0185,
+      "step": 7798
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.119660553166007e-06,
+      "loss": 1.6384,
+      "step": 7799
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.106501181412716e-06,
+      "loss": 0.5825,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.093355527864852e-06,
+      "loss": 1.2985,
+      "step": 7801
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.080223594443046e-06,
+      "loss": 1.1151,
+      "step": 7802
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.067105383065907e-06,
+      "loss": 1.1657,
+      "step": 7803
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.054000895650102e-06,
+      "loss": 0.6357,
+      "step": 7804
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.0409101341102114e-06,
+      "loss": 1.46,
+      "step": 7805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.027833100358837e-06,
+      "loss": 0.8317,
+      "step": 7806
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.014769796306608e-06,
+      "loss": 1.3017,
+      "step": 7807
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.001720223862118e-06,
+      "loss": 0.6106,
+      "step": 7808
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.988684384931953e-06,
+      "loss": 1.3539,
+      "step": 7809
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.975662281420713e-06,
+      "loss": 0.6476,
+      "step": 7810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.962653915230976e-06,
+      "loss": 1.5682,
+      "step": 7811
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.949659288263298e-06,
+      "loss": 1.6869,
+      "step": 7812
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.936678402416262e-06,
+      "loss": 1.3434,
+      "step": 7813
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.923711259586418e-06,
+      "loss": 1.0232,
+      "step": 7814
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.910757861668315e-06,
+      "loss": 0.4271,
+      "step": 7815
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.897818210554506e-06,
+      "loss": 1.2318,
+      "step": 7816
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.884892308135515e-06,
+      "loss": 1.0522,
+      "step": 7817
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.87198015629985e-06,
+      "loss": 1.2833,
+      "step": 7818
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.8590817569340705e-06,
+      "loss": 1.4859,
+      "step": 7819
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.846197111922658e-06,
+      "loss": 1.6485,
+      "step": 7820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.833326223148072e-06,
+      "loss": 1.5149,
+      "step": 7821
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.8204690924908655e-06,
+      "loss": 1.2134,
+      "step": 7822
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.807625721829491e-06,
+      "loss": 1.5601,
+      "step": 7823
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7947961130403905e-06,
+      "loss": 0.4767,
+      "step": 7824
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.781980267998044e-06,
+      "loss": 1.1069,
+      "step": 7825
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.769178188574897e-06,
+      "loss": 1.0726,
+      "step": 7826
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.756389876641388e-06,
+      "loss": 0.9168,
+      "step": 7827
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7436153340658995e-06,
+      "loss": 1.2152,
+      "step": 7828
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.730854562714882e-06,
+      "loss": 1.2361,
+      "step": 7829
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.718107564452735e-06,
+      "loss": 0.8407,
+      "step": 7830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7053743411418116e-06,
+      "loss": 0.5958,
+      "step": 7831
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.692654894642513e-06,
+      "loss": 1.4135,
+      "step": 7832
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.6799492268131975e-06,
+      "loss": 1.1426,
+      "step": 7833
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.667257339510201e-06,
+      "loss": 0.6838,
+      "step": 7834
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.6545792345878755e-06,
+      "loss": 1.5306,
+      "step": 7835
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.641914913898516e-06,
+      "loss": 1.281,
+      "step": 7836
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.6292643792924315e-06,
+      "loss": 0.7375,
+      "step": 7837
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.616627632617955e-06,
+      "loss": 1.4777,
+      "step": 7838
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.604004675721319e-06,
+      "loss": 1.5892,
+      "step": 7839
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.591395510446795e-06,
+      "loss": 0.6867,
+      "step": 7840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.578800138636664e-06,
+      "loss": 1.1219,
+      "step": 7841
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.56621856213112e-06,
+      "loss": 1.2654,
+      "step": 7842
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.553650782768372e-06,
+      "loss": 1.3013,
+      "step": 7843
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.541096802384671e-06,
+      "loss": 0.7706,
+      "step": 7844
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.528556622814185e-06,
+      "loss": 0.4471,
+      "step": 7845
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.516030245889048e-06,
+      "loss": 0.8928,
+      "step": 7846
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.50351767343944e-06,
+      "loss": 1.174,
+      "step": 7847
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.491018907293499e-06,
+      "loss": 0.7472,
+      "step": 7848
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.478533949277332e-06,
+      "loss": 1.3873,
+      "step": 7849
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.466062801215066e-06,
+      "loss": 0.6562,
+      "step": 7850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.453605464928735e-06,
+      "loss": 1.4558,
+      "step": 7851
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.44116194223846e-06,
+      "loss": 0.5809,
+      "step": 7852
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.428732234962241e-06,
+      "loss": 1.69,
+      "step": 7853
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.416316344916128e-06,
+      "loss": 0.9242,
+      "step": 7854
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4039142739141455e-06,
+      "loss": 1.1681,
+      "step": 7855
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.391526023768256e-06,
+      "loss": 0.9069,
+      "step": 7856
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3791515962884565e-06,
+      "loss": 1.3245,
+      "step": 7857
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.366790993282678e-06,
+      "loss": 1.1031,
+      "step": 7858
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.354444216556853e-06,
+      "loss": 0.9897,
+      "step": 7859
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.342111267914918e-06,
+      "loss": 1.4173,
+      "step": 7860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.32979214915873e-06,
+      "loss": 1.0009,
+      "step": 7861
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.317486862088183e-06,
+      "loss": 0.7819,
+      "step": 7862
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.30519540850114e-06,
+      "loss": 1.4263,
+      "step": 7863
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.292917790193408e-06,
+      "loss": 1.0608,
+      "step": 7864
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2806540089587745e-06,
+      "loss": 1.206,
+      "step": 7865
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.268404066589072e-06,
+      "loss": 1.3808,
+      "step": 7866
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.256167964874048e-06,
+      "loss": 1.3878,
+      "step": 7867
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.243945705601427e-06,
+      "loss": 1.0603,
+      "step": 7868
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2317372905569485e-06,
+      "loss": 1.317,
+      "step": 7869
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.219542721524307e-06,
+      "loss": 0.5425,
+      "step": 7870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.207362000285165e-06,
+      "loss": 0.5281,
+      "step": 7871
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.195195128619179e-06,
+      "loss": 1.1685,
+      "step": 7872
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1830421083039815e-06,
+      "loss": 1.104,
+      "step": 7873
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.170902941115175e-06,
+      "loss": 0.6291,
+      "step": 7874
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.158777628826339e-06,
+      "loss": 1.4129,
+      "step": 7875
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.146666173209014e-06,
+      "loss": 1.3309,
+      "step": 7876
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.134568576032761e-06,
+      "loss": 1.2677,
+      "step": 7877
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.122484839065045e-06,
+      "loss": 1.7529,
+      "step": 7878
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.110414964071397e-06,
+      "loss": 0.7822,
+      "step": 7879
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.098358952815219e-06,
+      "loss": 1.383,
+      "step": 7880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.086316807057978e-06,
+      "loss": 1.3251,
+      "step": 7881
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.074288528559068e-06,
+      "loss": 0.5835,
+      "step": 7882
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.062274119075872e-06,
+      "loss": 0.7074,
+      "step": 7883
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.050273580363729e-06,
+      "loss": 1.3139,
+      "step": 7884
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.038286914175982e-06,
+      "loss": 0.8678,
+      "step": 7885
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.02631412226392e-06,
+      "loss": 1.2046,
+      "step": 7886
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.014355206376798e-06,
+      "loss": 1.2773,
+      "step": 7887
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.002410168261895e-06,
+      "loss": 0.5216,
+      "step": 7888
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.990479009664406e-06,
+      "loss": 1.2364,
+      "step": 7889
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9785617323275245e-06,
+      "loss": 1.3837,
+      "step": 7890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.966658337992403e-06,
+      "loss": 1.3426,
+      "step": 7891
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.954768828398182e-06,
+      "loss": 1.4615,
+      "step": 7892
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.942893205281962e-06,
+      "loss": 1.395,
+      "step": 7893
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.931031470378822e-06,
+      "loss": 0.4483,
+      "step": 7894
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9191836254218085e-06,
+      "loss": 1.4418,
+      "step": 7895
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.907349672141926e-06,
+      "loss": 1.0214,
+      "step": 7896
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.895529612268179e-06,
+      "loss": 1.203,
+      "step": 7897
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8837234475275105e-06,
+      "loss": 1.061,
+      "step": 7898
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.87193117964484e-06,
+      "loss": 1.4205,
+      "step": 7899
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.860152810343099e-06,
+      "loss": 1.1519,
+      "step": 7900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.848388341343113e-06,
+      "loss": 1.3218,
+      "step": 7901
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.836637774363728e-06,
+      "loss": 1.546,
+      "step": 7902
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.82490111112176e-06,
+      "loss": 1.0535,
+      "step": 7903
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8131783533319815e-06,
+      "loss": 1.511,
+      "step": 7904
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.801469502707101e-06,
+      "loss": 1.6782,
+      "step": 7905
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.789774560957861e-06,
+      "loss": 0.8963,
+      "step": 7906
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.778093529792938e-06,
+      "loss": 1.452,
+      "step": 7907
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.766426410918967e-06,
+      "loss": 1.1675,
+      "step": 7908
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.754773206040531e-06,
+      "loss": 1.2846,
+      "step": 7909
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7431339168602655e-06,
+      "loss": 0.8783,
+      "step": 7910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.731508545078689e-06,
+      "loss": 1.2676,
+      "step": 7911
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7198970923942874e-06,
+      "loss": 1.2338,
+      "step": 7912
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.708299560503571e-06,
+      "loss": 0.636,
+      "step": 7913
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.696715951100983e-06,
+      "loss": 0.7814,
+      "step": 7914
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.685146265878915e-06,
+      "loss": 0.6251,
+      "step": 7915
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.673590506527759e-06,
+      "loss": 0.9349,
+      "step": 7916
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.662048674735864e-06,
+      "loss": 1.2785,
+      "step": 7917
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.650520772189504e-06,
+      "loss": 0.7416,
+      "step": 7918
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.639006800572998e-06,
+      "loss": 1.0382,
+      "step": 7919
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.627506761568534e-06,
+      "loss": 1.3304,
+      "step": 7920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.616020656856324e-06,
+      "loss": 1.0214,
+      "step": 7921
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.604548488114568e-06,
+      "loss": 0.9703,
+      "step": 7922
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.59309025701935e-06,
+      "loss": 0.7349,
+      "step": 7923
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.581645965244785e-06,
+      "loss": 1.0663,
+      "step": 7924
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.570215614462936e-06,
+      "loss": 0.485,
+      "step": 7925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.558799206343811e-06,
+      "loss": 1.2045,
+      "step": 7926
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.547396742555377e-06,
+      "loss": 1.7537,
+      "step": 7927
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.536008224763588e-06,
+      "loss": 0.9163,
+      "step": 7928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.524633654632371e-06,
+      "loss": 1.3332,
+      "step": 7929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.513273033823584e-06,
+      "loss": 1.359,
+      "step": 7930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.501926363997011e-06,
+      "loss": 0.7248,
+      "step": 7931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.490593646810526e-06,
+      "loss": 0.7501,
+      "step": 7932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.479274883919837e-06,
+      "loss": 1.0744,
+      "step": 7933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.467970076978656e-06,
+      "loss": 1.4644,
+      "step": 7934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4566792276386715e-06,
+      "loss": 1.0777,
+      "step": 7935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.44540233754952e-06,
+      "loss": 1.1178,
+      "step": 7936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.434139408358795e-06,
+      "loss": 1.4554,
+      "step": 7937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.422890441712046e-06,
+      "loss": 1.0411,
+      "step": 7938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.411655439252826e-06,
+      "loss": 0.7654,
+      "step": 7939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.400434402622566e-06,
+      "loss": 0.6408,
+      "step": 7940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.389227333460744e-06,
+      "loss": 1.6206,
+      "step": 7941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.378034233404715e-06,
+      "loss": 1.3486,
+      "step": 7942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.366855104089862e-06,
+      "loss": 1.5784,
+      "step": 7943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.355689947149499e-06,
+      "loss": 0.9404,
+      "step": 7944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.344538764214889e-06,
+      "loss": 0.7605,
+      "step": 7945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.333401556915262e-06,
+      "loss": 0.6539,
+      "step": 7946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3222783268778375e-06,
+      "loss": 1.319,
+      "step": 7947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3111690757277276e-06,
+      "loss": 1.808,
+      "step": 7948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.300073805088034e-06,
+      "loss": 0.915,
+      "step": 7949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2889925165798376e-06,
+      "loss": 1.0463,
+      "step": 7950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.277925211822164e-06,
+      "loss": 0.6993,
+      "step": 7951
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.266871892431967e-06,
+      "loss": 0.6849,
+      "step": 7952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2558325600241865e-06,
+      "loss": 0.6207,
+      "step": 7953
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.244807216211732e-06,
+      "loss": 0.7532,
+      "step": 7954
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2337958626054274e-06,
+      "loss": 0.8323,
+      "step": 7955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.222798500814074e-06,
+      "loss": 1.1775,
+      "step": 7956
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.21181513244443e-06,
+      "loss": 0.6222,
+      "step": 7957
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.200845759101235e-06,
+      "loss": 1.5108,
+      "step": 7958
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.189890382387118e-06,
+      "loss": 1.3824,
+      "step": 7959
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.178949003902721e-06,
+      "loss": 1.1781,
+      "step": 7960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.168021625246632e-06,
+      "loss": 0.4779,
+      "step": 7961
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1571082480153625e-06,
+      "loss": 0.8287,
+      "step": 7962
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.146208873803415e-06,
+      "loss": 0.6966,
+      "step": 7963
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.135323504203215e-06,
+      "loss": 0.6998,
+      "step": 7964
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.124452140805168e-06,
+      "loss": 1.3183,
+      "step": 7965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.113594785197639e-06,
+      "loss": 1.04,
+      "step": 7966
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.10275143896689e-06,
+      "loss": 0.9666,
+      "step": 7967
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.09192210369721e-06,
+      "loss": 1.1112,
+      "step": 7968
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0811067809708e-06,
+      "loss": 1.4573,
+      "step": 7969
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.070305472367819e-06,
+      "loss": 1.503,
+      "step": 7970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.059518179466359e-06,
+      "loss": 1.4305,
+      "step": 7971
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.048744903842516e-06,
+      "loss": 1.2247,
+      "step": 7972
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.037985647070297e-06,
+      "loss": 0.7592,
+      "step": 7973
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.027240410721667e-06,
+      "loss": 1.2302,
+      "step": 7974
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.016509196366536e-06,
+      "loss": 1.3105,
+      "step": 7975
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.005792005572817e-06,
+      "loss": 1.1879,
+      "step": 7976
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.995088839906291e-06,
+      "loss": 1.5187,
+      "step": 7977
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.984399700930741e-06,
+      "loss": 0.5412,
+      "step": 7978
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.973724590207905e-06,
+      "loss": 0.5069,
+      "step": 7979
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.96306350929746e-06,
+      "loss": 1.568,
+      "step": 7980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.952416459757002e-06,
+      "loss": 1.4056,
+      "step": 7981
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.941783443142133e-06,
+      "loss": 0.7649,
+      "step": 7982
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.931164461006387e-06,
+      "loss": 1.4358,
+      "step": 7983
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.920559514901201e-06,
+      "loss": 0.6109,
+      "step": 7984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.909968606376047e-06,
+      "loss": 1.6314,
+      "step": 7985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.899391736978253e-06,
+      "loss": 1.2856,
+      "step": 7986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.88882890825315e-06,
+      "loss": 0.7936,
+      "step": 7987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.878280121744049e-06,
+      "loss": 1.1973,
+      "step": 7988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.867745378992116e-06,
+      "loss": 1.3082,
+      "step": 7989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.857224681536542e-06,
+      "loss": 0.6922,
+      "step": 7990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.846718030914454e-06,
+      "loss": 1.4004,
+      "step": 7991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.836225428660889e-06,
+      "loss": 1.4053,
+      "step": 7992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.825746876308855e-06,
+      "loss": 0.7849,
+      "step": 7993
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.81528237538935e-06,
+      "loss": 1.2571,
+      "step": 7994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8048319274312404e-06,
+      "loss": 0.8119,
+      "step": 7995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7943955339613814e-06,
+      "loss": 1.3635,
+      "step": 7996
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7839731965045757e-06,
+      "loss": 1.1005,
+      "step": 7997
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.773564916583583e-06,
+      "loss": 1.0295,
+      "step": 7998
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.763170695719065e-06,
+      "loss": 0.8677,
+      "step": 7999
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7527905354296845e-06,
+      "loss": 0.6322,
+      "step": 8000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.742424437231995e-06,
+      "loss": 1.3085,
+      "step": 8001
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.732072402640563e-06,
+      "loss": 1.2082,
+      "step": 8002
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7217344331678118e-06,
+      "loss": 0.4959,
+      "step": 8003
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7114105303241998e-06,
+      "loss": 1.2112,
+      "step": 8004
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7011006956180762e-06,
+      "loss": 1.4524,
+      "step": 8005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6908049305557355e-06,
+      "loss": 1.0358,
+      "step": 8006
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.680523236641453e-06,
+      "loss": 1.4738,
+      "step": 8007
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6702556153773937e-06,
+      "loss": 0.6944,
+      "step": 8008
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6600020682637237e-06,
+      "loss": 1.0235,
+      "step": 8009
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6497625967985226e-06,
+      "loss": 1.2404,
+      "step": 8010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.639537202477794e-06,
+      "loss": 0.8124,
+      "step": 8011
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6293258867955315e-06,
+      "loss": 1.1022,
+      "step": 8012
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.619128651243653e-06,
+      "loss": 0.587,
+      "step": 8013
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6089454973119996e-06,
+      "loss": 1.2802,
+      "step": 8014
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5987764264883484e-06,
+      "loss": 0.9288,
+      "step": 8015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5886214402584993e-06,
+      "loss": 0.672,
+      "step": 8016
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.578480540106088e-06,
+      "loss": 1.1196,
+      "step": 8017
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.568353727512741e-06,
+      "loss": 1.3785,
+      "step": 8018
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5582410039580407e-06,
+      "loss": 1.1418,
+      "step": 8019
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.548142370919505e-06,
+      "loss": 1.2319,
+      "step": 8020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5380578298725543e-06,
+      "loss": 1.1754,
+      "step": 8021
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5279873822905875e-06,
+      "loss": 1.0285,
+      "step": 8022
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.517931029644961e-06,
+      "loss": 1.3234,
+      "step": 8023
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5078887734049103e-06,
+      "loss": 1.088,
+      "step": 8024
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.497860615037685e-06,
+      "loss": 1.3226,
+      "step": 8025
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4878465560084117e-06,
+      "loss": 0.3466,
+      "step": 8026
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.477846597780199e-06,
+      "loss": 0.9694,
+      "step": 8027
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.467860741814055e-06,
+      "loss": 0.8647,
+      "step": 8028
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.45788898956898e-06,
+      "loss": 1.3381,
+      "step": 8029
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4479313425018755e-06,
+      "loss": 1.4508,
+      "step": 8030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.437987802067577e-06,
+      "loss": 1.2218,
+      "step": 8031
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4280583697189115e-06,
+      "loss": 1.4515,
+      "step": 8032
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4181430469065743e-06,
+      "loss": 1.0494,
+      "step": 8033
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.40824183507924e-06,
+      "loss": 0.9238,
+      "step": 8034
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3983547356835287e-06,
+      "loss": 1.2587,
+      "step": 8035
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3884817501639745e-06,
+      "loss": 0.7506,
+      "step": 8036
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.378622879963034e-06,
+      "loss": 0.594,
+      "step": 8037
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.368778126521166e-06,
+      "loss": 1.3709,
+      "step": 8038
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3589474912767203e-06,
+      "loss": 0.604,
+      "step": 8039
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3491309756659704e-06,
+      "loss": 1.0007,
+      "step": 8040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.339328581123158e-06,
+      "loss": 1.1109,
+      "step": 8041
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.32954030908047e-06,
+      "loss": 0.9611,
+      "step": 8042
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.319766160967974e-06,
+      "loss": 1.0871,
+      "step": 8043
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.310006138213728e-06,
+      "loss": 1.3736,
+      "step": 8044
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.300260242243736e-06,
+      "loss": 0.9359,
+      "step": 8045
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.290528474481858e-06,
+      "loss": 1.6779,
+      "step": 8046
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.280810836350001e-06,
+      "loss": 0.8667,
+      "step": 8047
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.271107329267897e-06,
+      "loss": 1.4341,
+      "step": 8048
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2614179546532898e-06,
+      "loss": 1.1669,
+      "step": 8049
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.251742713921857e-06,
+      "loss": 1.271,
+      "step": 8050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2420816084871575e-06,
+      "loss": 1.0104,
+      "step": 8051
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2324346397607175e-06,
+      "loss": 0.5703,
+      "step": 8052
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2228018091520098e-06,
+      "loss": 1.1406,
+      "step": 8053
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.21318311806843e-06,
+      "loss": 1.166,
+      "step": 8054
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2035785679152885e-06,
+      "loss": 1.0575,
+      "step": 8055
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.193988160095862e-06,
+      "loss": 0.4899,
+      "step": 8056
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1844118960113413e-06,
+      "loss": 1.0235,
+      "step": 8057
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.174849777060873e-06,
+      "loss": 1.2709,
+      "step": 8058
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.165301804641474e-06,
+      "loss": 1.5011,
+      "step": 8059
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.155767980148194e-06,
+      "loss": 1.2291,
+      "step": 8060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.146248304973942e-06,
+      "loss": 1.7157,
+      "step": 8061
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1367427805095608e-06,
+      "loss": 1.1416,
+      "step": 8062
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.127251408143872e-06,
+      "loss": 0.5518,
+      "step": 8063
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1177741892635893e-06,
+      "loss": 0.8999,
+      "step": 8064
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1083111252533602e-06,
+      "loss": 0.9646,
+      "step": 8065
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0988622174957906e-06,
+      "loss": 1.7372,
+      "step": 8066
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0894274673714085e-06,
+      "loss": 1.2092,
+      "step": 8067
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0800068762586454e-06,
+      "loss": 1.2415,
+      "step": 8068
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.070600445533911e-06,
+      "loss": 1.1605,
+      "step": 8069
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0612081765715063e-06,
+      "loss": 0.8475,
+      "step": 8070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0518300707436663e-06,
+      "loss": 1.1522,
+      "step": 8071
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.042466129420607e-06,
+      "loss": 0.8909,
+      "step": 8072
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0331163539704e-06,
+      "loss": 1.2555,
+      "step": 8073
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.023780745759097e-06,
+      "loss": 1.3553,
+      "step": 8074
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.014459306150674e-06,
+      "loss": 1.3122,
+      "step": 8075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0051520365070196e-06,
+      "loss": 1.0511,
+      "step": 8076
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9958589381879675e-06,
+      "loss": 0.7556,
+      "step": 8077
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9865800125512544e-06,
+      "loss": 1.1576,
+      "step": 8078
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.977315260952618e-06,
+      "loss": 1.0321,
+      "step": 8079
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9680646847456306e-06,
+      "loss": 1.2133,
+      "step": 8080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9588282852818226e-06,
+      "loss": 1.6809,
+      "step": 8081
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9496060639107258e-06,
+      "loss": 0.9542,
+      "step": 8082
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.940398021979707e-06,
+      "loss": 1.1809,
+      "step": 8083
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9312041608341003e-06,
+      "loss": 1.5164,
+      "step": 8084
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9220244818171538e-06,
+      "loss": 0.7729,
+      "step": 8085
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.912858986270084e-06,
+      "loss": 1.1751,
+      "step": 8086
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9037076755319746e-06,
+      "loss": 1.2103,
+      "step": 8087
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.894570550939879e-06,
+      "loss": 0.9252,
+      "step": 8088
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.885447613828773e-06,
+      "loss": 1.0075,
+      "step": 8089
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8763388655315472e-06,
+      "loss": 1.4862,
+      "step": 8090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8672443073790355e-06,
+      "loss": 0.6341,
+      "step": 8091
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8581639406999762e-06,
+      "loss": 0.8718,
+      "step": 8092
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8490977668210406e-06,
+      "loss": 0.8418,
+      "step": 8093
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8400457870668475e-06,
+      "loss": 0.6513,
+      "step": 8094
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.831008002759916e-06,
+      "loss": 1.3441,
+      "step": 8095
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8219844152207022e-06,
+      "loss": 1.1459,
+      "step": 8096
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8129750257676058e-06,
+      "loss": 1.1365,
+      "step": 8097
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.803979835716908e-06,
+      "loss": 1.3923,
+      "step": 8098
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.794998846382846e-06,
+      "loss": 1.0725,
+      "step": 8099
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7860320590775925e-06,
+      "loss": 0.6674,
+      "step": 8100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7770794751112216e-06,
+      "loss": 1.1216,
+      "step": 8101
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7681410957917318e-06,
+      "loss": 1.0449,
+      "step": 8102
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7592169224250563e-06,
+      "loss": 1.1293,
+      "step": 8103
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7503069563150628e-06,
+      "loss": 0.3496,
+      "step": 8104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7414111987635327e-06,
+      "loss": 0.9417,
+      "step": 8105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.732529651070159e-06,
+      "loss": 1.3454,
+      "step": 8106
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7236623145325713e-06,
+      "loss": 1.1955,
+      "step": 8107
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7148091904463435e-06,
+      "loss": 0.9004,
+      "step": 8108
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.705970280104919e-06,
+      "loss": 1.1042,
+      "step": 8109
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6971455847997094e-06,
+      "loss": 0.8007,
+      "step": 8110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6883351058200612e-06,
+      "loss": 0.7448,
+      "step": 8111
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6795388444531777e-06,
+      "loss": 0.6453,
+      "step": 8112
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6707568019842644e-06,
+      "loss": 0.5979,
+      "step": 8113
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.661988979696384e-06,
+      "loss": 1.2683,
+      "step": 8114
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6532353788705666e-06,
+      "loss": 1.8064,
+      "step": 8115
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6444960007857456e-06,
+      "loss": 1.3328,
+      "step": 8116
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6357708467187656e-06,
+      "loss": 0.9914,
+      "step": 8117
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.627059917944408e-06,
+      "loss": 1.2247,
+      "step": 8118
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6183632157353977e-06,
+      "loss": 0.8343,
+      "step": 8119
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6096807413623303e-06,
+      "loss": 0.5049,
+      "step": 8120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.601012496093758e-06,
+      "loss": 0.9198,
+      "step": 8121
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.592358481196133e-06,
+      "loss": 1.2795,
+      "step": 8122
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5837186979338677e-06,
+      "loss": 1.1144,
+      "step": 8123
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.575093147569252e-06,
+      "loss": 0.8569,
+      "step": 8124
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5664818313624994e-06,
+      "loss": 0.8996,
+      "step": 8125
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5578847505717816e-06,
+      "loss": 0.662,
+      "step": 8126
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5493019064531607e-06,
+      "loss": 0.8354,
+      "step": 8127
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5407333002606005e-06,
+      "loss": 1.2606,
+      "step": 8128
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5321789332460324e-06,
+      "loss": 1.3878,
+      "step": 8129
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5236388066592786e-06,
+      "loss": 1.1845,
+      "step": 8130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.515112921748064e-06,
+      "loss": 0.887,
+      "step": 8131
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.506601279758081e-06,
+      "loss": 0.9836,
+      "step": 8132
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4981038819329118e-06,
+      "loss": 0.8669,
+      "step": 8133
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4896207295140304e-06,
+      "loss": 1.5349,
+      "step": 8134
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4811518237409013e-06,
+      "loss": 0.5375,
+      "step": 8135
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4726971658508237e-06,
+      "loss": 1.7607,
+      "step": 8136
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.464256757079075e-06,
+      "loss": 0.8968,
+      "step": 8137
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.455830598658848e-06,
+      "loss": 1.2744,
+      "step": 8138
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4474186918212016e-06,
+      "loss": 0.4668,
+      "step": 8139
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.439021037795164e-06,
+      "loss": 1.1081,
+      "step": 8140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4306376378076867e-06,
+      "loss": 1.2777,
+      "step": 8141
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4222684930835903e-06,
+      "loss": 1.461,
+      "step": 8142
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.41391360484563e-06,
+      "loss": 1.2096,
+      "step": 8143
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4055729743145185e-06,
+      "loss": 0.6582,
+      "step": 8144
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.397246602708847e-06,
+      "loss": 1.0201,
+      "step": 8145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3889344912451093e-06,
+      "loss": 1.3376,
+      "step": 8146
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3806366411377567e-06,
+      "loss": 1.0657,
+      "step": 8147
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3723530535991413e-06,
+      "loss": 0.8444,
+      "step": 8148
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.364083729839506e-06,
+      "loss": 0.969,
+      "step": 8149
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3558286710670506e-06,
+      "loss": 1.0594,
+      "step": 8150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.347587878487867e-06,
+      "loss": 0.7713,
+      "step": 8151
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.339361353305969e-06,
+      "loss": 0.9852,
+      "step": 8152
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3311490967232618e-06,
+      "loss": 1.3632,
+      "step": 8153
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3229511099396196e-06,
+      "loss": 1.0642,
+      "step": 8154
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3147673941527835e-06,
+      "loss": 1.3824,
+      "step": 8155
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.30659795055842e-06,
+      "loss": 0.9629,
+      "step": 8156
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2984427803501406e-06,
+      "loss": 1.3671,
+      "step": 8157
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2903018847194257e-06,
+      "loss": 1.2309,
+      "step": 8158
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2821752648557015e-06,
+      "loss": 0.688,
+      "step": 8159
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.274062921946296e-06,
+      "loss": 0.7193,
+      "step": 8160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2659648571764504e-06,
+      "loss": 0.5007,
+      "step": 8161
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2578810717293175e-06,
+      "loss": 1.1654,
+      "step": 8162
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2498115667859977e-06,
+      "loss": 1.0472,
+      "step": 8163
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2417563435254586e-06,
+      "loss": 0.4999,
+      "step": 8164
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.233715403124581e-06,
+      "loss": 1.046,
+      "step": 8165
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2256887467582145e-06,
+      "loss": 1.5933,
+      "step": 8166
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.217676375599065e-06,
+      "loss": 1.5434,
+      "step": 8167
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2096782908177625e-06,
+      "loss": 1.5304,
+      "step": 8168
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.201694493582862e-06,
+      "loss": 1.0839,
+      "step": 8169
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1937249850608523e-06,
+      "loss": 1.5215,
+      "step": 8170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1857697664160792e-06,
+      "loss": 1.0946,
+      "step": 8171
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1778288388108582e-06,
+      "loss": 1.5506,
+      "step": 8172
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1699022034053606e-06,
+      "loss": 1.1249,
+      "step": 8173
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.161989861357716e-06,
+      "loss": 1.1711,
+      "step": 8174
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.154091813823944e-06,
+      "loss": 1.0733,
+      "step": 8175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.146208061957977e-06,
+      "loss": 1.2238,
+      "step": 8176
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1383386069116827e-06,
+      "loss": 0.8722,
+      "step": 8177
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.130483449834786e-06,
+      "loss": 1.1169,
+      "step": 8178
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1226425918749794e-06,
+      "loss": 1.0678,
+      "step": 8179
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1148160341778354e-06,
+      "loss": 1.0461,
+      "step": 8180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.107003777886829e-06,
+      "loss": 1.1915,
+      "step": 8181
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.099205824143402e-06,
+      "loss": 0.6013,
+      "step": 8182
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.091422174086821e-06,
+      "loss": 1.1682,
+      "step": 8183
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0836528288543323e-06,
+      "loss": 1.1876,
+      "step": 8184
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0758977895810716e-06,
+      "loss": 1.3205,
+      "step": 8185
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0681570574000663e-06,
+      "loss": 1.1232,
+      "step": 8186
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0604306334422565e-06,
+      "loss": 1.2703,
+      "step": 8187
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0527185188365385e-06,
+      "loss": 0.8129,
+      "step": 8188
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0450207147096665e-06,
+      "loss": 1.3141,
+      "step": 8189
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0373372221862974e-06,
+      "loss": 0.9882,
+      "step": 8190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0296680423890436e-06,
+      "loss": 1.1669,
+      "step": 8191
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.022013176438398e-06,
+      "loss": 1.178,
+      "step": 8192
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.014372625452765e-06,
+      "loss": 1.1935,
+      "step": 8193
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.006746390548453e-06,
+      "loss": 1.4054,
+      "step": 8194
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9991344728396923e-06,
+      "loss": 0.464,
+      "step": 8195
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9915368734386154e-06,
+      "loss": 0.8267,
+      "step": 8196
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9839535934552566e-06,
+      "loss": 0.7653,
+      "step": 8197
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9763846339975523e-06,
+      "loss": 1.5353,
+      "step": 8198
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.968829996171373e-06,
+      "loss": 1.6189,
+      "step": 8199
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9612896810804803e-06,
+      "loss": 0.9659,
+      "step": 8200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.953763689826538e-06,
+      "loss": 1.1846,
+      "step": 8201
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9462520235091096e-06,
+      "loss": 1.2666,
+      "step": 8202
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9387546832257076e-06,
+      "loss": 1.1486,
+      "step": 8203
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.931271670071699e-06,
+      "loss": 0.5501,
+      "step": 8204
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9238029851403883e-06,
+      "loss": 0.6902,
+      "step": 8205
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.916348629522979e-06,
+      "loss": 1.371,
+      "step": 8206
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.90890860430859e-06,
+      "loss": 0.8041,
+      "step": 8207
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9014829105842403e-06,
+      "loss": 0.7649,
+      "step": 8208
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8940715494348172e-06,
+      "loss": 0.5479,
+      "step": 8209
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.886674521943188e-06,
+      "loss": 1.1121,
+      "step": 8210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8792918291900774e-06,
+      "loss": 0.9981,
+      "step": 8211
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8719234722541224e-06,
+      "loss": 1.2047,
+      "step": 8212
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8645694522118618e-06,
+      "loss": 0.9673,
+      "step": 8213
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8572297701377582e-06,
+      "loss": 1.5546,
+      "step": 8214
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8499044271041655e-06,
+      "loss": 0.5894,
+      "step": 8215
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8425934241813381e-06,
+      "loss": 0.8399,
+      "step": 8216
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8352967624374552e-06,
+      "loss": 0.4527,
+      "step": 8217
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8280144429385637e-06,
+      "loss": 0.9117,
+      "step": 8218
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8207464667486685e-06,
+      "loss": 0.8266,
+      "step": 8219
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8134928349296199e-06,
+      "loss": 0.8488,
+      "step": 8220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8062535485412147e-06,
+      "loss": 0.6807,
+      "step": 8221
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7990286086411623e-06,
+      "loss": 1.2468,
+      "step": 8222
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7918180162850184e-06,
+      "loss": 0.9492,
+      "step": 8223
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.784621772526296e-06,
+      "loss": 0.8016,
+      "step": 8224
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7774398784163982e-06,
+      "loss": 0.8141,
+      "step": 8225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.770272335004619e-06,
+      "loss": 0.609,
+      "step": 8226
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7631191433381656e-06,
+      "loss": 0.4796,
+      "step": 8227
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.755980304462157e-06,
+      "loss": 1.362,
+      "step": 8228
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7488558194195926e-06,
+      "loss": 0.7504,
+      "step": 8229
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.741745689251395e-06,
+      "loss": 1.3963,
+      "step": 8230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7346499149963669e-06,
+      "loss": 1.3805,
+      "step": 8231
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7275684976912566e-06,
+      "loss": 0.6998,
+      "step": 8232
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7205014383706697e-06,
+      "loss": 1.2251,
+      "step": 8233
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.713448738067125e-06,
+      "loss": 0.5609,
+      "step": 8234
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7064103978110535e-06,
+      "loss": 0.5639,
+      "step": 8235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6993864186307995e-06,
+      "loss": 1.2637,
+      "step": 8236
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6923768015525754e-06,
+      "loss": 1.2144,
+      "step": 8237
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6853815476005174e-06,
+      "loss": 1.5009,
+      "step": 8238
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.678400657796675e-06,
+      "loss": 1.2996,
+      "step": 8239
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.671434133160954e-06,
+      "loss": 1.1994,
+      "step": 8240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6644819747112184e-06,
+      "loss": 1.4294,
+      "step": 8241
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6575441834631778e-06,
+      "loss": 0.8112,
+      "step": 8242
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.650620760430499e-06,
+      "loss": 1.521,
+      "step": 8243
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6437117066247064e-06,
+      "loss": 0.5487,
+      "step": 8244
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.636817023055237e-06,
+      "loss": 0.464,
+      "step": 8245
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6299367107294404e-06,
+      "loss": 1.2866,
+      "step": 8246
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6230707706525572e-06,
+      "loss": 1.2187,
+      "step": 8247
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.616219203827718e-06,
+      "loss": 0.8619,
+      "step": 8248
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6093820112559667e-06,
+      "loss": 1.4968,
+      "step": 8249
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6025591939362484e-06,
+      "loss": 1.17,
+      "step": 8250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5957507528653993e-06,
+      "loss": 1.0916,
+      "step": 8251
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5889566890381568e-06,
+      "loss": 1.4015,
+      "step": 8252
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5821770034471718e-06,
+      "loss": 1.126,
+      "step": 8253
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5754116970829736e-06,
+      "loss": 1.3455,
+      "step": 8254
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.568660770933994e-06,
+      "loss": 1.3786,
+      "step": 8255
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5619242259865773e-06,
+      "loss": 0.4432,
+      "step": 8256
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5552020632249587e-06,
+      "loss": 0.9404,
+      "step": 8257
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5484942836312744e-06,
+      "loss": 0.7477,
+      "step": 8258
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5418008881855517e-06,
+      "loss": 1.661,
+      "step": 8259
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5351218778657195e-06,
+      "loss": 0.7261,
+      "step": 8260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5284572536476195e-06,
+      "loss": 1.6013,
+      "step": 8261
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5218070165049725e-06,
+      "loss": 1.2563,
+      "step": 8262
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5151711674094015e-06,
+      "loss": 0.7923,
+      "step": 8263
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.50854970733042e-06,
+      "loss": 0.667,
+      "step": 8264
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.501942637235465e-06,
+      "loss": 1.0534,
+      "step": 8265
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4953499580898645e-06,
+      "loss": 1.4,
+      "step": 8266
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4887716708568035e-06,
+      "loss": 1.078,
+      "step": 8267
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4822077764974241e-06,
+      "loss": 1.4072,
+      "step": 8268
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.475658275970726e-06,
+      "loss": 0.5934,
+      "step": 8269
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4691231702336107e-06,
+      "loss": 1.3422,
+      "step": 8270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4626024602408917e-06,
+      "loss": 0.8509,
+      "step": 8271
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4560961469452516e-06,
+      "loss": 0.7568,
+      "step": 8272
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.449604231297319e-06,
+      "loss": 1.2001,
+      "step": 8273
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4431267142455685e-06,
+      "loss": 1.0702,
+      "step": 8274
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4366635967363874e-06,
+      "loss": 1.2908,
+      "step": 8275
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4302148797140758e-06,
+      "loss": 0.9068,
+      "step": 8276
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4237805641208024e-06,
+      "loss": 0.6069,
+      "step": 8277
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4173606508966597e-06,
+      "loss": 1.059,
+      "step": 8278
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4109551409796195e-06,
+      "loss": 0.4623,
+      "step": 8279
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4045640353055334e-06,
+      "loss": 1.1348,
+      "step": 8280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3981873348081876e-06,
+      "loss": 1.2045,
+      "step": 8281
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3918250404192368e-06,
+      "loss": 1.2584,
+      "step": 8282
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3854771530682375e-06,
+      "loss": 1.3967,
+      "step": 8283
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3791436736826257e-06,
+      "loss": 0.8702,
+      "step": 8284
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3728246031877834e-06,
+      "loss": 1.0357,
+      "step": 8285
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3665199425069053e-06,
+      "loss": 0.9913,
+      "step": 8286
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3602296925611547e-06,
+      "loss": 0.7484,
+      "step": 8287
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.353953854269574e-06,
+      "loss": 1.5352,
+      "step": 8288
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3476924285490521e-06,
+      "loss": 1.5457,
+      "step": 8289
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3414454163144351e-06,
+      "loss": 0.8568,
+      "step": 8290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3352128184784263e-06,
+      "loss": 1.2298,
+      "step": 8291
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3289946359516413e-06,
+      "loss": 0.9214,
+      "step": 8292
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.322790869642554e-06,
+      "loss": 0.5941,
+      "step": 8293
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3166015204575943e-06,
+      "loss": 1.313,
+      "step": 8294
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.310426589301028e-06,
+      "loss": 1.3805,
+      "step": 8295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3042660770750448e-06,
+      "loss": 1.1824,
+      "step": 8296
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.298119984679702e-06,
+      "loss": 0.8439,
+      "step": 8297
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.291988313013004e-06,
+      "loss": 0.6113,
+      "step": 8298
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2858710629707782e-06,
+      "loss": 1.471,
+      "step": 8299
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2797682354467877e-06,
+      "loss": 1.156,
+      "step": 8300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.273679831332686e-06,
+      "loss": 0.5277,
+      "step": 8301
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2676058515180056e-06,
+      "loss": 1.3438,
+      "step": 8302
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2615462968901926e-06,
+      "loss": 1.6187,
+      "step": 8303
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2555011683345497e-06,
+      "loss": 1.3028,
+      "step": 8304
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.249470466734315e-06,
+      "loss": 1.4273,
+      "step": 8305
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2434541929705724e-06,
+      "loss": 0.5889,
+      "step": 8306
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.237452347922352e-06,
+      "loss": 1.2926,
+      "step": 8307
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2314649324665195e-06,
+      "loss": 1.4066,
+      "step": 8308
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2254919474778747e-06,
+      "loss": 0.8847,
+      "step": 8309
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2195333938290975e-06,
+      "loss": 0.5849,
+      "step": 8310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.213589272390747e-06,
+      "loss": 1.5719,
+      "step": 8311
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2076595840312843e-06,
+      "loss": 1.1029,
+      "step": 8312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2017443296170716e-06,
+      "loss": 1.0306,
+      "step": 8313
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.195843510012329e-06,
+      "loss": 1.2359,
+      "step": 8314
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1899571260791996e-06,
+      "loss": 1.274,
+      "step": 8315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1840851786777185e-06,
+      "loss": 1.3645,
+      "step": 8316
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1782276686657768e-06,
+      "loss": 1.1189,
+      "step": 8317
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1723845968992009e-06,
+      "loss": 1.2268,
+      "step": 8318
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1665559642316636e-06,
+      "loss": 1.6088,
+      "step": 8319
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1607417715147838e-06,
+      "loss": 0.8958,
+      "step": 8320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1549420195980042e-06,
+      "loss": 1.1334,
+      "step": 8321
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1491567093287137e-06,
+      "loss": 1.3539,
+      "step": 8322
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.143385841552158e-06,
+      "loss": 1.1995,
+      "step": 8323
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1376294171114742e-06,
+      "loss": 1.3041,
+      "step": 8324
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1318874368477227e-06,
+      "loss": 0.7074,
+      "step": 8325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1261599015998103e-06,
+      "loss": 1.1432,
+      "step": 8326
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1204468122045564e-06,
+      "loss": 1.2052,
+      "step": 8327
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1147481694966821e-06,
+      "loss": 1.1616,
+      "step": 8328
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.109063974308755e-06,
+      "loss": 1.2975,
+      "step": 8329
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.103394227471266e-06,
+      "loss": 1.5629,
+      "step": 8330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0977389298125972e-06,
+      "loss": 0.9149,
+      "step": 8331
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0920980821589988e-06,
+      "loss": 1.2251,
+      "step": 8332
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0864716853346335e-06,
+      "loss": 0.6987,
+      "step": 8333
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0808597401615217e-06,
+      "loss": 0.5072,
+      "step": 8334
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0752622474595963e-06,
+      "loss": 0.7123,
+      "step": 8335
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0696792080466921e-06,
+      "loss": 1.1662,
+      "step": 8336
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.064110622738479e-06,
+      "loss": 0.76,
+      "step": 8337
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0585564923485725e-06,
+      "loss": 0.9829,
+      "step": 8338
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.053016817688457e-06,
+      "loss": 1.1364,
+      "step": 8339
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.047491599567485e-06,
+      "loss": 1.0693,
+      "step": 8340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0419808387929108e-06,
+      "loss": 1.1825,
+      "step": 8341
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0364845361698904e-06,
+      "loss": 1.2541,
+      "step": 8342
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0310026925014481e-06,
+      "loss": 0.8081,
+      "step": 8343
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.02553530858851e-06,
+      "loss": 1.4703,
+      "step": 8344
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0200823852298703e-06,
+      "loss": 1.2085,
+      "step": 8345
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.014643923222236e-06,
+      "loss": 1.0127,
+      "step": 8346
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0092199233601717e-06,
+      "loss": 1.2088,
+      "step": 8347
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0038103864361549e-06,
+      "loss": 1.335,
+      "step": 8348
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.984153132405417e-07,
+      "loss": 1.0329,
+      "step": 8349
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.930347045615684e-07,
+      "loss": 1.1401,
+      "step": 8350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.87668561185373e-07,
+      "loss": 1.1181,
+      "step": 8351
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.823168838959396e-07,
+      "loss": 0.6667,
+      "step": 8352
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.76979673475209e-07,
+      "loss": 0.5362,
+      "step": 8353
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.716569307029467e-07,
+      "loss": 0.9769,
+      "step": 8354
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.663486563568303e-07,
+      "loss": 1.3335,
+      "step": 8355
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.610548512124062e-07,
+      "loss": 1.3488,
+      "step": 8356
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.557755160431448e-07,
+      "loss": 1.0379,
+      "step": 8357
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.505106516203732e-07,
+      "loss": 1.0035,
+      "step": 8358
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.452602587132764e-07,
+      "loss": 0.5009,
+      "step": 8359
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.400243380889961e-07,
+      "loss": 1.0712,
+      "step": 8360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.348028905125095e-07,
+      "loss": 1.2739,
+      "step": 8361
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.295959167466839e-07,
+      "loss": 1.1927,
+      "step": 8362
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.244034175522776e-07,
+      "loss": 0.956,
+      "step": 8363
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.192253936879391e-07,
+      "loss": 0.8316,
+      "step": 8364
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.140618459101969e-07,
+      "loss": 1.4359,
+      "step": 8365
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.089127749734583e-07,
+      "loss": 0.9089,
+      "step": 8366
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.037781816300439e-07,
+      "loss": 1.4786,
+      "step": 8367
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.986580666301203e-07,
+      "loss": 1.0009,
+      "step": 8368
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.935524307217557e-07,
+      "loss": 0.9064,
+      "step": 8369
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.884612746509091e-07,
+      "loss": 0.6116,
+      "step": 8370
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.833845991614187e-07,
+      "loss": 1.2487,
+      "step": 8371
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.783224049950023e-07,
+      "loss": 1.3437,
+      "step": 8372
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.732746928912682e-07,
+      "loss": 1.1737,
+      "step": 8373
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.682414635877046e-07,
+      "loss": 0.9547,
+      "step": 8374
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.632227178197005e-07,
+      "loss": 0.6089,
+      "step": 8375
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.58218456320492e-07,
+      "loss": 1.6526,
+      "step": 8376
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.532286798212275e-07,
+      "loss": 0.9382,
+      "step": 8377
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.482533890509347e-07,
+      "loss": 0.8501,
+      "step": 8378
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.432925847365436e-07,
+      "loss": 0.9676,
+      "step": 8379
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.383462676028186e-07,
+      "loss": 1.4585,
+      "step": 8380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.334144383724263e-07,
+      "loss": 1.0945,
+      "step": 8381
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.284970977659678e-07,
+      "loss": 1.2342,
+      "step": 8382
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.235942465018575e-07,
+      "loss": 1.0506,
+      "step": 8383
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.187058852964224e-07,
+      "loss": 1.3191,
+      "step": 8384
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.1383201486388e-07,
+      "loss": 0.7161,
+      "step": 8385
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.089726359163164e-07,
+      "loss": 1.1731,
+      "step": 8386
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.041277491636967e-07,
+      "loss": 0.8417,
+      "step": 8387
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.992973553138882e-07,
+      "loss": 1.1825,
+      "step": 8388
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.944814550726487e-07,
+      "loss": 1.325,
+      "step": 8389
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.896800491435597e-07,
+      "loss": 1.5024,
+      "step": 8390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.8489313822816e-07,
+      "loss": 1.1637,
+      "step": 8391
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.801207230258123e-07,
+      "loss": 1.3243,
+      "step": 8392
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.753628042337924e-07,
+      "loss": 1.0223,
+      "step": 8393
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.706193825472663e-07,
+      "loss": 0.7818,
+      "step": 8394
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.658904586592463e-07,
+      "loss": 0.9612,
+      "step": 8395
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.611760332606465e-07,
+      "loss": 1.3313,
+      "step": 8396
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.564761070402826e-07,
+      "loss": 1.3722,
+      "step": 8397
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.517906806848163e-07,
+      "loss": 1.3651,
+      "step": 8398
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.471197548788001e-07,
+      "loss": 1.2669,
+      "step": 8399
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.42463330304688e-07,
+      "loss": 1.415,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.378214076428136e-07,
+      "loss": 0.8684,
+      "step": 8401
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.331939875713456e-07,
+      "loss": 1.4111,
+      "step": 8402
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.285810707663876e-07,
+      "loss": 1.4156,
+      "step": 8403
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.239826579019115e-07,
+      "loss": 1.2006,
+      "step": 8404
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.193987496497468e-07,
+      "loss": 1.4053,
+      "step": 8405
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.148293466796352e-07,
+      "loss": 1.0059,
+      "step": 8406
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.102744496591762e-07,
+      "loss": 1.3313,
+      "step": 8407
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.057340592538598e-07,
+      "loss": 0.9439,
+      "step": 8408
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.01208176127055e-07,
+      "loss": 0.5502,
+      "step": 8409
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.966968009399999e-07,
+      "loss": 1.2095,
+      "step": 8410
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.921999343518449e-07,
+      "loss": 0.6147,
+      "step": 8411
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.877175770195865e-07,
+      "loss": 1.2572,
+      "step": 8412
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.832497295981233e-07,
+      "loss": 1.2136,
+      "step": 8413
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.787963927402108e-07,
+      "loss": 0.7269,
+      "step": 8414
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.743575670965175e-07,
+      "loss": 0.599,
+      "step": 8415
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.699332533155577e-07,
+      "loss": 1.2748,
+      "step": 8416
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.65523452043737e-07,
+      "loss": 1.3104,
+      "step": 8417
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.611281639253619e-07,
+      "loss": 1.1856,
+      "step": 8418
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.567473896025967e-07,
+      "loss": 0.7227,
+      "step": 8419
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.523811297154847e-07,
+      "loss": 1.3672,
+      "step": 8420
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.480293849019492e-07,
+      "loss": 0.4607,
+      "step": 8421
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.436921557978037e-07,
+      "loss": 1.882,
+      "step": 8422
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.393694430367414e-07,
+      "loss": 1.2247,
+      "step": 8423
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.350612472503015e-07,
+      "loss": 1.1864,
+      "step": 8424
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.307675690679582e-07,
+      "loss": 1.4448,
+      "step": 8425
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.264884091170209e-07,
+      "loss": 1.4043,
+      "step": 8426
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.222237680226895e-07,
+      "loss": 0.7318,
+      "step": 8427
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.179736464080433e-07,
+      "loss": 1.1099,
+      "step": 8428
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.137380448940522e-07,
+      "loss": 1.2829,
+      "step": 8429
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.095169640995546e-07,
+      "loss": 1.2839,
+      "step": 8430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.053104046412461e-07,
+      "loss": 1.1601,
+      "step": 8431
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.01118367133735e-07,
+      "loss": 0.9434,
+      "step": 8432
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.969408521895092e-07,
+      "loss": 1.2684,
+      "step": 8433
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.927778604188916e-07,
+      "loss": 1.2363,
+      "step": 8434
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.886293924301289e-07,
+      "loss": 0.533,
+      "step": 8435
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.844954488293253e-07,
+      "loss": 1.3476,
+      "step": 8436
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.803760302204642e-07,
+      "loss": 0.9673,
+      "step": 8437
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.762711372054197e-07,
+      "loss": 1.3585,
+      "step": 8438
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.721807703839122e-07,
+      "loss": 0.6353,
+      "step": 8439
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.681049303535746e-07,
+      "loss": 0.5018,
+      "step": 8440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.640436177098973e-07,
+      "loss": 1.2875,
+      "step": 8441
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.599968330462612e-07,
+      "loss": 1.1607,
+      "step": 8442
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.559645769538934e-07,
+      "loss": 1.1735,
+      "step": 8443
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.519468500219671e-07,
+      "loss": 1.035,
+      "step": 8444
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.479436528374459e-07,
+      "loss": 1.4551,
+      "step": 8445
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.439549859852289e-07,
+      "loss": 1.1045,
+      "step": 8446
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.39980850048083e-07,
+      "loss": 0.7557,
+      "step": 8447
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.360212456066327e-07,
+      "loss": 1.65,
+      "step": 8448
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.32076173239393e-07,
+      "loss": 0.5651,
+      "step": 8449
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.281456335227586e-07,
+      "loss": 1.536,
+      "step": 8450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.242296270309921e-07,
+      "loss": 1.128,
+      "step": 8451
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.20328154336236e-07,
+      "loss": 1.2145,
+      "step": 8452
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.164412160085341e-07,
+      "loss": 1.1415,
+      "step": 8453
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.125688126157435e-07,
+      "loss": 0.6415,
+      "step": 8454
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.087109447236671e-07,
+      "loss": 0.9374,
+      "step": 8455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.048676128959428e-07,
+      "loss": 1.5632,
+      "step": 8456
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.010388176941105e-07,
+      "loss": 0.7129,
+      "step": 8457
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.972245596775449e-07,
+      "loss": 1.1821,
+      "step": 8458
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.934248394035445e-07,
+      "loss": 1.1585,
+      "step": 8459
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.896396574272655e-07,
+      "loss": 1.162,
+      "step": 8460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.858690143017208e-07,
+      "loss": 1.0764,
+      "step": 8461
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.821129105778255e-07,
+      "loss": 1.0066,
+      "step": 8462
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.783713468043627e-07,
+      "loss": 0.8551,
+      "step": 8463
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7464432352799516e-07,
+      "loss": 1.4511,
+      "step": 8464
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7093184129323175e-07,
+      "loss": 1.2776,
+      "step": 8465
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6723390064251636e-07,
+      "loss": 1.1953,
+      "step": 8466
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6355050211610573e-07,
+      "loss": 1.4474,
+      "step": 8467
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.598816462521582e-07,
+      "loss": 1.0689,
+      "step": 8468
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5622733358672285e-07,
+      "loss": 0.7814,
+      "step": 8469
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.525875646537059e-07,
+      "loss": 1.0288,
+      "step": 8470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4896233998488193e-07,
+      "loss": 0.8481,
+      "step": 8471
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.453516601099161e-07,
+      "loss": 1.385,
+      "step": 8472
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.417555255563421e-07,
+      "loss": 1.1568,
+      "step": 8473
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.38173936849573e-07,
+      "loss": 0.7564,
+      "step": 8474
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3460689451289005e-07,
+      "loss": 1.1028,
+      "step": 8475
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3105439906744315e-07,
+      "loss": 1.138,
+      "step": 8476
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2751645103228376e-07,
+      "loss": 1.1412,
+      "step": 8477
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2399305092430954e-07,
+      "loss": 1.3261,
+      "step": 8478
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.204841992582975e-07,
+      "loss": 0.9843,
+      "step": 8479
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1698989654691543e-07,
+      "loss": 1.0078,
+      "step": 8480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1351014330068827e-07,
+      "loss": 1.2385,
+      "step": 8481
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.100449400280204e-07,
+      "loss": 1.5471,
+      "step": 8482
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.065942872351958e-07,
+      "loss": 1.5748,
+      "step": 8483
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0315818542637775e-07,
+      "loss": 0.7419,
+      "step": 8484
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9973663510357585e-07,
+      "loss": 0.4942,
+      "step": 8485
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9632963676669024e-07,
+      "loss": 0.4922,
+      "step": 8486
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9293719091351153e-07,
+      "loss": 1.4311,
+      "step": 8487
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8955929803968783e-07,
+      "loss": 0.7036,
+      "step": 8488
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.861959586387354e-07,
+      "loss": 0.6601,
+      "step": 8489
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8284717320205e-07,
+      "loss": 1.3905,
+      "step": 8490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.79512942218907e-07,
+      "loss": 1.248,
+      "step": 8491
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7619326617645e-07,
+      "loss": 1.3019,
+      "step": 8492
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.728881455597022e-07,
+      "loss": 1.066,
+      "step": 8493
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6959758085153285e-07,
+      "loss": 1.3264,
+      "step": 8494
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.663215725327351e-07,
+      "loss": 1.4307,
+      "step": 8495
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6306012108192623e-07,
+      "loss": 0.8621,
+      "step": 8496
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.59813226975636e-07,
+      "loss": 0.528,
+      "step": 8497
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5658089068821843e-07,
+      "loss": 1.3217,
+      "step": 8498
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5336311269196233e-07,
+      "loss": 1.2747,
+      "step": 8499
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.501598934569805e-07,
+      "loss": 1.5014,
+      "step": 8500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4697123345127646e-07,
+      "loss": 1.3617,
+      "step": 8501
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4379713314072195e-07,
+      "loss": 1.0933,
+      "step": 8502
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4063759298907927e-07,
+      "loss": 1.0643,
+      "step": 8503
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3749261345795703e-07,
+      "loss": 1.5104,
+      "step": 8504
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.343621950068543e-07,
+      "loss": 1.2015,
+      "step": 8505
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3124633809313853e-07,
+      "loss": 1.0389,
+      "step": 8506
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2814504317204563e-07,
+      "loss": 1.0562,
+      "step": 8507
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.250583106966909e-07,
+      "loss": 1.5523,
+      "step": 8508
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2198614111805804e-07,
+      "loss": 0.8408,
+      "step": 8509
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1892853488498795e-07,
+      "loss": 0.6354,
+      "step": 8510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.158854924442345e-07,
+      "loss": 1.2413,
+      "step": 8511
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.128570142403753e-07,
+      "loss": 1.5075,
+      "step": 8512
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0984310071588974e-07,
+      "loss": 0.6912,
+      "step": 8513
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.068437523111367e-07,
+      "loss": 0.5341,
+      "step": 8514
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0385896946431015e-07,
+      "loss": 0.8269,
+      "step": 8515
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.008887526115056e-07,
+      "loss": 1.3363,
+      "step": 8516
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9793310218669824e-07,
+      "loss": 1.2041,
+      "step": 8517
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9499201862169813e-07,
+      "loss": 0.4879,
+      "step": 8518
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.920655023462282e-07,
+      "loss": 0.5922,
+      "step": 8519
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8915355378784646e-07,
+      "loss": 1.0407,
+      "step": 8520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.862561733720015e-07,
+      "loss": 1.267,
+      "step": 8521
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8337336152203243e-07,
+      "loss": 1.2093,
+      "step": 8522
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8050511865911343e-07,
+      "loss": 1.061,
+      "step": 8523
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7765144520229826e-07,
+      "loss": 0.6975,
+      "step": 8524
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.748123415685422e-07,
+      "loss": 1.4396,
+      "step": 8525
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7198780817263567e-07,
+      "loss": 1.0857,
+      "step": 8526
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6917784542724866e-07,
+      "loss": 1.0484,
+      "step": 8527
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6638245374293047e-07,
+      "loss": 1.429,
+      "step": 8528
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6360163352810995e-07,
+      "loss": 1.4979,
+      "step": 8529
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6083538518906213e-07,
+      "loss": 1.3994,
+      "step": 8530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.580837091299526e-07,
+      "loss": 1.299,
+      "step": 8531
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5534660575281534e-07,
+      "loss": 1.2171,
+      "step": 8532
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5262407545755265e-07,
+      "loss": 1.1035,
+      "step": 8533
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.499161186419241e-07,
+      "loss": 0.7548,
+      "step": 8534
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4722273570157995e-07,
+      "loss": 1.5929,
+      "step": 8535
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.445439270300387e-07,
+      "loss": 1.6622,
+      "step": 8536
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4187969301867617e-07,
+      "loss": 1.0242,
+      "step": 8537
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.392300340567588e-07,
+      "loss": 0.6379,
+      "step": 8538
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3659495053141022e-07,
+      "loss": 1.5368,
+      "step": 8539
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3397444282761137e-07,
+      "loss": 1.0859,
+      "step": 8540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.313685113282449e-07,
+      "loss": 0.6486,
+      "step": 8541
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.287771564140284e-07,
+      "loss": 1.1537,
+      "step": 8542
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2620037846359243e-07,
+      "loss": 1.3021,
+      "step": 8543
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2363817785340246e-07,
+      "loss": 0.3947,
+      "step": 8544
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2109055495781461e-07,
+      "loss": 1.3312,
+      "step": 8545
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1855751014903115e-07,
+      "loss": 1.3893,
+      "step": 8546
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.160390437971449e-07,
+      "loss": 0.7828,
+      "step": 8547
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.135351562701282e-07,
+      "loss": 1.2175,
+      "step": 8548
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1104584793378847e-07,
+      "loss": 0.9643,
+      "step": 8549
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.085711191518236e-07,
+      "loss": 0.9436,
+      "step": 8550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0611097028582217e-07,
+      "loss": 0.5934,
+      "step": 8551
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0366540169520776e-07,
+      "loss": 1.0391,
+      "step": 8552
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0123441373728347e-07,
+      "loss": 1.1328,
+      "step": 8553
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9881800676723183e-07,
+      "loss": 0.5109,
+      "step": 8554
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9641618113810377e-07,
+      "loss": 1.3211,
+      "step": 8555
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.940289372008075e-07,
+      "loss": 1.1346,
+      "step": 8556
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9165627530413067e-07,
+      "loss": 0.9728,
+      "step": 8557
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.892981957947293e-07,
+      "loss": 1.5377,
+      "step": 8558
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8695469901712782e-07,
+      "loss": 1.3583,
+      "step": 8559
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.84625785313719e-07,
+      "loss": 1.5213,
+      "step": 8560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.82311455024764e-07,
+      "loss": 1.1732,
+      "step": 8561
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8001170848840344e-07,
+      "loss": 1.1325,
+      "step": 8562
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7772654604063522e-07,
+      "loss": 1.361,
+      "step": 8563
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.754559680153256e-07,
+      "loss": 0.6245,
+      "step": 8564
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7319997474422033e-07,
+      "loss": 0.5572,
+      "step": 8565
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7095856655692245e-07,
+      "loss": 0.8996,
+      "step": 8566
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.687317437809255e-07,
+      "loss": 1.0223,
+      "step": 8567
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6651950674155815e-07,
+      "loss": 0.651,
+      "step": 8568
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6432185576205072e-07,
+      "loss": 0.8569,
+      "step": 8569
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6213879116349083e-07,
+      "loss": 0.4241,
+      "step": 8570
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5997031326481227e-07,
+      "loss": 1.4287,
+      "step": 8571
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5781642238286155e-07,
+      "loss": 0.5797,
+      "step": 8572
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.556771188323092e-07,
+      "loss": 1.1162,
+      "step": 8573
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5355240292573848e-07,
+      "loss": 1.3486,
+      "step": 8574
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5144227497355667e-07,
+      "loss": 1.189,
+      "step": 8575
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4934673528407273e-07,
+      "loss": 1.306,
+      "step": 8576
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4726578416346394e-07,
+      "loss": 1.0467,
+      "step": 8577
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4519942191574264e-07,
+      "loss": 1.1097,
+      "step": 8578
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4314764884282295e-07,
+      "loss": 1.0757,
+      "step": 8579
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4111046524447612e-07,
+      "loss": 0.3669,
+      "step": 8580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.39087871418353e-07,
+      "loss": 1.4923,
+      "step": 8581
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.370798676599505e-07,
+      "loss": 1.3044,
+      "step": 8582
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.350864542626451e-07,
+      "loss": 1.2928,
+      "step": 8583
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3310763151769268e-07,
+      "loss": 1.4462,
+      "step": 8584
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.311433997141953e-07,
+      "loss": 1.4727,
+      "step": 8585
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2919375913913457e-07,
+      "loss": 1.2199,
+      "step": 8586
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2725871007737145e-07,
+      "loss": 0.725,
+      "step": 8587
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2533825281161315e-07,
+      "loss": 1.435,
+      "step": 8588
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2343238762245745e-07,
+      "loss": 1.193,
+      "step": 8589
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.215411147883483e-07,
+      "loss": 0.973,
+      "step": 8590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1966443458562015e-07,
+      "loss": 1.3017,
+      "step": 8591
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1780234728844264e-07,
+      "loss": 0.5719,
+      "step": 8592
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.159548531688981e-07,
+      "loss": 1.0077,
+      "step": 8593
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1412195249689284e-07,
+      "loss": 1.1876,
+      "step": 8594
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1230364554022376e-07,
+      "loss": 1.1537,
+      "step": 8595
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1049993256456725e-07,
+      "loss": 0.6169,
+      "step": 8596
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0871081383344583e-07,
+      "loss": 1.1782,
+      "step": 8597
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0693628960825042e-07,
+      "loss": 1.448,
+      "step": 8598
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0517636014824028e-07,
+      "loss": 1.069,
+      "step": 8599
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0343102571056529e-07,
+      "loss": 1.0483,
+      "step": 8600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0170028655022145e-07,
+      "loss": 1.4439,
+      "step": 8601
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.998414292007318e-08,
+      "loss": 1.1318,
+      "step": 8602
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.828259507085324e-08,
+      "loss": 1.4613,
+      "step": 8603
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.659564325116277e-08,
+      "loss": 1.0441,
+      "step": 8604
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.492328770747128e-08,
+      "loss": 1.139,
+      "step": 8605
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.326552868412775e-08,
+      "loss": 1.2034,
+      "step": 8606
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.162236642332734e-08,
+      "loss": 1.0954,
+      "step": 8607
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.999380116514466e-08,
+      "loss": 1.3254,
+      "step": 8608
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.837983314752274e-08,
+      "loss": 1.0821,
+      "step": 8609
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.678046260627293e-08,
+      "loss": 1.0302,
+      "step": 8610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.519568977505276e-08,
+      "loss": 1.3003,
+      "step": 8611
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.362551488541038e-08,
+      "loss": 0.9971,
+      "step": 8612
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.206993816676223e-08,
+      "loss": 1.4802,
+      "step": 8613
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.052895984638208e-08,
+      "loss": 0.4392,
+      "step": 8614
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.900258014941208e-08,
+      "loss": 1.3893,
+      "step": 8615
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.749079929885161e-08,
+      "loss": 1.3589,
+      "step": 8616
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.599361751559065e-08,
+      "loss": 1.3731,
+      "step": 8617
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.451103501837641e-08,
+      "loss": 1.3157,
+      "step": 8618
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.304305202380234e-08,
+      "loss": 1.2029,
+      "step": 8619
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.158966874636353e-08,
+      "loss": 0.7707,
+      "step": 8620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.01508853983901e-08,
+      "loss": 1.0035,
+      "step": 8621
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.87267021901028e-08,
+      "loss": 1.3371,
+      "step": 8622
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.731711932959073e-08,
+      "loss": 0.7046,
+      "step": 8623
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.592213702277805e-08,
+      "loss": 0.5693,
+      "step": 8624
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.454175547349062e-08,
+      "loss": 0.5683,
+      "step": 8625
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.317597488340043e-08,
+      "loss": 0.997,
+      "step": 8626
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.182479545205899e-08,
+      "loss": 0.5618,
+      "step": 8627
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.048821737687504e-08,
+      "loss": 1.1855,
+      "step": 8628
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9166240853136824e-08,
+      "loss": 1.3421,
+      "step": 8629
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.7858866073978725e-08,
+      "loss": 1.1278,
+      "step": 8630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.656609323041462e-08,
+      "loss": 1.1364,
+      "step": 8631
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.528792251132675e-08,
+      "loss": 1.0724,
+      "step": 8632
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.402435410346574e-08,
+      "loss": 1.1808,
+      "step": 8633
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.277538819143946e-08,
+      "loss": 1.2226,
+      "step": 8634
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.154102495772417e-08,
+      "loss": 1.3998,
+      "step": 8635
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0321264582664504e-08,
+      "loss": 0.6621,
+      "step": 8636
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9116107244484565e-08,
+      "loss": 1.2178,
+      "step": 8637
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.792555311924351e-08,
+      "loss": 1.3425,
+      "step": 8638
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.67496023808911e-08,
+      "loss": 1.336,
+      "step": 8639
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.558825520125653e-08,
+      "loss": 1.1993,
+      "step": 8640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4441511749993e-08,
+      "loss": 1.4484,
+      "step": 8641
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3309372194655365e-08,
+      "loss": 1.061,
+      "step": 8642
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.219183670065574e-08,
+      "loss": 1.9227,
+      "step": 8643
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1088905431263534e-08,
+      "loss": 1.2731,
+      "step": 8644
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.000057854762762e-08,
+      "loss": 0.5489,
+      "step": 8645
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.892685620875414e-08,
+      "loss": 1.046,
+      "step": 8646
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7867738571517595e-08,
+      "loss": 1.6426,
+      "step": 8647
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.682322579064978e-08,
+      "loss": 0.9357,
+      "step": 8648
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5793318018784164e-08,
+      "loss": 0.6321,
+      "step": 8649
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.477801540636705e-08,
+      "loss": 1.4043,
+      "step": 8650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.377731810175755e-08,
+      "loss": 1.0862,
+      "step": 8651
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.279122625113873e-08,
+      "loss": 0.8878,
+      "step": 8652
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.181973999860643e-08,
+      "loss": 1.221,
+      "step": 8653
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0862859486091576e-08,
+      "loss": 1.3238,
+      "step": 8654
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9920584853393444e-08,
+      "loss": 0.376,
+      "step": 8655
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.899291623817968e-08,
+      "loss": 1.6948,
+      "step": 8656
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8079853775997423e-08,
+      "loss": 0.9104,
+      "step": 8657
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7181397600239966e-08,
+      "loss": 1.5065,
+      "step": 8658
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.629754784216898e-08,
+      "loss": 1.1748,
+      "step": 8659
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5428304630936706e-08,
+      "loss": 0.6351,
+      "step": 8660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4573668093530456e-08,
+      "loss": 1.2397,
+      "step": 8661
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3733638354817013e-08,
+      "loss": 1.1299,
+      "step": 8662
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.290821553753153e-08,
+      "loss": 1.149,
+      "step": 8663
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2097399762266437e-08,
+      "loss": 0.8579,
+      "step": 8664
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1301191147482526e-08,
+      "loss": 1.2806,
+      "step": 8665
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0519589809520067e-08,
+      "loss": 1.2083,
+      "step": 8666
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9752595862565505e-08,
+      "loss": 1.1235,
+      "step": 8667
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9000209418684745e-08,
+      "loss": 1.1065,
+      "step": 8668
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8262430587800972e-08,
+      "loss": 1.2938,
+      "step": 8669
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.753925947771684e-08,
+      "loss": 1.3014,
+      "step": 8670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6830696194070074e-08,
+      "loss": 1.3112,
+      "step": 8671
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6136740840400068e-08,
+      "loss": 1.2954,
+      "step": 8672
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5457393518092388e-08,
+      "loss": 0.9409,
+      "step": 8673
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4792654326389877e-08,
+      "loss": 1.4866,
+      "step": 8674
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4142523362437043e-08,
+      "loss": 1.0484,
+      "step": 8675
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.350700072121347e-08,
+      "loss": 1.3156,
+      "step": 8676
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2886086495556005e-08,
+      "loss": 1.3673,
+      "step": 8677
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2279780776203176e-08,
+      "loss": 1.2944,
+      "step": 8678
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1688083651717475e-08,
+      "loss": 1.2947,
+      "step": 8679
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1110995208574172e-08,
+      "loss": 1.2475,
+      "step": 8680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0548515531061398e-08,
+      "loss": 1.0671,
+      "step": 8681
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0000644701380069e-08,
+      "loss": 1.3518,
+      "step": 8682
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.46738279956616e-09,
+      "loss": 1.0009,
+      "step": 8683
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.948729903535125e-09,
+      "loss": 1.1419,
+      "step": 8684
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.444686089048582e-09,
+      "loss": 0.803,
+      "step": 8685
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.95525142978093e-09,
+      "loss": 0.7656,
+      "step": 8686
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.48042599721943e-09,
+      "loss": 1.2918,
+      "step": 8687
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.02020986073082e-09,
+      "loss": 1.7849,
+      "step": 8688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.574603087572406e-09,
+      "loss": 1.4075,
+      "step": 8689
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.143605742836567e-09,
+      "loss": 0.5129,
+      "step": 8690
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.727217889506253e-09,
+      "loss": 0.4931,
+      "step": 8691
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.325439588410586e-09,
+      "loss": 1.2862,
+      "step": 8692
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9382708982359526e-09,
+      "loss": 0.6253,
+      "step": 8693
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.565711875581524e-09,
+      "loss": 1.4611,
+      "step": 8694
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2077625748482284e-09,
+      "loss": 1.7055,
+      "step": 8695
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.864423048349775e-09,
+      "loss": 1.4663,
+      "step": 8696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5356933462460385e-09,
+      "loss": 1.2401,
+      "step": 8697
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.221573516565268e-09,
+      "loss": 1.4874,
+      "step": 8698
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.922063605204084e-09,
+      "loss": 0.731,
+      "step": 8699
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.637163655927477e-09,
+      "loss": 1.051,
+      "step": 8700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3668737103466064e-09,
+      "loss": 0.4403,
+      "step": 8701
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1111938079632078e-09,
+      "loss": 1.1463,
+      "step": 8702
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8701239861251828e-09,
+      "loss": 1.5949,
+      "step": 8703
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6436642800599089e-09,
+      "loss": 0.4679,
+      "step": 8704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4318147228409295e-09,
+      "loss": 1.2534,
+      "step": 8705
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2345753454434672e-09,
+      "loss": 1.1259,
+      "step": 8706
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0519461766667072e-09,
+      "loss": 0.4933,
+      "step": 8707
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.839272432004109e-10,
+      "loss": 1.307,
+      "step": 8708
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.305185695916095e-10,
+      "loss": 1.0688,
+      "step": 8709
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.917201782557058e-10,
+      "loss": 1.312,
+      "step": 8710
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.675320894653723e-10,
+      "loss": 1.5121,
+      "step": 8711
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.579543213727554e-10,
+      "loss": 0.648,
+      "step": 8712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.629868899761689e-10,
+      "loss": 0.5611,
+      "step": 8713
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.826298091645029e-10,
+      "loss": 0.9669,
+      "step": 8714
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1688309067281468e-10,
+      "loss": 1.0046,
+      "step": 8715
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.574674410453341e-11,
+      "loss": 0.5706,
+      "step": 8716
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.922077694256231e-11,
+      "loss": 1.282,
+      "step": 8717
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.30519450486966e-12,
+      "loss": 1.3111,
+      "step": 8718
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.5905,
+      "step": 8719
+    },
+    {
+      "epoch": 1.0,
+      "step": 8719,
+      "total_flos": 1.4738059356587164e+19,
+      "train_loss": 1.107404184918746,
+      "train_runtime": 127760.4241,
+      "train_samples_per_second": 0.068,
+      "train_steps_per_second": 0.068
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 8719,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 1.4738059356587164e+19,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}