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": 5.000000000000001e-07,
+      "loss": 1.5884,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 1.557,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-06,
+      "loss": 1.19,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.4482,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 1.0548,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 1.3078,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5000000000000004e-06,
+      "loss": 1.1797,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.5884,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-06,
+      "loss": 0.8202,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 1.2825,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 1.2916,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 0.8961,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 0.9632,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 1.4184,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.5e-06,
+      "loss": 1.2376,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.3428,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.500000000000002e-06,
+      "loss": 1.1264,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-06,
+      "loss": 0.9978,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.5e-06,
+      "loss": 1.0725,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 1.4059,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.05e-05,
+      "loss": 1.9265,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 1.3093,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1500000000000002e-05,
+      "loss": 0.7429,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 1.2852,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.25e-05,
+      "loss": 0.9292,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 1.6453,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 1.481,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 1.0882,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.45e-05,
+      "loss": 1.3001,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 1.0973,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.55e-05,
+      "loss": 1.4475,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.0719,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.65e-05,
+      "loss": 1.351,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 1.1896,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.75e-05,
+      "loss": 1.5056,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-05,
+      "loss": 1.369,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.85e-05,
+      "loss": 1.435,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9e-05,
+      "loss": 1.1323,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9500000000000003e-05,
+      "loss": 0.8013,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.8014,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.05e-05,
+      "loss": 1.2288,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1e-05,
+      "loss": 0.9482,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.15e-05,
+      "loss": 1.2256,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 1.1583,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.25e-05,
+      "loss": 0.5951,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 1.4167,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.35e-05,
+      "loss": 1.2015,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-05,
+      "loss": 0.9047,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.45e-05,
+      "loss": 0.8149,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-05,
+      "loss": 1.2908,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5500000000000003e-05,
+      "loss": 1.2686,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.5878,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6500000000000004e-05,
+      "loss": 1.2825,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 1.3749,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 1.1801,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 1.3112,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 0.4748,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9e-05,
+      "loss": 0.7116,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.95e-05,
+      "loss": 1.0059,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 0.7261,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.05e-05,
+      "loss": 1.5567,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1e-05,
+      "loss": 1.4507,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.15e-05,
+      "loss": 1.7152,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.0488,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 1.2642,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3e-05,
+      "loss": 1.3988,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.35e-05,
+      "loss": 0.9996,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 1.3613,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.45e-05,
+      "loss": 1.4942,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5e-05,
+      "loss": 1.2362,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.55e-05,
+      "loss": 1.3143,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-05,
+      "loss": 1.3011,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.65e-05,
+      "loss": 1.141,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7e-05,
+      "loss": 0.8567,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.9715,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8e-05,
+      "loss": 1.2306,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.85e-05,
+      "loss": 1.4444,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 1.3807,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9500000000000005e-05,
+      "loss": 1.3726,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-05,
+      "loss": 1.0251,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.05e-05,
+      "loss": 0.9919,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1e-05,
+      "loss": 0.8163,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.15e-05,
+      "loss": 0.9911,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2e-05,
+      "loss": 1.3138,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.25e-05,
+      "loss": 1.2602,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3e-05,
+      "loss": 1.2896,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.35e-05,
+      "loss": 1.3603,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 1.0338,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4500000000000004e-05,
+      "loss": 1.2492,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5e-05,
+      "loss": 1.1808,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.55e-05,
+      "loss": 1.1323,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 0.7277,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6500000000000005e-05,
+      "loss": 1.2923,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7e-05,
+      "loss": 1.4806,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.75e-05,
+      "loss": 1.4874,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8e-05,
+      "loss": 1.1879,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.85e-05,
+      "loss": 1.1386,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9e-05,
+      "loss": 1.1471,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9500000000000004e-05,
+      "loss": 1.356,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-05,
+      "loss": 1.4908,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.05e-05,
+      "loss": 1.3574,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.1000000000000006e-05,
+      "loss": 1.5173,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.1500000000000005e-05,
+      "loss": 1.3344,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 0.996,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.25e-05,
+      "loss": 1.3407,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.300000000000001e-05,
+      "loss": 1.3285,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.3500000000000006e-05,
+      "loss": 1.3159,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 1.0723,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.45e-05,
+      "loss": 0.678,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 0.9342,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.550000000000001e-05,
+      "loss": 0.5789,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 1.2818,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.65e-05,
+      "loss": 1.095,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 0.9139,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7499999999999995e-05,
+      "loss": 1.5349,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.8e-05,
+      "loss": 1.3035,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.85e-05,
+      "loss": 1.5355,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9e-05,
+      "loss": 1.2594,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.95e-05,
+      "loss": 1.0445,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-05,
+      "loss": 0.8948,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.05e-05,
+      "loss": 1.1551,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.1e-05,
+      "loss": 1.0026,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.15e-05,
+      "loss": 1.4377,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.2e-05,
+      "loss": 1.4777,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.25e-05,
+      "loss": 0.8209,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.3e-05,
+      "loss": 0.7917,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.35e-05,
+      "loss": 1.6241,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 1.8283,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.450000000000001e-05,
+      "loss": 0.5513,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 1.1312,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.55e-05,
+      "loss": 0.9688,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.6e-05,
+      "loss": 1.0379,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.65e-05,
+      "loss": 1.0558,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.7e-05,
+      "loss": 0.9088,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.750000000000001e-05,
+      "loss": 1.0406,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 1.2543,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.850000000000001e-05,
+      "loss": 1.3826,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.9e-05,
+      "loss": 1.1313,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.95e-05,
+      "loss": 1.3691,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7e-05,
+      "loss": 1.3661,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.05e-05,
+      "loss": 1.4231,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.1e-05,
+      "loss": 1.278,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.15e-05,
+      "loss": 1.0915,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.2e-05,
+      "loss": 1.1608,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.25e-05,
+      "loss": 1.0823,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.3e-05,
+      "loss": 1.8188,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.35e-05,
+      "loss": 1.3276,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.4e-05,
+      "loss": 0.6877,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.450000000000001e-05,
+      "loss": 1.5497,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 1.066,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.55e-05,
+      "loss": 1.3565,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.6e-05,
+      "loss": 1.4165,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.65e-05,
+      "loss": 1.2677,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.7e-05,
+      "loss": 0.6844,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.75e-05,
+      "loss": 1.3172,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 1.1638,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.850000000000001e-05,
+      "loss": 0.5805,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.900000000000001e-05,
+      "loss": 1.0484,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.950000000000001e-05,
+      "loss": 1.2836,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8e-05,
+      "loss": 1.4505,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.05e-05,
+      "loss": 1.6705,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.1e-05,
+      "loss": 0.9863,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.15e-05,
+      "loss": 1.5563,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.2e-05,
+      "loss": 1.377,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.25e-05,
+      "loss": 0.642,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.3e-05,
+      "loss": 0.916,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.35e-05,
+      "loss": 1.6484,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.4e-05,
+      "loss": 0.6917,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.450000000000001e-05,
+      "loss": 0.8816,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.5e-05,
+      "loss": 0.7474,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.55e-05,
+      "loss": 1.19,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.6e-05,
+      "loss": 1.3613,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.65e-05,
+      "loss": 0.9675,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.7e-05,
+      "loss": 0.7473,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.75e-05,
+      "loss": 0.9498,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 0.7143,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.850000000000001e-05,
+      "loss": 1.4176,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.900000000000001e-05,
+      "loss": 1.4528,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.950000000000001e-05,
+      "loss": 1.3837,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9e-05,
+      "loss": 1.6324,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.05e-05,
+      "loss": 1.2994,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.1e-05,
+      "loss": 1.0913,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.15e-05,
+      "loss": 0.7607,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 1.4538,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.250000000000001e-05,
+      "loss": 1.4173,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.300000000000001e-05,
+      "loss": 1.9281,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.350000000000001e-05,
+      "loss": 1.1176,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.4e-05,
+      "loss": 1.24,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.449999999999999e-05,
+      "loss": 1.1693,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.5e-05,
+      "loss": 1.3903,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.55e-05,
+      "loss": 1.3537,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.6e-05,
+      "loss": 1.1501,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.65e-05,
+      "loss": 1.3329,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.7e-05,
+      "loss": 1.4665,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.75e-05,
+      "loss": 1.3539,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.8e-05,
+      "loss": 1.4233,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.850000000000001e-05,
+      "loss": 1.4529,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.900000000000001e-05,
+      "loss": 1.1142,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.95e-05,
+      "loss": 1.4062,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001,
+      "loss": 0.6816,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999999660012843e-05,
+      "loss": 0.8952,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999998640051416e-05,
+      "loss": 1.214,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999996940115859e-05,
+      "loss": 1.3403,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999994560206403e-05,
+      "loss": 0.6464,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999991500323372e-05,
+      "loss": 1.3563,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999987760467183e-05,
+      "loss": 1.5646,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999983340638342e-05,
+      "loss": 0.683,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999978240837451e-05,
+      "loss": 0.608,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999972461065203e-05,
+      "loss": 1.3142,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999966001322387e-05,
+      "loss": 1.235,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999958861609878e-05,
+      "loss": 1.2528,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99995104192865e-05,
+      "loss": 1.4009,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999942542279763e-05,
+      "loss": 0.9948,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999933362664376e-05,
+      "loss": 1.1872,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999923503083736e-05,
+      "loss": 0.9207,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999912963539185e-05,
+      "loss": 0.9779,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999901744032153e-05,
+      "loss": 1.0676,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99988984456417e-05,
+      "loss": 1.6958,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99987726513685e-05,
+      "loss": 0.5053,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999864005751908e-05,
+      "loss": 0.6833,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999850066411147e-05,
+      "loss": 1.0865,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999835447116459e-05,
+      "loss": 0.9784,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999820147869834e-05,
+      "loss": 1.178,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999804168673353e-05,
+      "loss": 1.4837,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99978750952919e-05,
+      "loss": 1.5693,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999770170439608e-05,
+      "loss": 0.7003,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999752151406967e-05,
+      "loss": 0.769,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999733452433717e-05,
+      "loss": 0.6939,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999714073522401e-05,
+      "loss": 1.4563,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999694014675653e-05,
+      "loss": 1.4483,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999673275896204e-05,
+      "loss": 1.125,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999651857186872e-05,
+      "loss": 1.0573,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99962975855057e-05,
+      "loss": 1.2787,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999606979990306e-05,
+      "loss": 1.623,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999583521509173e-05,
+      "loss": 1.4727,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999559383110364e-05,
+      "loss": 0.8807,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999534564797162e-05,
+      "loss": 1.3286,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999509066572943e-05,
+      "loss": 0.6759,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999482888441172e-05,
+      "loss": 1.6883,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99945603040541e-05,
+      "loss": 1.6611,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99942849246931e-05,
+      "loss": 1.1791,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999400274636617e-05,
+      "loss": 1.6935,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999371376911169e-05,
+      "loss": 1.2084,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999341799296894e-05,
+      "loss": 1.0383,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999311541797816e-05,
+      "loss": 0.6134,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999280604418051e-05,
+      "loss": 1.3884,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999248987161803e-05,
+      "loss": 1.2489,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999216690033375e-05,
+      "loss": 1.002,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999183713037159e-05,
+      "loss": 1.2775,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999150056177637e-05,
+      "loss": 1.2298,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999115719459386e-05,
+      "loss": 1.1124,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99908070288708e-05,
+      "loss": 1.2102,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99904500646548e-05,
+      "loss": 1.2831,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999008630199436e-05,
+      "loss": 1.4734,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.998971574093899e-05,
+      "loss": 1.1609,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.998933838153908e-05,
+      "loss": 1.2522,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.998895422384593e-05,
+      "loss": 1.6348,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.998856326791181e-05,
+      "loss": 0.8174,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.998816551378988e-05,
+      "loss": 1.5131,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.998776096153421e-05,
+      "loss": 0.92,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.998734961119984e-05,
+      "loss": 1.4555,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.998693146284273e-05,
+      "loss": 1.0493,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.998650651651969e-05,
+      "loss": 1.2802,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.998607477228857e-05,
+      "loss": 1.4147,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.998563623020803e-05,
+      "loss": 1.3871,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.998519089033776e-05,
+      "loss": 1.2754,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.998473875273829e-05,
+      "loss": 1.3142,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.998427981747111e-05,
+      "loss": 1.4297,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.998381408459865e-05,
+      "loss": 1.406,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.998334155418425e-05,
+      "loss": 1.2616,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.998286222629213e-05,
+      "loss": 1.1345,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.998237610098754e-05,
+      "loss": 1.1626,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.998188317833655e-05,
+      "loss": 1.2743,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.998138345840621e-05,
+      "loss": 1.0215,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.998087694126445e-05,
+      "loss": 0.6795,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99803636269802e-05,
+      "loss": 1.0368,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.997984351562323e-05,
+      "loss": 1.2455,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.997931660726429e-05,
+      "loss": 1.0486,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.997878290197504e-05,
+      "loss": 1.3916,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.997824239982804e-05,
+      "loss": 0.9824,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.997769510089683e-05,
+      "loss": 0.7325,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99771410052558e-05,
+      "loss": 0.8018,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.997658011298034e-05,
+      "loss": 0.8046,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.997601242414669e-05,
+      "loss": 0.9704,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99754379388321e-05,
+      "loss": 1.0454,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.997485665711466e-05,
+      "loss": 1.8743,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.997426857907344e-05,
+      "loss": 0.8478,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.997367370478841e-05,
+      "loss": 1.1584,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.997307203434045e-05,
+      "loss": 1.2554,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.997246356781143e-05,
+      "loss": 0.7563,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.997184830528406e-05,
+      "loss": 1.3431,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9971226246842e-05,
+      "loss": 1.3189,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.997059739256991e-05,
+      "loss": 1.0724,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.996996174255325e-05,
+      "loss": 0.961,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.996931929687849e-05,
+      "loss": 1.3819,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.996867005563301e-05,
+      "loss": 0.8472,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.996801401890507e-05,
+      "loss": 1.1912,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.996735118678392e-05,
+      "loss": 1.0044,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.996668155935969e-05,
+      "loss": 1.2633,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.996600513672344e-05,
+      "loss": 1.3408,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.996532191896717e-05,
+      "loss": 1.4845,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.996463190618377e-05,
+      "loss": 0.8216,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.996393509846712e-05,
+      "loss": 1.1004,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.996323149591196e-05,
+      "loss": 1.0951,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.996252109861394e-05,
+      "loss": 1.3594,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996180390666975e-05,
+      "loss": 1.398,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996107992017687e-05,
+      "loss": 1.214,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996034913923376e-05,
+      "loss": 1.053,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99596115639398e-05,
+      "loss": 0.9925,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995886719439533e-05,
+      "loss": 1.1578,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995811603070155e-05,
+      "loss": 1.5099,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995735807296062e-05,
+      "loss": 1.3316,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995659332127563e-05,
+      "loss": 1.3256,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995582177575058e-05,
+      "loss": 1.39,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995504343649037e-05,
+      "loss": 1.325,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995425830360088e-05,
+      "loss": 1.2894,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995346637718887e-05,
+      "loss": 0.9088,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995266765736205e-05,
+      "loss": 1.211,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995186214422902e-05,
+      "loss": 1.2807,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995104983789936e-05,
+      "loss": 1.5239,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99502307384835e-05,
+      "loss": 1.1935,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994940484609285e-05,
+      "loss": 0.9268,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994857216083974e-05,
+      "loss": 0.5388,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99477326828374e-05,
+      "loss": 1.1857,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99468864122e-05,
+      "loss": 1.3607,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994603334904261e-05,
+      "loss": 1.5327,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994517349348127e-05,
+      "loss": 1.1077,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994430684563289e-05,
+      "loss": 1.0419,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994343340561534e-05,
+      "loss": 0.7475,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99425531735474e-05,
+      "loss": 1.536,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99416661495488e-05,
+      "loss": 1.3034,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994077233374014e-05,
+      "loss": 1.2248,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9939871726243e-05,
+      "loss": 1.2088,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.993896432717981e-05,
+      "loss": 0.7454,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.993805013667403e-05,
+      "loss": 1.1787,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.993712915484997e-05,
+      "loss": 1.4041,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.993620138183287e-05,
+      "loss": 1.2839,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.993526681774888e-05,
+      "loss": 1.5503,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.993432546272514e-05,
+      "loss": 0.9466,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.993337731688963e-05,
+      "loss": 1.0105,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.993242238037132e-05,
+      "loss": 1.226,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.993146065330008e-05,
+      "loss": 1.2443,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.993049213580668e-05,
+      "loss": 1.4895,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.992951682802283e-05,
+      "loss": 0.845,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.992853473008117e-05,
+      "loss": 1.9573,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99275458421153e-05,
+      "loss": 1.2565,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.992655016425964e-05,
+      "loss": 0.9356,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.992554769664965e-05,
+      "loss": 0.8197,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.992453843942162e-05,
+      "loss": 0.8327,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.992352239271284e-05,
+      "loss": 1.548,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.992249955666147e-05,
+      "loss": 0.8146,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99214699314066e-05,
+      "loss": 0.7456,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.992043351708827e-05,
+      "loss": 1.0017,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991939031384742e-05,
+      "loss": 1.5275,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991834032182592e-05,
+      "loss": 0.5566,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991728354116656e-05,
+      "loss": 1.2789,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991621997201307e-05,
+      "loss": 1.2486,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991514961451008e-05,
+      "loss": 1.2267,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991407246880315e-05,
+      "loss": 1.4233,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991298853503878e-05,
+      "loss": 1.2175,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991189781336437e-05,
+      "loss": 0.9514,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991080030392823e-05,
+      "loss": 0.8844,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.990969600687968e-05,
+      "loss": 0.9589,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.990858492236882e-05,
+      "loss": 1.4615,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99074670505468e-05,
+      "loss": 1.2794,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.990634239156565e-05,
+      "loss": 0.6135,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.990521094557828e-05,
+      "loss": 1.8493,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.990407271273858e-05,
+      "loss": 1.1832,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.990292769320136e-05,
+      "loss": 1.1362,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.990177588712232e-05,
+      "loss": 1.39,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.990061729465811e-05,
+      "loss": 1.1998,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.989945191596628e-05,
+      "loss": 1.4347,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.989827975120533e-05,
+      "loss": 0.7793,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.989710080053465e-05,
+      "loss": 0.7483,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.989591506411457e-05,
+      "loss": 1.2062,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.989472254210636e-05,
+      "loss": 1.0486,
+      "step": 376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.989352323467221e-05,
+      "loss": 1.2309,
+      "step": 377
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.989231714197518e-05,
+      "loss": 0.7526,
+      "step": 378
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.989110426417933e-05,
+      "loss": 0.7665,
+      "step": 379
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988988460144959e-05,
+      "loss": 0.4088,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.98886581539518e-05,
+      "loss": 1.2818,
+      "step": 381
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988742492185282e-05,
+      "loss": 1.525,
+      "step": 382
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988618490532029e-05,
+      "loss": 0.6147,
+      "step": 383
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988493810452289e-05,
+      "loss": 0.7003,
+      "step": 384
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988368451963015e-05,
+      "loss": 1.1302,
+      "step": 385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988242415081257e-05,
+      "loss": 1.2855,
+      "step": 386
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988115699824156e-05,
+      "loss": 1.4127,
+      "step": 387
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.987988306208942e-05,
+      "loss": 0.9269,
+      "step": 388
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.987860234252943e-05,
+      "loss": 1.0446,
+      "step": 389
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.987731483973575e-05,
+      "loss": 0.9185,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.987602055388346e-05,
+      "loss": 1.5635,
+      "step": 391
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.987471948514857e-05,
+      "loss": 1.2144,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987341163370806e-05,
+      "loss": 1.2171,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987209699973976e-05,
+      "loss": 0.8214,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987077558342245e-05,
+      "loss": 0.6057,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986944738493585e-05,
+      "loss": 1.3233,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986811240446059e-05,
+      "loss": 0.8188,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98667706421782e-05,
+      "loss": 1.3493,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986542209827118e-05,
+      "loss": 1.4346,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98640667729229e-05,
+      "loss": 0.979,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98627046663177e-05,
+      "loss": 1.5917,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986133577864081e-05,
+      "loss": 1.335,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985996011007838e-05,
+      "loss": 1.2419,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98585776608175e-05,
+      "loss": 1.3577,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985718843104619e-05,
+      "loss": 1.3073,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985579242095337e-05,
+      "loss": 1.1479,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985438963072887e-05,
+      "loss": 1.3896,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98529800605635e-05,
+      "loss": 1.4273,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985156371064892e-05,
+      "loss": 1.4633,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985014058117778e-05,
+      "loss": 1.5096,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984871067234357e-05,
+      "loss": 1.359,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984727398434081e-05,
+      "loss": 1.7109,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984583051736485e-05,
+      "loss": 0.4647,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984438027161198e-05,
+      "loss": 1.1025,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984292324727945e-05,
+      "loss": 1.3641,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98414594445654e-05,
+      "loss": 0.9137,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983998886366889e-05,
+      "loss": 1.3942,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983851150478993e-05,
+      "loss": 1.0875,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983702736812941e-05,
+      "loss": 0.4291,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98355364538892e-05,
+      "loss": 1.3205,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983403876227202e-05,
+      "loss": 1.5215,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983253429348157e-05,
+      "loss": 0.6945,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983102304772245e-05,
+      "loss": 0.6543,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.982950502520016e-05,
+      "loss": 1.0651,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.982798022612117e-05,
+      "loss": 0.8181,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.982644865069284e-05,
+      "loss": 0.9925,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.982491029912345e-05,
+      "loss": 0.7406,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.982336517162219e-05,
+      "loss": 1.3011,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.982181326839922e-05,
+      "loss": 0.8639,
+      "step": 429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.982025458966557e-05,
+      "loss": 1.1516,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.981868913563324e-05,
+      "loss": 1.1786,
+      "step": 431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.981711690651509e-05,
+      "loss": 1.2685,
+      "step": 432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.981553790252495e-05,
+      "loss": 1.4447,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.981395212387756e-05,
+      "loss": 1.4742,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.981235957078856e-05,
+      "loss": 0.8204,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.981076024347456e-05,
+      "loss": 1.0938,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.980915414215304e-05,
+      "loss": 1.0454,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.980754126704242e-05,
+      "loss": 1.2776,
+      "step": 438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.980592161836204e-05,
+      "loss": 1.8951,
+      "step": 439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.980429519633219e-05,
+      "loss": 1.0689,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.980266200117402e-05,
+      "loss": 1.137,
+      "step": 441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.980102203310966e-05,
+      "loss": 0.9002,
+      "step": 442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.979937529236211e-05,
+      "loss": 1.1194,
+      "step": 443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.979772177915536e-05,
+      "loss": 1.1091,
+      "step": 444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.979606149371426e-05,
+      "loss": 1.3395,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.979439443626459e-05,
+      "loss": 1.2467,
+      "step": 446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.979272060703307e-05,
+      "loss": 0.5327,
+      "step": 447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.979104000624733e-05,
+      "loss": 1.1765,
+      "step": 448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.978935263413592e-05,
+      "loss": 0.8882,
+      "step": 449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.978765849092833e-05,
+      "loss": 1.0445,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.978595757685492e-05,
+      "loss": 1.2776,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.978424989214705e-05,
+      "loss": 1.3161,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.978253543703693e-05,
+      "loss": 1.0487,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.97808142117577e-05,
+      "loss": 0.5152,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.977908621654349e-05,
+      "loss": 1.1015,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.977735145162925e-05,
+      "loss": 1.3112,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.977560991725093e-05,
+      "loss": 1.1706,
+      "step": 457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.977386161364535e-05,
+      "loss": 1.3369,
+      "step": 458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.977210654105027e-05,
+      "loss": 1.0337,
+      "step": 459
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.977034469970438e-05,
+      "loss": 1.2715,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.976857608984729e-05,
+      "loss": 1.4055,
+      "step": 461
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.976680071171951e-05,
+      "loss": 1.121,
+      "step": 462
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.976501856556249e-05,
+      "loss": 1.2267,
+      "step": 463
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.976322965161856e-05,
+      "loss": 1.4043,
+      "step": 464
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.976143397013104e-05,
+      "loss": 1.1387,
+      "step": 465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.975963152134413e-05,
+      "loss": 0.5705,
+      "step": 466
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.975782230550295e-05,
+      "loss": 1.475,
+      "step": 467
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.975600632285353e-05,
+      "loss": 0.7616,
+      "step": 468
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.975418357364285e-05,
+      "loss": 1.5536,
+      "step": 469
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.975235405811879e-05,
+      "loss": 1.4424,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.975051777653015e-05,
+      "loss": 0.9515,
+      "step": 471
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.974867472912665e-05,
+      "loss": 1.3916,
+      "step": 472
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.974682491615897e-05,
+      "loss": 1.5862,
+      "step": 473
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.974496833787863e-05,
+      "loss": 1.1275,
+      "step": 474
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.974310499453814e-05,
+      "loss": 1.0443,
+      "step": 475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.97412348863909e-05,
+      "loss": 1.3923,
+      "step": 476
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.973935801369123e-05,
+      "loss": 0.5746,
+      "step": 477
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.973747437669439e-05,
+      "loss": 1.1023,
+      "step": 478
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.973558397565652e-05,
+      "loss": 0.9694,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973368681083473e-05,
+      "loss": 1.3027,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973178288248701e-05,
+      "loss": 1.1321,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972987219087228e-05,
+      "loss": 0.5934,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97279547362504e-05,
+      "loss": 1.7202,
+      "step": 483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972603051888213e-05,
+      "loss": 1.3503,
+      "step": 484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972409953902914e-05,
+      "loss": 1.3318,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972216179695407e-05,
+      "loss": 1.0153,
+      "step": 486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972021729292038e-05,
+      "loss": 1.034,
+      "step": 487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971826602719258e-05,
+      "loss": 0.6318,
+      "step": 488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971630800003599e-05,
+      "loss": 1.4953,
+      "step": 489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971434321171689e-05,
+      "loss": 1.152,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971237166250252e-05,
+      "loss": 1.3961,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971039335266094e-05,
+      "loss": 1.128,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970840828246125e-05,
+      "loss": 1.2542,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970641645217338e-05,
+      "loss": 1.3756,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970441786206821e-05,
+      "loss": 0.8164,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970241251241754e-05,
+      "loss": 1.1612,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970040040349409e-05,
+      "loss": 1.1613,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96983815355715e-05,
+      "loss": 1.2339,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96963559089243e-05,
+      "loss": 1.3371,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9694323523828e-05,
+      "loss": 0.8579,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969228438055897e-05,
+      "loss": 1.3644,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969023847939455e-05,
+      "loss": 1.1915,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968818582061295e-05,
+      "loss": 1.1373,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968612640449331e-05,
+      "loss": 1.1478,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96840602313157e-05,
+      "loss": 1.3863,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968198730136117e-05,
+      "loss": 0.9801,
+      "step": 506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.967990761491155e-05,
+      "loss": 1.2719,
+      "step": 507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.967782117224971e-05,
+      "loss": 1.2636,
+      "step": 508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.967572797365938e-05,
+      "loss": 1.1269,
+      "step": 509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.967362801942522e-05,
+      "loss": 0.8058,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.967152130983284e-05,
+      "loss": 1.169,
+      "step": 511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96694078451687e-05,
+      "loss": 1.2844,
+      "step": 512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.966728762572026e-05,
+      "loss": 0.5389,
+      "step": 513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.966516065177583e-05,
+      "loss": 0.5572,
+      "step": 514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.966302692362469e-05,
+      "loss": 0.3958,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9660886441557e-05,
+      "loss": 1.4549,
+      "step": 516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.965873920586387e-05,
+      "loss": 0.9737,
+      "step": 517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96565852168373e-05,
+      "loss": 0.8437,
+      "step": 518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96544244747702e-05,
+      "loss": 0.5308,
+      "step": 519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.965225697995647e-05,
+      "loss": 0.6395,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.965008273269085e-05,
+      "loss": 1.8094,
+      "step": 521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.964790173326902e-05,
+      "loss": 1.178,
+      "step": 522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96457139819876e-05,
+      "loss": 0.9442,
+      "step": 523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96435194791441e-05,
+      "loss": 1.2998,
+      "step": 524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.964131822503697e-05,
+      "loss": 0.6665,
+      "step": 525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.963911021996559e-05,
+      "loss": 1.1651,
+      "step": 526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.963689546423018e-05,
+      "loss": 1.3563,
+      "step": 527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9634673958132e-05,
+      "loss": 1.032,
+      "step": 528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96324457019731e-05,
+      "loss": 0.6555,
+      "step": 529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.963021069605657e-05,
+      "loss": 0.7052,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.962796894068634e-05,
+      "loss": 1.482,
+      "step": 531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.962572043616725e-05,
+      "loss": 1.4243,
+      "step": 532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.962346518280511e-05,
+      "loss": 1.0238,
+      "step": 533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.962120318090662e-05,
+      "loss": 1.216,
+      "step": 534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.961893443077941e-05,
+      "loss": 1.1963,
+      "step": 535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9616658932732e-05,
+      "loss": 1.3647,
+      "step": 536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.961437668707385e-05,
+      "loss": 0.8142,
+      "step": 537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.961208769411534e-05,
+      "loss": 0.6334,
+      "step": 538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.960979195416776e-05,
+      "loss": 1.2609,
+      "step": 539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.960748946754333e-05,
+      "loss": 1.2124,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.960518023455515e-05,
+      "loss": 1.0836,
+      "step": 541
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.960286425551728e-05,
+      "loss": 0.6611,
+      "step": 542
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.960054153074468e-05,
+      "loss": 0.7677,
+      "step": 543
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.959821206055323e-05,
+      "loss": 1.1671,
+      "step": 544
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.959587584525973e-05,
+      "loss": 1.8016,
+      "step": 545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.959353288518187e-05,
+      "loss": 1.3495,
+      "step": 546
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.959118318063832e-05,
+      "loss": 0.8806,
+      "step": 547
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.958882673194859e-05,
+      "loss": 1.2594,
+      "step": 548
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.958646353943317e-05,
+      "loss": 1.4037,
+      "step": 549
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.958409360341343e-05,
+      "loss": 1.0497,
+      "step": 550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.958171692421167e-05,
+      "loss": 1.2631,
+      "step": 551
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.95793335021511e-05,
+      "loss": 1.5571,
+      "step": 552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.957694333755587e-05,
+      "loss": 1.0927,
+      "step": 553
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.957454643075101e-05,
+      "loss": 1.2629,
+      "step": 554
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.957214278206251e-05,
+      "loss": 1.1831,
+      "step": 555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.956973239181725e-05,
+      "loss": 0.8073,
+      "step": 556
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.956731526034301e-05,
+      "loss": 1.6605,
+      "step": 557
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.956489138796852e-05,
+      "loss": 1.4699,
+      "step": 558
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.956246077502341e-05,
+      "loss": 1.1422,
+      "step": 559
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.956002342183824e-05,
+      "loss": 1.2026,
+      "step": 560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.955757932874448e-05,
+      "loss": 1.1061,
+      "step": 561
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.95551284960745e-05,
+      "loss": 1.3314,
+      "step": 562
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.95526709241616e-05,
+      "loss": 1.3114,
+      "step": 563
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.955020661334002e-05,
+      "loss": 1.7359,
+      "step": 564
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.954773556394488e-05,
+      "loss": 1.3365,
+      "step": 565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.95452577763122e-05,
+      "loss": 1.5416,
+      "step": 566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9542773250779e-05,
+      "loss": 1.2911,
+      "step": 567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954028198768314e-05,
+      "loss": 1.2085,
+      "step": 568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953778398736341e-05,
+      "loss": 1.0492,
+      "step": 569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953527925015952e-05,
+      "loss": 0.5514,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953276777641213e-05,
+      "loss": 0.5067,
+      "step": 571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953024956646277e-05,
+      "loss": 0.7107,
+      "step": 572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95277246206539e-05,
+      "loss": 1.4096,
+      "step": 573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95251929393289e-05,
+      "loss": 0.9978,
+      "step": 574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952265452283207e-05,
+      "loss": 1.2465,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95201093715086e-05,
+      "loss": 0.4878,
+      "step": 576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951755748570467e-05,
+      "loss": 1.0715,
+      "step": 577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951499886576728e-05,
+      "loss": 0.8784,
+      "step": 578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95124335120444e-05,
+      "loss": 1.1077,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950986142488491e-05,
+      "loss": 0.8527,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950728260463859e-05,
+      "loss": 1.3972,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950469705165617e-05,
+      "loss": 1.1219,
+      "step": 582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950210476628923e-05,
+      "loss": 1.0252,
+      "step": 583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949950574889035e-05,
+      "loss": 1.206,
+      "step": 584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949689999981295e-05,
+      "loss": 0.9822,
+      "step": 585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949428751941143e-05,
+      "loss": 1.1797,
+      "step": 586
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949166830804107e-05,
+      "loss": 1.2972,
+      "step": 587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948904236605803e-05,
+      "loss": 0.9469,
+      "step": 588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948640969381945e-05,
+      "loss": 0.6194,
+      "step": 589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948377029168338e-05,
+      "loss": 1.5669,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948112416000874e-05,
+      "loss": 0.7401,
+      "step": 591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94784712991554e-05,
+      "loss": 1.4454,
+      "step": 592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947581170948412e-05,
+      "loss": 0.9153,
+      "step": 593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947314539135662e-05,
+      "loss": 0.7473,
+      "step": 594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947047234513547e-05,
+      "loss": 0.8094,
+      "step": 595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.946779257118422e-05,
+      "loss": 1.0856,
+      "step": 596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94651060698673e-05,
+      "loss": 0.7135,
+      "step": 597
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.946241284155004e-05,
+      "loss": 1.0856,
+      "step": 598
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.945971288659874e-05,
+      "loss": 1.1863,
+      "step": 599
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.945700620538055e-05,
+      "loss": 1.4036,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.945429279826357e-05,
+      "loss": 0.8333,
+      "step": 601
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.945157266561682e-05,
+      "loss": 1.4825,
+      "step": 602
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.944884580781021e-05,
+      "loss": 1.4864,
+      "step": 603
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.944611222521459e-05,
+      "loss": 1.2899,
+      "step": 604
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.944337191820172e-05,
+      "loss": 0.8993,
+      "step": 605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.944062488714425e-05,
+      "loss": 1.3244,
+      "step": 606
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.943787113241577e-05,
+      "loss": 1.2205,
+      "step": 607
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.943511065439079e-05,
+      "loss": 0.6993,
+      "step": 608
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.943234345344469e-05,
+      "loss": 0.7718,
+      "step": 609
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.942956952995381e-05,
+      "loss": 1.3843,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94267888842954e-05,
+      "loss": 1.4947,
+      "step": 611
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94240015168476e-05,
+      "loss": 1.5738,
+      "step": 612
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.942120742798951e-05,
+      "loss": 0.9434,
+      "step": 613
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.941840661810106e-05,
+      "loss": 0.5556,
+      "step": 614
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.941559908756318e-05,
+      "loss": 0.6567,
+      "step": 615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.941278483675765e-05,
+      "loss": 0.8071,
+      "step": 616
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.940996386606724e-05,
+      "loss": 0.6088,
+      "step": 617
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.940713617587554e-05,
+      "loss": 1.466,
+      "step": 618
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.940430176656714e-05,
+      "loss": 1.1442,
+      "step": 619
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.940146063852748e-05,
+      "loss": 1.706,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.939861279214294e-05,
+      "loss": 0.7533,
+      "step": 621
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.939575822780082e-05,
+      "loss": 1.0296,
+      "step": 622
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.939289694588933e-05,
+      "loss": 0.7621,
+      "step": 623
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.939002894679759e-05,
+      "loss": 1.1627,
+      "step": 624
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.938715423091563e-05,
+      "loss": 0.823,
+      "step": 625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.938427279863439e-05,
+      "loss": 1.2383,
+      "step": 626
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.938138465034574e-05,
+      "loss": 0.9729,
+      "step": 627
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.937848978644244e-05,
+      "loss": 1.0935,
+      "step": 628
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.937558820731818e-05,
+      "loss": 1.6986,
+      "step": 629
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.937267991336757e-05,
+      "loss": 0.9049,
+      "step": 630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.936976490498613e-05,
+      "loss": 1.4906,
+      "step": 631
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.936684318257026e-05,
+      "loss": 0.7642,
+      "step": 632
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.936391474651732e-05,
+      "loss": 0.8279,
+      "step": 633
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.936097959722555e-05,
+      "loss": 0.7628,
+      "step": 634
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.935803773509411e-05,
+      "loss": 1.2688,
+      "step": 635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.93550891605231e-05,
+      "loss": 1.4385,
+      "step": 636
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.93521338739135e-05,
+      "loss": 0.9901,
+      "step": 637
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.934917187566721e-05,
+      "loss": 0.7324,
+      "step": 638
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.934620316618706e-05,
+      "loss": 1.4282,
+      "step": 639
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.934322774587674e-05,
+      "loss": 1.2066,
+      "step": 640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.934024561514095e-05,
+      "loss": 0.7095,
+      "step": 641
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.933725677438522e-05,
+      "loss": 0.8802,
+      "step": 642
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.933426122401599e-05,
+      "loss": 0.5751,
+      "step": 643
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.933125896444066e-05,
+      "loss": 0.6812,
+      "step": 644
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.932824999606754e-05,
+      "loss": 0.9225,
+      "step": 645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.932523431930581e-05,
+      "loss": 1.1778,
+      "step": 646
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.932221193456561e-05,
+      "loss": 1.5258,
+      "step": 647
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.931918284225794e-05,
+      "loss": 1.1581,
+      "step": 648
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.931614704279475e-05,
+      "loss": 1.5132,
+      "step": 649
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.93131045365889e-05,
+      "loss": 1.2052,
+      "step": 650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.931005532405416e-05,
+      "loss": 0.6995,
+      "step": 651
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.930699940560519e-05,
+      "loss": 0.5955,
+      "step": 652
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.93039367816576e-05,
+      "loss": 1.3731,
+      "step": 653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930086745262788e-05,
+      "loss": 0.9155,
+      "step": 654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929779141893343e-05,
+      "loss": 1.8638,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92947086809926e-05,
+      "loss": 0.9568,
+      "step": 656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92916192392246e-05,
+      "loss": 0.6935,
+      "step": 657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928852309404962e-05,
+      "loss": 1.3102,
+      "step": 658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928542024588868e-05,
+      "loss": 1.3675,
+      "step": 659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928231069516373e-05,
+      "loss": 1.3565,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927919444229774e-05,
+      "loss": 1.2645,
+      "step": 661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927607148771443e-05,
+      "loss": 1.2049,
+      "step": 662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927294183183852e-05,
+      "loss": 1.4003,
+      "step": 663
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926980547509564e-05,
+      "loss": 1.0584,
+      "step": 664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92666624179123e-05,
+      "loss": 0.8861,
+      "step": 665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926351266071596e-05,
+      "loss": 1.5824,
+      "step": 666
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926035620393498e-05,
+      "loss": 1.1526,
+      "step": 667
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925719304799858e-05,
+      "loss": 1.4304,
+      "step": 668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925402319333699e-05,
+      "loss": 1.6985,
+      "step": 669
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925084664038123e-05,
+      "loss": 1.2224,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924766338956334e-05,
+      "loss": 1.3029,
+      "step": 671
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924447344131621e-05,
+      "loss": 1.4367,
+      "step": 672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924127679607367e-05,
+      "loss": 1.0282,
+      "step": 673
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923807345427042e-05,
+      "loss": 1.4001,
+      "step": 674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923486341634212e-05,
+      "loss": 1.6722,
+      "step": 675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923164668272531e-05,
+      "loss": 0.8197,
+      "step": 676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922842325385746e-05,
+      "loss": 1.3708,
+      "step": 677
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922519313017693e-05,
+      "loss": 1.157,
+      "step": 678
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9221956312123e-05,
+      "loss": 1.1068,
+      "step": 679
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921871280013586e-05,
+      "loss": 1.191,
+      "step": 680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921546259465662e-05,
+      "loss": 0.5613,
+      "step": 681
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921220569612728e-05,
+      "loss": 1.0909,
+      "step": 682
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.920894210499077e-05,
+      "loss": 1.246,
+      "step": 683
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.920567182169091e-05,
+      "loss": 1.3475,
+      "step": 684
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.920239484667246e-05,
+      "loss": 0.6819,
+      "step": 685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.919911118038107e-05,
+      "loss": 1.1234,
+      "step": 686
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.919582082326328e-05,
+      "loss": 0.7206,
+      "step": 687
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.919252377576658e-05,
+      "loss": 1.7297,
+      "step": 688
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.918922003833934e-05,
+      "loss": 1.2568,
+      "step": 689
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.918590961143088e-05,
+      "loss": 1.4557,
+      "step": 690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.918259249549135e-05,
+      "loss": 1.1766,
+      "step": 691
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.91792686909719e-05,
+      "loss": 1.4406,
+      "step": 692
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.917593819832456e-05,
+      "loss": 1.4695,
+      "step": 693
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.91726010180022e-05,
+      "loss": 1.3058,
+      "step": 694
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.916925715045872e-05,
+      "loss": 1.3999,
+      "step": 695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.916590659614886e-05,
+      "loss": 1.2502,
+      "step": 696
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.916254935552827e-05,
+      "loss": 0.7439,
+      "step": 697
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.915918542905349e-05,
+      "loss": 1.0708,
+      "step": 698
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.915581481718204e-05,
+      "loss": 1.5537,
+      "step": 699
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.915243752037227e-05,
+      "loss": 0.6507,
+      "step": 700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.914905353908351e-05,
+      "loss": 0.5866,
+      "step": 701
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.914566287377593e-05,
+      "loss": 0.687,
+      "step": 702
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.914226552491066e-05,
+      "loss": 1.1859,
+      "step": 703
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.913886149294973e-05,
+      "loss": 1.3395,
+      "step": 704
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.913545077835604e-05,
+      "loss": 1.0804,
+      "step": 705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.913203338159347e-05,
+      "loss": 1.026,
+      "step": 706
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.912860930312676e-05,
+      "loss": 1.5098,
+      "step": 707
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.912517854342154e-05,
+      "loss": 1.2779,
+      "step": 708
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.912174110294438e-05,
+      "loss": 0.938,
+      "step": 709
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.911829698216279e-05,
+      "loss": 1.3275,
+      "step": 710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.911484618154513e-05,
+      "loss": 0.8508,
+      "step": 711
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.911138870156068e-05,
+      "loss": 1.234,
+      "step": 712
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.910792454267965e-05,
+      "loss": 0.8204,
+      "step": 713
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.910445370537316e-05,
+      "loss": 1.2721,
+      "step": 714
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.91009761901132e-05,
+      "loss": 1.5441,
+      "step": 715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.909749199737271e-05,
+      "loss": 0.8533,
+      "step": 716
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.909400112762552e-05,
+      "loss": 0.9627,
+      "step": 717
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.909050358134637e-05,
+      "loss": 1.0365,
+      "step": 718
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.908699935901092e-05,
+      "loss": 1.0355,
+      "step": 719
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.90834884610957e-05,
+      "loss": 1.6009,
+      "step": 720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.907997088807822e-05,
+      "loss": 1.3295,
+      "step": 721
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.907644664043678e-05,
+      "loss": 1.4153,
+      "step": 722
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.907291571865072e-05,
+      "loss": 1.2117,
+      "step": 723
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.90693781232002e-05,
+      "loss": 1.0373,
+      "step": 724
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.906583385456634e-05,
+      "loss": 1.1905,
+      "step": 725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.906228291323111e-05,
+      "loss": 1.3493,
+      "step": 726
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.905872529967744e-05,
+      "loss": 1.2636,
+      "step": 727
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.905516101438914e-05,
+      "loss": 1.2191,
+      "step": 728
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.905159005785094e-05,
+      "loss": 1.1226,
+      "step": 729
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.904801243054847e-05,
+      "loss": 0.7128,
+      "step": 730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.904442813296827e-05,
+      "loss": 1.094,
+      "step": 731
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.904083716559778e-05,
+      "loss": 1.412,
+      "step": 732
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.903723952892535e-05,
+      "loss": 1.1606,
+      "step": 733
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.903363522344024e-05,
+      "loss": 1.1101,
+      "step": 734
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.903002424963265e-05,
+      "loss": 1.4023,
+      "step": 735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.902640660799361e-05,
+      "loss": 0.7062,
+      "step": 736
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.902278229901512e-05,
+      "loss": 1.4683,
+      "step": 737
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.901915132319006e-05,
+      "loss": 0.7432,
+      "step": 738
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.901551368101225e-05,
+      "loss": 1.4027,
+      "step": 739
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.901186937297635e-05,
+      "loss": 1.355,
+      "step": 740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.900821839957799e-05,
+      "loss": 0.5342,
+      "step": 741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900456076131369e-05,
+      "loss": 1.6817,
+      "step": 742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900089645868086e-05,
+      "loss": 1.3817,
+      "step": 743
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899722549217783e-05,
+      "loss": 1.3886,
+      "step": 744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899354786230381e-05,
+      "loss": 0.9074,
+      "step": 745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898986356955897e-05,
+      "loss": 1.4865,
+      "step": 746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898617261444436e-05,
+      "loss": 1.1739,
+      "step": 747
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898247499746189e-05,
+      "loss": 1.5633,
+      "step": 748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897877071911445e-05,
+      "loss": 1.2765,
+      "step": 749
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89750597799058e-05,
+      "loss": 1.1617,
+      "step": 750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89713421803406e-05,
+      "loss": 1.3865,
+      "step": 751
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896761792092443e-05,
+      "loss": 1.1067,
+      "step": 752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896388700216377e-05,
+      "loss": 1.3553,
+      "step": 753
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.8960149424566e-05,
+      "loss": 1.8459,
+      "step": 754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895640518863941e-05,
+      "loss": 1.2136,
+      "step": 755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89526542948932e-05,
+      "loss": 1.1385,
+      "step": 756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89488967438375e-05,
+      "loss": 1.3495,
+      "step": 757
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894513253598326e-05,
+      "loss": 1.4127,
+      "step": 758
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894136167184243e-05,
+      "loss": 1.0627,
+      "step": 759
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893758415192783e-05,
+      "loss": 1.5481,
+      "step": 760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893379997675317e-05,
+      "loss": 0.6776,
+      "step": 761
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893000914683309e-05,
+      "loss": 0.8183,
+      "step": 762
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892621166268312e-05,
+      "loss": 1.3707,
+      "step": 763
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892240752481969e-05,
+      "loss": 0.6617,
+      "step": 764
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891859673376015e-05,
+      "loss": 0.9674,
+      "step": 765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891477929002275e-05,
+      "loss": 0.4237,
+      "step": 766
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891095519412664e-05,
+      "loss": 1.4323,
+      "step": 767
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890712444659187e-05,
+      "loss": 0.5574,
+      "step": 768
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890328704793943e-05,
+      "loss": 1.0745,
+      "step": 769
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.889944299869113e-05,
+      "loss": 0.7244,
+      "step": 770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.889559229936981e-05,
+      "loss": 1.2455,
+      "step": 771
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.88917349504991e-05,
+      "loss": 1.4272,
+      "step": 772
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.88878709526036e-05,
+      "loss": 0.8442,
+      "step": 773
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.888400030620876e-05,
+      "loss": 1.3159,
+      "step": 774
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.888012301184102e-05,
+      "loss": 0.5708,
+      "step": 775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.887623907002763e-05,
+      "loss": 1.0275,
+      "step": 776
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.887234848129682e-05,
+      "loss": 1.0915,
+      "step": 777
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.886845124617765e-05,
+      "loss": 0.7544,
+      "step": 778
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.886454736520017e-05,
+      "loss": 1.1113,
+      "step": 779
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.886063683889524e-05,
+      "loss": 1.1607,
+      "step": 780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.885671966779471e-05,
+      "loss": 1.4713,
+      "step": 781
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.885279585243127e-05,
+      "loss": 1.2423,
+      "step": 782
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.884886539333857e-05,
+      "loss": 1.2545,
+      "step": 783
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.88449282910511e-05,
+      "loss": 1.2634,
+      "step": 784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.88409845461043e-05,
+      "loss": 0.7507,
+      "step": 785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.883703415903448e-05,
+      "loss": 0.7488,
+      "step": 786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.88330771303789e-05,
+      "loss": 1.2322,
+      "step": 787
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.882911346067568e-05,
+      "loss": 1.1541,
+      "step": 788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.882514315046387e-05,
+      "loss": 1.4526,
+      "step": 789
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.88211662002834e-05,
+      "loss": 1.1324,
+      "step": 790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.881718261067512e-05,
+      "loss": 0.7131,
+      "step": 791
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.881319238218077e-05,
+      "loss": 1.1655,
+      "step": 792
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.880919551534302e-05,
+      "loss": 0.9722,
+      "step": 793
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.88051920107054e-05,
+      "loss": 1.0475,
+      "step": 794
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.880118186881239e-05,
+      "loss": 1.2413,
+      "step": 795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.879716509020934e-05,
+      "loss": 0.9867,
+      "step": 796
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.879314167544248e-05,
+      "loss": 1.209,
+      "step": 797
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.878911162505901e-05,
+      "loss": 1.026,
+      "step": 798
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.878507493960701e-05,
+      "loss": 1.1504,
+      "step": 799
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.87810316196354e-05,
+      "loss": 1.0953,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.877698166569409e-05,
+      "loss": 1.6145,
+      "step": 801
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.877292507833382e-05,
+      "loss": 0.5722,
+      "step": 802
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.87688618581063e-05,
+      "loss": 1.1267,
+      "step": 803
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.87647920055641e-05,
+      "loss": 1.2298,
+      "step": 804
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.876071552126067e-05,
+      "loss": 1.4486,
+      "step": 805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.875663240575042e-05,
+      "loss": 0.6025,
+      "step": 806
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.875254265958862e-05,
+      "loss": 0.9344,
+      "step": 807
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.874844628333147e-05,
+      "loss": 1.3513,
+      "step": 808
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.874434327753604e-05,
+      "loss": 1.208,
+      "step": 809
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.874023364276031e-05,
+      "loss": 1.1289,
+      "step": 810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.87361173795632e-05,
+      "loss": 1.3382,
+      "step": 811
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.873199448850447e-05,
+      "loss": 1.1454,
+      "step": 812
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.872786497014483e-05,
+      "loss": 1.0482,
+      "step": 813
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.872372882504586e-05,
+      "loss": 1.2976,
+      "step": 814
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.871958605377008e-05,
+      "loss": 1.2127,
+      "step": 815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.871543665688084e-05,
+      "loss": 1.4977,
+      "step": 816
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.871128063494249e-05,
+      "loss": 1.1701,
+      "step": 817
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.870711798852019e-05,
+      "loss": 1.2422,
+      "step": 818
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.870294871818004e-05,
+      "loss": 1.0877,
+      "step": 819
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.869877282448905e-05,
+      "loss": 1.1912,
+      "step": 820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.869459030801513e-05,
+      "loss": 0.816,
+      "step": 821
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.869040116932708e-05,
+      "loss": 1.415,
+      "step": 822
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.868620540899458e-05,
+      "loss": 0.7215,
+      "step": 823
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.868200302758825e-05,
+      "loss": 1.769,
+      "step": 824
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.867779402567958e-05,
+      "loss": 1.4871,
+      "step": 825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.867357840384097e-05,
+      "loss": 1.3793,
+      "step": 826
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.866935616264575e-05,
+      "loss": 1.2764,
+      "step": 827
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.866512730266809e-05,
+      "loss": 0.6827,
+      "step": 828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866089182448311e-05,
+      "loss": 1.6019,
+      "step": 829
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865664972866682e-05,
+      "loss": 1.3976,
+      "step": 830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865240101579611e-05,
+      "loss": 1.5852,
+      "step": 831
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864814568644878e-05,
+      "loss": 0.4911,
+      "step": 832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864388374120355e-05,
+      "loss": 0.9404,
+      "step": 833
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863961518064e-05,
+      "loss": 0.7144,
+      "step": 834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863534000533867e-05,
+      "loss": 1.2858,
+      "step": 835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863105821588093e-05,
+      "loss": 1.2566,
+      "step": 836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862676981284907e-05,
+      "loss": 1.1244,
+      "step": 837
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862247479682633e-05,
+      "loss": 0.9796,
+      "step": 838
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861817316839676e-05,
+      "loss": 1.0903,
+      "step": 839
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861386492814542e-05,
+      "loss": 1.5415,
+      "step": 840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860955007665818e-05,
+      "loss": 1.4509,
+      "step": 841
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860522861452181e-05,
+      "loss": 0.4796,
+      "step": 842
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860090054232405e-05,
+      "loss": 1.1375,
+      "step": 843
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859656586065346e-05,
+      "loss": 0.4782,
+      "step": 844
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859222457009957e-05,
+      "loss": 1.2125,
+      "step": 845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858787667125276e-05,
+      "loss": 1.1038,
+      "step": 846
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858352216470429e-05,
+      "loss": 0.979,
+      "step": 847
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857916105104639e-05,
+      "loss": 1.0542,
+      "step": 848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857479333087214e-05,
+      "loss": 0.5461,
+      "step": 849
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857041900477552e-05,
+      "loss": 0.9702,
+      "step": 850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856603807335143e-05,
+      "loss": 1.3991,
+      "step": 851
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856165053719564e-05,
+      "loss": 0.5651,
+      "step": 852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855725639690483e-05,
+      "loss": 1.0757,
+      "step": 853
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855285565307659e-05,
+      "loss": 1.4828,
+      "step": 854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85484483063094e-05,
+      "loss": 1.1449,
+      "step": 855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854403435720261e-05,
+      "loss": 1.0361,
+      "step": 856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853961380635654e-05,
+      "loss": 0.6657,
+      "step": 857
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853518665437234e-05,
+      "loss": 0.6396,
+      "step": 858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853075290185206e-05,
+      "loss": 0.7999,
+      "step": 859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85263125493987e-05,
+      "loss": 1.1527,
+      "step": 860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.852186559761611e-05,
+      "loss": 1.1965,
+      "step": 861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.851741204710904e-05,
+      "loss": 0.5702,
+      "step": 862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.851295189848318e-05,
+      "loss": 1.1225,
+      "step": 863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.850848515234508e-05,
+      "loss": 1.2772,
+      "step": 864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.850401180930217e-05,
+      "loss": 1.3445,
+      "step": 865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.849953186996282e-05,
+      "loss": 0.9571,
+      "step": 866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.849504533493627e-05,
+      "loss": 1.5727,
+      "step": 867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.849055220483267e-05,
+      "loss": 1.4245,
+      "step": 868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.848605248026307e-05,
+      "loss": 1.3858,
+      "step": 869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.848154616183942e-05,
+      "loss": 1.4785,
+      "step": 870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.847703325017452e-05,
+      "loss": 1.1358,
+      "step": 871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.847251374588213e-05,
+      "loss": 1.0628,
+      "step": 872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.846798764957687e-05,
+      "loss": 1.3597,
+      "step": 873
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.846345496187427e-05,
+      "loss": 1.5676,
+      "step": 874
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.845891568339074e-05,
+      "loss": 1.0949,
+      "step": 875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.845436981474362e-05,
+      "loss": 1.6245,
+      "step": 876
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.844981735655113e-05,
+      "loss": 1.2093,
+      "step": 877
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.844525830943233e-05,
+      "loss": 0.5738,
+      "step": 878
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.844069267400729e-05,
+      "loss": 1.386,
+      "step": 879
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.843612045089686e-05,
+      "loss": 1.3512,
+      "step": 880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.843154164072288e-05,
+      "loss": 1.0136,
+      "step": 881
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.842695624410802e-05,
+      "loss": 0.9069,
+      "step": 882
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.842236426167588e-05,
+      "loss": 1.1216,
+      "step": 883
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.841776569405095e-05,
+      "loss": 0.6272,
+      "step": 884
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.84131605418586e-05,
+      "loss": 0.72,
+      "step": 885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.84085488057251e-05,
+      "loss": 1.1488,
+      "step": 886
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.840393048627767e-05,
+      "loss": 0.85,
+      "step": 887
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.839930558414432e-05,
+      "loss": 0.7965,
+      "step": 888
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.839467409995403e-05,
+      "loss": 0.7927,
+      "step": 889
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.839003603433667e-05,
+      "loss": 1.2185,
+      "step": 890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8385391387923e-05,
+      "loss": 1.284,
+      "step": 891
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.838074016134464e-05,
+      "loss": 0.8616,
+      "step": 892
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.837608235523416e-05,
+      "loss": 1.4601,
+      "step": 893
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.837141797022498e-05,
+      "loss": 0.8338,
+      "step": 894
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.836674700695143e-05,
+      "loss": 1.3842,
+      "step": 895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.836206946604876e-05,
+      "loss": 1.1217,
+      "step": 896
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.835738534815307e-05,
+      "loss": 1.3646,
+      "step": 897
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.835269465390137e-05,
+      "loss": 1.4525,
+      "step": 898
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.834799738393161e-05,
+      "loss": 1.2791,
+      "step": 899
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.834329353888254e-05,
+      "loss": 1.656,
+      "step": 900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.83385831193939e-05,
+      "loss": 0.6471,
+      "step": 901
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.833386612610628e-05,
+      "loss": 1.3343,
+      "step": 902
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.832914255966115e-05,
+      "loss": 0.8403,
+      "step": 903
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.832441242070088e-05,
+      "loss": 1.1793,
+      "step": 904
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.831967570986877e-05,
+      "loss": 1.4122,
+      "step": 905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8314932427809e-05,
+      "loss": 1.2418,
+      "step": 906
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.83101825751666e-05,
+      "loss": 1.5355,
+      "step": 907
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.830542615258753e-05,
+      "loss": 1.2902,
+      "step": 908
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.830066316071864e-05,
+      "loss": 0.8062,
+      "step": 909
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.82958936002077e-05,
+      "loss": 0.9269,
+      "step": 910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.829111747170332e-05,
+      "loss": 1.3984,
+      "step": 911
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.828633477585501e-05,
+      "loss": 0.44,
+      "step": 912
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.828154551331324e-05,
+      "loss": 0.8843,
+      "step": 913
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.827674968472929e-05,
+      "loss": 0.736,
+      "step": 914
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.827194729075538e-05,
+      "loss": 0.5785,
+      "step": 915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826713833204462e-05,
+      "loss": 1.264,
+      "step": 916
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826232280925098e-05,
+      "loss": 1.3228,
+      "step": 917
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825750072302937e-05,
+      "loss": 1.4425,
+      "step": 918
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825267207403555e-05,
+      "loss": 1.1469,
+      "step": 919
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82478368629262e-05,
+      "loss": 1.4007,
+      "step": 920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82429950903589e-05,
+      "loss": 1.5237,
+      "step": 921
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823814675699207e-05,
+      "loss": 1.3564,
+      "step": 922
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823329186348509e-05,
+      "loss": 1.064,
+      "step": 923
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822843041049819e-05,
+      "loss": 1.1765,
+      "step": 924
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822356239869249e-05,
+      "loss": 1.1569,
+      "step": 925
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821868782873004e-05,
+      "loss": 0.6835,
+      "step": 926
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821380670127374e-05,
+      "loss": 1.0799,
+      "step": 927
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82089190169874e-05,
+      "loss": 0.9833,
+      "step": 928
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820402477653572e-05,
+      "loss": 1.2697,
+      "step": 929
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819912398058429e-05,
+      "loss": 0.8313,
+      "step": 930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819421662979959e-05,
+      "loss": 0.7932,
+      "step": 931
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818930272484902e-05,
+      "loss": 1.1768,
+      "step": 932
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81843822664008e-05,
+      "loss": 0.5268,
+      "step": 933
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817945525512414e-05,
+      "loss": 0.9847,
+      "step": 934
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817452169168906e-05,
+      "loss": 1.2513,
+      "step": 935
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816958157676648e-05,
+      "loss": 1.0429,
+      "step": 936
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816463491102827e-05,
+      "loss": 0.6801,
+      "step": 937
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815968169514712e-05,
+      "loss": 0.7584,
+      "step": 938
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815472192979666e-05,
+      "loss": 1.2141,
+      "step": 939
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814975561565139e-05,
+      "loss": 1.2645,
+      "step": 940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814478275338667e-05,
+      "loss": 0.9273,
+      "step": 941
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813980334367885e-05,
+      "loss": 1.23,
+      "step": 942
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813481738720506e-05,
+      "loss": 1.5339,
+      "step": 943
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812982488464337e-05,
+      "loss": 0.6888,
+      "step": 944
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812482583667273e-05,
+      "loss": 1.382,
+      "step": 945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.811982024397299e-05,
+      "loss": 0.6064,
+      "step": 946
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81148081072249e-05,
+      "loss": 1.367,
+      "step": 947
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.810978942711007e-05,
+      "loss": 0.7157,
+      "step": 948
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8104764204311e-05,
+      "loss": 1.4035,
+      "step": 949
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.809973243951113e-05,
+      "loss": 1.4037,
+      "step": 950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.809469413339471e-05,
+      "loss": 0.7921,
+      "step": 951
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.808964928664698e-05,
+      "loss": 1.2334,
+      "step": 952
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.808459789995397e-05,
+      "loss": 1.2136,
+      "step": 953
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.807953997400264e-05,
+      "loss": 0.7776,
+      "step": 954
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.807447550948088e-05,
+      "loss": 0.8362,
+      "step": 955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.806940450707738e-05,
+      "loss": 1.1311,
+      "step": 956
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.806432696748182e-05,
+      "loss": 1.3836,
+      "step": 957
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.805924289138469e-05,
+      "loss": 0.7817,
+      "step": 958
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.80541522794774e-05,
+      "loss": 1.2884,
+      "step": 959
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.804905513245226e-05,
+      "loss": 1.5801,
+      "step": 960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.804395145100245e-05,
+      "loss": 1.2666,
+      "step": 961
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.803884123582204e-05,
+      "loss": 1.3838,
+      "step": 962
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.803372448760599e-05,
+      "loss": 0.753,
+      "step": 963
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.802860120705016e-05,
+      "loss": 0.5846,
+      "step": 964
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.802347139485129e-05,
+      "loss": 1.3518,
+      "step": 965
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.801833505170701e-05,
+      "loss": 1.1426,
+      "step": 966
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.801319217831582e-05,
+      "loss": 1.5856,
+      "step": 967
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.800804277537714e-05,
+      "loss": 0.7619,
+      "step": 968
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.800288684359127e-05,
+      "loss": 1.2201,
+      "step": 969
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.799772438365936e-05,
+      "loss": 1.0553,
+      "step": 970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.799255539628351e-05,
+      "loss": 1.3948,
+      "step": 971
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.798737988216665e-05,
+      "loss": 1.011,
+      "step": 972
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.798219784201264e-05,
+      "loss": 1.0506,
+      "step": 973
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.797700927652622e-05,
+      "loss": 1.017,
+      "step": 974
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.797181418641297e-05,
+      "loss": 0.6953,
+      "step": 975
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.796661257237944e-05,
+      "loss": 0.9801,
+      "step": 976
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.7961404435133e-05,
+      "loss": 0.8359,
+      "step": 977
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.795618977538193e-05,
+      "loss": 1.186,
+      "step": 978
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.79509685938354e-05,
+      "loss": 1.3964,
+      "step": 979
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.794574089120345e-05,
+      "loss": 1.3429,
+      "step": 980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.794050666819705e-05,
+      "loss": 0.9803,
+      "step": 981
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.793526592552801e-05,
+      "loss": 1.4323,
+      "step": 982
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.793001866390903e-05,
+      "loss": 1.1613,
+      "step": 983
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.792476488405375e-05,
+      "loss": 0.6444,
+      "step": 984
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.791950458667662e-05,
+      "loss": 0.8816,
+      "step": 985
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.791423777249303e-05,
+      "loss": 0.9504,
+      "step": 986
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.790896444221924e-05,
+      "loss": 1.4082,
+      "step": 987
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.79036845965724e-05,
+      "loss": 1.3982,
+      "step": 988
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.789839823627052e-05,
+      "loss": 0.4711,
+      "step": 989
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.789310536203253e-05,
+      "loss": 1.2809,
+      "step": 990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.788780597457825e-05,
+      "loss": 1.0997,
+      "step": 991
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.788250007462835e-05,
+      "loss": 1.6317,
+      "step": 992
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.78771876629044e-05,
+      "loss": 0.615,
+      "step": 993
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.787186874012888e-05,
+      "loss": 1.059,
+      "step": 994
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.786654330702514e-05,
+      "loss": 1.5444,
+      "step": 995
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.786121136431738e-05,
+      "loss": 0.8459,
+      "step": 996
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.785587291273074e-05,
+      "loss": 0.879,
+      "step": 997
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.785052795299123e-05,
+      "loss": 0.9301,
+      "step": 998
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.78451764858257e-05,
+      "loss": 1.083,
+      "step": 999
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.783981851196197e-05,
+      "loss": 1.4806,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.783445403212867e-05,
+      "loss": 1.4506,
+      "step": 1001
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.782908304705533e-05,
+      "loss": 1.1542,
+      "step": 1002
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782370555747241e-05,
+      "loss": 0.6584,
+      "step": 1003
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78183215641112e-05,
+      "loss": 1.4159,
+      "step": 1004
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781293106770388e-05,
+      "loss": 1.3748,
+      "step": 1005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780753406898357e-05,
+      "loss": 0.8346,
+      "step": 1006
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78021305686842e-05,
+      "loss": 1.0486,
+      "step": 1007
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779672056754064e-05,
+      "loss": 0.9494,
+      "step": 1008
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77913040662886e-05,
+      "loss": 0.639,
+      "step": 1009
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778588106566473e-05,
+      "loss": 1.1647,
+      "step": 1010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778045156640649e-05,
+      "loss": 1.4535,
+      "step": 1011
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777501556925229e-05,
+      "loss": 1.2065,
+      "step": 1012
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776957307494139e-05,
+      "loss": 1.1359,
+      "step": 1013
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776412408421394e-05,
+      "loss": 0.9047,
+      "step": 1014
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775866859781098e-05,
+      "loss": 1.5589,
+      "step": 1015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775320661647443e-05,
+      "loss": 0.8766,
+      "step": 1016
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774773814094709e-05,
+      "loss": 1.2505,
+      "step": 1017
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774226317197263e-05,
+      "loss": 0.7122,
+      "step": 1018
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773678171029563e-05,
+      "loss": 1.0941,
+      "step": 1019
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773129375666155e-05,
+      "loss": 1.0741,
+      "step": 1020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772579931181669e-05,
+      "loss": 0.8813,
+      "step": 1021
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772029837650831e-05,
+      "loss": 1.588,
+      "step": 1022
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771479095148448e-05,
+      "loss": 0.9321,
+      "step": 1023
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770927703749419e-05,
+      "loss": 1.2114,
+      "step": 1024
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770375663528731e-05,
+      "loss": 1.3409,
+      "step": 1025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769822974561458e-05,
+      "loss": 1.4803,
+      "step": 1026
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769269636922762e-05,
+      "loss": 0.5387,
+      "step": 1027
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768715650687896e-05,
+      "loss": 1.1414,
+      "step": 1028
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768161015932198e-05,
+      "loss": 1.4061,
+      "step": 1029
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767605732731095e-05,
+      "loss": 1.7186,
+      "step": 1030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767049801160104e-05,
+      "loss": 0.7723,
+      "step": 1031
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766493221294828e-05,
+      "loss": 1.3866,
+      "step": 1032
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76593599321096e-05,
+      "loss": 0.8718,
+      "step": 1033
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.765378116984277e-05,
+      "loss": 1.4517,
+      "step": 1034
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.764819592690652e-05,
+      "loss": 1.1495,
+      "step": 1035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.764260420406037e-05,
+      "loss": 1.2777,
+      "step": 1036
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76370060020648e-05,
+      "loss": 1.2173,
+      "step": 1037
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.763140132168114e-05,
+      "loss": 0.2127,
+      "step": 1038
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.762579016367156e-05,
+      "loss": 0.8025,
+      "step": 1039
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.762017252879917e-05,
+      "loss": 1.4471,
+      "step": 1040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.761454841782795e-05,
+      "loss": 0.6607,
+      "step": 1041
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.760891783152273e-05,
+      "loss": 0.5799,
+      "step": 1042
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.760328077064924e-05,
+      "loss": 0.8854,
+      "step": 1043
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.759763723597413e-05,
+      "loss": 1.6591,
+      "step": 1044
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.759198722826483e-05,
+      "loss": 0.525,
+      "step": 1045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.758633074828978e-05,
+      "loss": 1.4363,
+      "step": 1046
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.758066779681817e-05,
+      "loss": 1.2427,
+      "step": 1047
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.757499837462017e-05,
+      "loss": 1.0269,
+      "step": 1048
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.756932248246678e-05,
+      "loss": 0.7438,
+      "step": 1049
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.75636401211299e-05,
+      "loss": 1.4203,
+      "step": 1050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.75579512913823e-05,
+      "loss": 1.0289,
+      "step": 1051
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.755225599399762e-05,
+      "loss": 0.9627,
+      "step": 1052
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.754655422975038e-05,
+      "loss": 0.5922,
+      "step": 1053
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.754084599941605e-05,
+      "loss": 1.1798,
+      "step": 1054
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.753513130377086e-05,
+      "loss": 1.2863,
+      "step": 1055
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.752941014359199e-05,
+      "loss": 0.5683,
+      "step": 1056
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.752368251965751e-05,
+      "loss": 1.0222,
+      "step": 1057
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.751794843274633e-05,
+      "loss": 0.8965,
+      "step": 1058
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.751220788363827e-05,
+      "loss": 1.1212,
+      "step": 1059
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.750646087311399e-05,
+      "loss": 1.0996,
+      "step": 1060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.750070740195507e-05,
+      "loss": 1.7908,
+      "step": 1061
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.749494747094396e-05,
+      "loss": 0.7174,
+      "step": 1062
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.748918108086398e-05,
+      "loss": 1.1066,
+      "step": 1063
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.748340823249932e-05,
+      "loss": 0.5175,
+      "step": 1064
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.747762892663506e-05,
+      "loss": 1.1036,
+      "step": 1065
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.747184316405716e-05,
+      "loss": 0.8831,
+      "step": 1066
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.746605094555243e-05,
+      "loss": 0.8164,
+      "step": 1067
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.746025227190862e-05,
+      "loss": 1.2229,
+      "step": 1068
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.745444714391431e-05,
+      "loss": 1.5454,
+      "step": 1069
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.744863556235895e-05,
+      "loss": 1.4736,
+      "step": 1070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.744281752803291e-05,
+      "loss": 1.3785,
+      "step": 1071
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.743699304172737e-05,
+      "loss": 1.1463,
+      "step": 1072
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.743116210423448e-05,
+      "loss": 0.8906,
+      "step": 1073
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.742532471634719e-05,
+      "loss": 1.1509,
+      "step": 1074
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.741948087885935e-05,
+      "loss": 1.53,
+      "step": 1075
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.741363059256572e-05,
+      "loss": 1.4203,
+      "step": 1076
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.740777385826189e-05,
+      "loss": 1.169,
+      "step": 1077
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.740191067674434e-05,
+      "loss": 0.9801,
+      "step": 1078
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.739604104881044e-05,
+      "loss": 1.3952,
+      "step": 1079
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.739016497525843e-05,
+      "loss": 1.1849,
+      "step": 1080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.738428245688743e-05,
+      "loss": 1.3664,
+      "step": 1081
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.737839349449742e-05,
+      "loss": 1.4802,
+      "step": 1082
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.737249808888928e-05,
+      "loss": 1.3108,
+      "step": 1083
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.736659624086475e-05,
+      "loss": 1.2813,
+      "step": 1084
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.736068795122647e-05,
+      "loss": 1.3427,
+      "step": 1085
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.735477322077788e-05,
+      "loss": 0.7857,
+      "step": 1086
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.734885205032342e-05,
+      "loss": 1.6211,
+      "step": 1087
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.73429244406683e-05,
+      "loss": 0.9352,
+      "step": 1088
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.733699039261866e-05,
+      "loss": 1.3735,
+      "step": 1089
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73310499069815e-05,
+      "loss": 1.0273,
+      "step": 1090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732510298456468e-05,
+      "loss": 1.2351,
+      "step": 1091
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731914962617696e-05,
+      "loss": 1.2912,
+      "step": 1092
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731318983262798e-05,
+      "loss": 1.2428,
+      "step": 1093
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730722360472822e-05,
+      "loss": 0.7071,
+      "step": 1094
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730125094328905e-05,
+      "loss": 0.9752,
+      "step": 1095
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729527184912277e-05,
+      "loss": 1.9263,
+      "step": 1096
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728928632304246e-05,
+      "loss": 0.7885,
+      "step": 1097
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728329436586213e-05,
+      "loss": 1.4797,
+      "step": 1098
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727729597839666e-05,
+      "loss": 1.1173,
+      "step": 1099
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727129116146179e-05,
+      "loss": 0.7841,
+      "step": 1100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726527991587416e-05,
+      "loss": 0.6581,
+      "step": 1101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725926224245125e-05,
+      "loss": 0.5575,
+      "step": 1102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725323814201147e-05,
+      "loss": 0.8214,
+      "step": 1103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724720761537403e-05,
+      "loss": 1.432,
+      "step": 1104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724117066335905e-05,
+      "loss": 1.4913,
+      "step": 1105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723512728678754e-05,
+      "loss": 0.6718,
+      "step": 1106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722907748648138e-05,
+      "loss": 0.7256,
+      "step": 1107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722302126326329e-05,
+      "loss": 1.6901,
+      "step": 1108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.721695861795688e-05,
+      "loss": 1.2758,
+      "step": 1109
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.721088955138664e-05,
+      "loss": 0.7155,
+      "step": 1110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720481406437797e-05,
+      "loss": 0.7264,
+      "step": 1111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719873215775706e-05,
+      "loss": 1.2377,
+      "step": 1112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719264383235104e-05,
+      "loss": 1.6372,
+      "step": 1113
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718654908898787e-05,
+      "loss": 0.9805,
+      "step": 1114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718044792849645e-05,
+      "loss": 1.5587,
+      "step": 1115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717434035170646e-05,
+      "loss": 1.2557,
+      "step": 1116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716822635944851e-05,
+      "loss": 0.8216,
+      "step": 1117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716210595255408e-05,
+      "loss": 1.1168,
+      "step": 1118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715597913185551e-05,
+      "loss": 1.1153,
+      "step": 1119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.714984589818603e-05,
+      "loss": 1.1554,
+      "step": 1120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71437062523797e-05,
+      "loss": 0.5872,
+      "step": 1121
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.713756019527149e-05,
+      "loss": 1.2307,
+      "step": 1122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.713140772769724e-05,
+      "loss": 1.515,
+      "step": 1123
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.712524885049367e-05,
+      "loss": 1.2392,
+      "step": 1124
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.711908356449833e-05,
+      "loss": 1.2696,
+      "step": 1125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.711291187054967e-05,
+      "loss": 1.5188,
+      "step": 1126
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7106733769487e-05,
+      "loss": 1.6201,
+      "step": 1127
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.710054926215053e-05,
+      "loss": 1.3772,
+      "step": 1128
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.709435834938133e-05,
+      "loss": 1.2392,
+      "step": 1129
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.70881610320213e-05,
+      "loss": 0.1847,
+      "step": 1130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.708195731091327e-05,
+      "loss": 0.5306,
+      "step": 1131
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.70757471869009e-05,
+      "loss": 1.3034,
+      "step": 1132
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.706953066082873e-05,
+      "loss": 1.124,
+      "step": 1133
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.70633077335422e-05,
+      "loss": 0.7884,
+      "step": 1134
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.705707840588757e-05,
+      "loss": 1.1826,
+      "step": 1135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.705084267871201e-05,
+      "loss": 0.4312,
+      "step": 1136
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.704460055286356e-05,
+      "loss": 1.1708,
+      "step": 1137
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.703835202919108e-05,
+      "loss": 0.8684,
+      "step": 1138
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.703209710854435e-05,
+      "loss": 1.4309,
+      "step": 1139
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.702583579177405e-05,
+      "loss": 1.0036,
+      "step": 1140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.701956807973163e-05,
+      "loss": 0.9368,
+      "step": 1141
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.701329397326948e-05,
+      "loss": 0.6928,
+      "step": 1142
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.700701347324086e-05,
+      "loss": 1.3201,
+      "step": 1143
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.700072658049991e-05,
+      "loss": 1.2221,
+      "step": 1144
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.699443329590155e-05,
+      "loss": 1.3085,
+      "step": 1145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.698813362030169e-05,
+      "loss": 0.9531,
+      "step": 1146
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.698182755455703e-05,
+      "loss": 0.5244,
+      "step": 1147
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.697551509952517e-05,
+      "loss": 1.0955,
+      "step": 1148
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.696919625606455e-05,
+      "loss": 1.5004,
+      "step": 1149
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.696287102503453e-05,
+      "loss": 0.6347,
+      "step": 1150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.69565394072953e-05,
+      "loss": 1.697,
+      "step": 1151
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.695020140370792e-05,
+      "loss": 1.2004,
+      "step": 1152
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.694385701513434e-05,
+      "loss": 0.9202,
+      "step": 1153
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.693750624243733e-05,
+      "loss": 1.3523,
+      "step": 1154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.693114908648063e-05,
+      "loss": 1.3576,
+      "step": 1155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.692478554812871e-05,
+      "loss": 0.7423,
+      "step": 1156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.6918415628247e-05,
+      "loss": 0.9127,
+      "step": 1157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.69120393277018e-05,
+      "loss": 0.6148,
+      "step": 1158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.690565664736022e-05,
+      "loss": 1.1161,
+      "step": 1159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.68992675880903e-05,
+      "loss": 0.634,
+      "step": 1160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.68928721507609e-05,
+      "loss": 1.4233,
+      "step": 1161
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.688647033624179e-05,
+      "loss": 1.2932,
+      "step": 1162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.688006214540356e-05,
+      "loss": 0.8184,
+      "step": 1163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.68736475791177e-05,
+      "loss": 0.6603,
+      "step": 1164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.686722663825656e-05,
+      "loss": 1.6156,
+      "step": 1165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.686079932369334e-05,
+      "loss": 1.3121,
+      "step": 1166
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.685436563630215e-05,
+      "loss": 0.6405,
+      "step": 1167
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.684792557695791e-05,
+      "loss": 0.9139,
+      "step": 1168
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.684147914653646e-05,
+      "loss": 1.3056,
+      "step": 1169
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.683502634591445e-05,
+      "loss": 1.2866,
+      "step": 1170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.682856717596947e-05,
+      "loss": 1.3659,
+      "step": 1171
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.682210163757989e-05,
+      "loss": 1.3746,
+      "step": 1172
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.681562973162503e-05,
+      "loss": 1.2743,
+      "step": 1173
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.6809151458985e-05,
+      "loss": 0.799,
+      "step": 1174
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.680266682054084e-05,
+      "loss": 1.288,
+      "step": 1175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.679617581717442e-05,
+      "loss": 1.1249,
+      "step": 1176
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.678967844976846e-05,
+      "loss": 1.3814,
+      "step": 1177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678317471920659e-05,
+      "loss": 1.2385,
+      "step": 1178
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67766646263733e-05,
+      "loss": 1.0679,
+      "step": 1179
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67701481721539e-05,
+      "loss": 1.2442,
+      "step": 1180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67636253574346e-05,
+      "loss": 0.6445,
+      "step": 1181
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67570961831025e-05,
+      "loss": 1.3477,
+      "step": 1182
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675056065004548e-05,
+      "loss": 1.1784,
+      "step": 1183
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674401875915237e-05,
+      "loss": 1.285,
+      "step": 1184
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673747051131285e-05,
+      "loss": 1.1073,
+      "step": 1185
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67309159074174e-05,
+      "loss": 1.1524,
+      "step": 1186
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672435494835748e-05,
+      "loss": 0.5365,
+      "step": 1187
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671778763502528e-05,
+      "loss": 1.1162,
+      "step": 1188
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671121396831395e-05,
+      "loss": 0.8919,
+      "step": 1189
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670463394911749e-05,
+      "loss": 1.7122,
+      "step": 1190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669804757833072e-05,
+      "loss": 1.4443,
+      "step": 1191
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669145485684936e-05,
+      "loss": 1.5166,
+      "step": 1192
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668485578557001e-05,
+      "loss": 1.1991,
+      "step": 1193
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667825036539008e-05,
+      "loss": 1.2838,
+      "step": 1194
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667163859720789e-05,
+      "loss": 1.3736,
+      "step": 1195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666502048192259e-05,
+      "loss": 1.3988,
+      "step": 1196
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665839602043424e-05,
+      "loss": 1.3468,
+      "step": 1197
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665176521364369e-05,
+      "loss": 1.0266,
+      "step": 1198
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664512806245274e-05,
+      "loss": 0.7954,
+      "step": 1199
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663848456776398e-05,
+      "loss": 0.9506,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66318347304809e-05,
+      "loss": 0.3819,
+      "step": 1201
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662517855150784e-05,
+      "loss": 1.1153,
+      "step": 1202
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661851603175002e-05,
+      "loss": 1.319,
+      "step": 1203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66118471721135e-05,
+      "loss": 0.9269,
+      "step": 1204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660517197350518e-05,
+      "loss": 0.7448,
+      "step": 1205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.65984904368329e-05,
+      "loss": 1.271,
+      "step": 1206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.659180256300528e-05,
+      "loss": 1.1888,
+      "step": 1207
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.658510835293188e-05,
+      "loss": 1.3338,
+      "step": 1208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.657840780752303e-05,
+      "loss": 0.5668,
+      "step": 1209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.657170092768999e-05,
+      "loss": 1.0222,
+      "step": 1210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.656498771434486e-05,
+      "loss": 0.6326,
+      "step": 1211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.655826816840062e-05,
+      "loss": 0.8603,
+      "step": 1212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.655154229077106e-05,
+      "loss": 1.2686,
+      "step": 1213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.65448100823709e-05,
+      "loss": 1.2968,
+      "step": 1214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.653807154411565e-05,
+      "loss": 0.8266,
+      "step": 1215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.653132667692173e-05,
+      "loss": 0.6711,
+      "step": 1216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.652457548170644e-05,
+      "loss": 0.6416,
+      "step": 1217
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.651781795938786e-05,
+      "loss": 1.089,
+      "step": 1218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.651105411088499e-05,
+      "loss": 0.7812,
+      "step": 1219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.65042839371177e-05,
+      "loss": 1.1172,
+      "step": 1220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.649750743900669e-05,
+      "loss": 1.1723,
+      "step": 1221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.64907246174735e-05,
+      "loss": 0.7454,
+      "step": 1222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.648393547344062e-05,
+      "loss": 1.0081,
+      "step": 1223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.647714000783127e-05,
+      "loss": 1.4765,
+      "step": 1224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.647033822156964e-05,
+      "loss": 1.0978,
+      "step": 1225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.646353011558073e-05,
+      "loss": 1.016,
+      "step": 1226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.64567156907904e-05,
+      "loss": 1.3024,
+      "step": 1227
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.64498949481254e-05,
+      "loss": 1.1975,
+      "step": 1228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.644306788851328e-05,
+      "loss": 1.746,
+      "step": 1229
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.643623451288251e-05,
+      "loss": 0.7799,
+      "step": 1230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.642939482216239e-05,
+      "loss": 1.1311,
+      "step": 1231
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.642254881728308e-05,
+      "loss": 1.2543,
+      "step": 1232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.641569649917559e-05,
+      "loss": 0.6726,
+      "step": 1233
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.640883786877183e-05,
+      "loss": 0.8091,
+      "step": 1234
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.64019729270045e-05,
+      "loss": 0.9464,
+      "step": 1235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.639510167480723e-05,
+      "loss": 1.0276,
+      "step": 1236
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.638822411311447e-05,
+      "loss": 1.2587,
+      "step": 1237
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.638134024286151e-05,
+      "loss": 1.7258,
+      "step": 1238
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.637445006498456e-05,
+      "loss": 1.0597,
+      "step": 1239
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.63675535804206e-05,
+      "loss": 1.0155,
+      "step": 1240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.636065079010755e-05,
+      "loss": 1.2465,
+      "step": 1241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.635374169498415e-05,
+      "loss": 0.8883,
+      "step": 1242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.634682629599e-05,
+      "loss": 1.359,
+      "step": 1243
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.633990459406553e-05,
+      "loss": 1.2043,
+      "step": 1244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.63329765901521e-05,
+      "loss": 0.8057,
+      "step": 1245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.632604228519187e-05,
+      "loss": 1.1288,
+      "step": 1246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.631910168012786e-05,
+      "loss": 1.6907,
+      "step": 1247
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.631215477590395e-05,
+      "loss": 1.091,
+      "step": 1248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.63052015734649e-05,
+      "loss": 1.1575,
+      "step": 1249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.629824207375632e-05,
+      "loss": 1.3936,
+      "step": 1250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.629127627772463e-05,
+      "loss": 0.9326,
+      "step": 1251
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.628430418631718e-05,
+      "loss": 1.1419,
+      "step": 1252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.627732580048209e-05,
+      "loss": 1.1318,
+      "step": 1253
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.627034112116846e-05,
+      "loss": 0.9212,
+      "step": 1254
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.62633501493261e-05,
+      "loss": 0.9996,
+      "step": 1255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.625635288590578e-05,
+      "loss": 0.863,
+      "step": 1256
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.624934933185909e-05,
+      "loss": 1.1776,
+      "step": 1257
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.624233948813845e-05,
+      "loss": 1.4211,
+      "step": 1258
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.62353233556972e-05,
+      "loss": 1.216,
+      "step": 1259
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.622830093548949e-05,
+      "loss": 0.9989,
+      "step": 1260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.622127222847033e-05,
+      "loss": 0.8855,
+      "step": 1261
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.621423723559556e-05,
+      "loss": 0.7995,
+      "step": 1262
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.620719595782195e-05,
+      "loss": 1.1177,
+      "step": 1263
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.620014839610704e-05,
+      "loss": 0.6044,
+      "step": 1264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.619309455140928e-05,
+      "loss": 1.3294,
+      "step": 1265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618603442468796e-05,
+      "loss": 0.7721,
+      "step": 1266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617896801690321e-05,
+      "loss": 1.2069,
+      "step": 1267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617189532901603e-05,
+      "loss": 1.34,
+      "step": 1268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.616481636198825e-05,
+      "loss": 1.3948,
+      "step": 1269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615773111678262e-05,
+      "loss": 1.3036,
+      "step": 1270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615063959436265e-05,
+      "loss": 1.4449,
+      "step": 1271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.614354179569279e-05,
+      "loss": 1.0989,
+      "step": 1272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613643772173827e-05,
+      "loss": 1.102,
+      "step": 1273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612932737346521e-05,
+      "loss": 1.5711,
+      "step": 1274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61222107518406e-05,
+      "loss": 1.1821,
+      "step": 1275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611508785783225e-05,
+      "loss": 1.5055,
+      "step": 1276
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.610795869240885e-05,
+      "loss": 0.6842,
+      "step": 1277
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.610082325653993e-05,
+      "loss": 0.4584,
+      "step": 1278
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.609368155119587e-05,
+      "loss": 1.6214,
+      "step": 1279
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608653357734787e-05,
+      "loss": 1.3071,
+      "step": 1280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607937933596809e-05,
+      "loss": 1.5221,
+      "step": 1281
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60722188280294e-05,
+      "loss": 1.2592,
+      "step": 1282
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606505205450563e-05,
+      "loss": 0.781,
+      "step": 1283
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605787901637142e-05,
+      "loss": 1.0593,
+      "step": 1284
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605069971460225e-05,
+      "loss": 1.447,
+      "step": 1285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604351415017448e-05,
+      "loss": 0.7617,
+      "step": 1286
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60363223240653e-05,
+      "loss": 0.6102,
+      "step": 1287
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60291242372528e-05,
+      "loss": 1.3352,
+      "step": 1288
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602191989071583e-05,
+      "loss": 0.6126,
+      "step": 1289
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601470928543418e-05,
+      "loss": 1.0623,
+      "step": 1290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.600749242238845e-05,
+      "loss": 1.3385,
+      "step": 1291
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.600026930256007e-05,
+      "loss": 1.3306,
+      "step": 1292
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.599303992693138e-05,
+      "loss": 1.4937,
+      "step": 1293
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.598580429648552e-05,
+      "loss": 1.2521,
+      "step": 1294
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.597856241220648e-05,
+      "loss": 0.6959,
+      "step": 1295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.597131427507917e-05,
+      "loss": 1.1759,
+      "step": 1296
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.596405988608927e-05,
+      "loss": 0.9625,
+      "step": 1297
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.595679924622334e-05,
+      "loss": 0.7328,
+      "step": 1298
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.59495323564688e-05,
+      "loss": 1.3154,
+      "step": 1299
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.594225921781388e-05,
+      "loss": 1.3584,
+      "step": 1300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.593497983124772e-05,
+      "loss": 0.9512,
+      "step": 1301
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.592769419776026e-05,
+      "loss": 1.4172,
+      "step": 1302
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.592040231834234e-05,
+      "loss": 1.1792,
+      "step": 1303
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.591310419398558e-05,
+      "loss": 1.1065,
+      "step": 1304
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.59057998256825e-05,
+      "loss": 1.322,
+      "step": 1305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.589848921442647e-05,
+      "loss": 1.1176,
+      "step": 1306
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.589117236121168e-05,
+      "loss": 1.5079,
+      "step": 1307
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.58838492670332e-05,
+      "loss": 1.2978,
+      "step": 1308
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.587651993288691e-05,
+      "loss": 0.7151,
+      "step": 1309
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.586918435976958e-05,
+      "loss": 1.1063,
+      "step": 1310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.58618425486788e-05,
+      "loss": 0.7674,
+      "step": 1311
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.585449450061303e-05,
+      "loss": 1.1921,
+      "step": 1312
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.584714021657157e-05,
+      "loss": 0.6956,
+      "step": 1313
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.583977969755456e-05,
+      "loss": 1.0612,
+      "step": 1314
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.583241294456296e-05,
+      "loss": 0.4909,
+      "step": 1315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.582503995859866e-05,
+      "loss": 1.4006,
+      "step": 1316
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.581766074066431e-05,
+      "loss": 0.664,
+      "step": 1317
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.581027529176347e-05,
+      "loss": 0.6128,
+      "step": 1318
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.580288361290053e-05,
+      "loss": 0.6552,
+      "step": 1319
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.57954857050807e-05,
+      "loss": 1.4543,
+      "step": 1320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.578808156931006e-05,
+      "loss": 0.6731,
+      "step": 1321
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.578067120659553e-05,
+      "loss": 1.1749,
+      "step": 1322
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.577325461794491e-05,
+      "loss": 1.1479,
+      "step": 1323
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.576583180436677e-05,
+      "loss": 0.9863,
+      "step": 1324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.575840276687064e-05,
+      "loss": 1.4635,
+      "step": 1325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.575096750646677e-05,
+      "loss": 0.559,
+      "step": 1326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.574352602416634e-05,
+      "loss": 1.5002,
+      "step": 1327
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.573607832098135e-05,
+      "loss": 1.1624,
+      "step": 1328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.572862439792466e-05,
+      "loss": 1.4147,
+      "step": 1329
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.572116425600996e-05,
+      "loss": 0.939,
+      "step": 1330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.571369789625177e-05,
+      "loss": 0.6187,
+      "step": 1331
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.570622531966552e-05,
+      "loss": 0.4979,
+      "step": 1332
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.569874652726742e-05,
+      "loss": 1.3417,
+      "step": 1333
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.569126152007453e-05,
+      "loss": 0.3954,
+      "step": 1334
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.568377029910477e-05,
+      "loss": 1.0601,
+      "step": 1335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.567627286537696e-05,
+      "loss": 1.2029,
+      "step": 1336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.566876921991067e-05,
+      "loss": 1.1129,
+      "step": 1337
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.566125936372637e-05,
+      "loss": 1.0992,
+      "step": 1338
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.565374329784534e-05,
+      "loss": 0.8361,
+      "step": 1339
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.564622102328976e-05,
+      "loss": 1.807,
+      "step": 1340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.563869254108258e-05,
+      "loss": 0.8572,
+      "step": 1341
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.563115785224768e-05,
+      "loss": 1.2969,
+      "step": 1342
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.56236169578097e-05,
+      "loss": 1.5549,
+      "step": 1343
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.561606985879419e-05,
+      "loss": 1.1016,
+      "step": 1344
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.560851655622752e-05,
+      "loss": 1.0262,
+      "step": 1345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.560095705113686e-05,
+      "loss": 0.8302,
+      "step": 1346
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.55933913445503e-05,
+      "loss": 0.8269,
+      "step": 1347
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.558581943749675e-05,
+      "loss": 0.6217,
+      "step": 1348
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.557824133100592e-05,
+      "loss": 1.481,
+      "step": 1349
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.557065702610839e-05,
+      "loss": 1.1522,
+      "step": 1350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.55630665238356e-05,
+      "loss": 1.0824,
+      "step": 1351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555546982521983e-05,
+      "loss": 0.5789,
+      "step": 1352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.554786693129418e-05,
+      "loss": 1.038,
+      "step": 1353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.554025784309259e-05,
+      "loss": 1.3472,
+      "step": 1354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.553264256164989e-05,
+      "loss": 1.0642,
+      "step": 1355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55250210880017e-05,
+      "loss": 1.1532,
+      "step": 1356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.551739342318449e-05,
+      "loss": 1.1918,
+      "step": 1357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55097595682356e-05,
+      "loss": 1.2,
+      "step": 1358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55021195241932e-05,
+      "loss": 1.0277,
+      "step": 1359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.549447329209629e-05,
+      "loss": 1.2563,
+      "step": 1360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54868208729847e-05,
+      "loss": 1.1234,
+      "step": 1361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.547916226789915e-05,
+      "loss": 1.0243,
+      "step": 1362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.547149747788114e-05,
+      "loss": 1.14,
+      "step": 1363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.546382650397307e-05,
+      "loss": 1.152,
+      "step": 1364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.545614934721813e-05,
+      "loss": 1.1804,
+      "step": 1365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54484660086604e-05,
+      "loss": 0.6918,
+      "step": 1366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544077648934475e-05,
+      "loss": 1.2731,
+      "step": 1367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.543308079031693e-05,
+      "loss": 1.1608,
+      "step": 1368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.542537891262352e-05,
+      "loss": 1.4063,
+      "step": 1369
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54176708573119e-05,
+      "loss": 0.9677,
+      "step": 1370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.540995662543038e-05,
+      "loss": 1.2825,
+      "step": 1371
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.540223621802801e-05,
+      "loss": 0.8409,
+      "step": 1372
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.539450963615474e-05,
+      "loss": 1.2556,
+      "step": 1373
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.538677688086137e-05,
+      "loss": 0.9783,
+      "step": 1374
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.537903795319948e-05,
+      "loss": 0.8019,
+      "step": 1375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.537129285422155e-05,
+      "loss": 1.3794,
+      "step": 1376
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.536354158498086e-05,
+      "loss": 1.3768,
+      "step": 1377
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.535578414653153e-05,
+      "loss": 1.1789,
+      "step": 1378
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.534802053992856e-05,
+      "loss": 1.0337,
+      "step": 1379
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.534025076622775e-05,
+      "loss": 1.4086,
+      "step": 1380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533247482648574e-05,
+      "loss": 1.1107,
+      "step": 1381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.532469272176004e-05,
+      "loss": 1.2063,
+      "step": 1382
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.531690445310894e-05,
+      "loss": 0.9013,
+      "step": 1383
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.530911002159164e-05,
+      "loss": 1.1061,
+      "step": 1384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.530130942826811e-05,
+      "loss": 1.2022,
+      "step": 1385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.529350267419923e-05,
+      "loss": 0.9269,
+      "step": 1386
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.528568976044666e-05,
+      "loss": 0.706,
+      "step": 1387
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.527787068807289e-05,
+      "loss": 1.2792,
+      "step": 1388
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.527004545814131e-05,
+      "loss": 1.1639,
+      "step": 1389
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.52622140717161e-05,
+      "loss": 1.1425,
+      "step": 1390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.525437652986228e-05,
+      "loss": 1.2032,
+      "step": 1391
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.524653283364572e-05,
+      "loss": 1.5611,
+      "step": 1392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.523868298413314e-05,
+      "loss": 1.4501,
+      "step": 1393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.523082698239207e-05,
+      "loss": 0.9388,
+      "step": 1394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.522296482949086e-05,
+      "loss": 1.4845,
+      "step": 1395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.521509652649874e-05,
+      "loss": 1.1218,
+      "step": 1396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.520722207448577e-05,
+      "loss": 0.9105,
+      "step": 1397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.519934147452283e-05,
+      "loss": 1.5522,
+      "step": 1398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.519145472768163e-05,
+      "loss": 1.0412,
+      "step": 1399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.518356183503473e-05,
+      "loss": 1.1821,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.517566279765553e-05,
+      "loss": 0.6932,
+      "step": 1401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.516775761661825e-05,
+      "loss": 1.3044,
+      "step": 1402
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.515984629299796e-05,
+      "loss": 1.1443,
+      "step": 1403
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.515192882787057e-05,
+      "loss": 1.2479,
+      "step": 1404
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.51440052223128e-05,
+      "loss": 1.1209,
+      "step": 1405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.513607547740222e-05,
+      "loss": 1.1859,
+      "step": 1406
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.512813959421725e-05,
+      "loss": 0.91,
+      "step": 1407
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.51201975738371e-05,
+      "loss": 1.3782,
+      "step": 1408
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.511224941734186e-05,
+      "loss": 1.6977,
+      "step": 1409
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.510429512581245e-05,
+      "loss": 0.6953,
+      "step": 1410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.509633470033059e-05,
+      "loss": 1.5499,
+      "step": 1411
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.508836814197888e-05,
+      "loss": 0.6408,
+      "step": 1412
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.508039545184071e-05,
+      "loss": 0.5597,
+      "step": 1413
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.507241663100034e-05,
+      "loss": 1.2624,
+      "step": 1414
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.506443168054284e-05,
+      "loss": 0.7785,
+      "step": 1415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.505644060155414e-05,
+      "loss": 0.667,
+      "step": 1416
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.504844339512095e-05,
+      "loss": 1.0514,
+      "step": 1417
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.504044006233089e-05,
+      "loss": 1.0859,
+      "step": 1418
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.503243060427235e-05,
+      "loss": 1.0762,
+      "step": 1419
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.502441502203458e-05,
+      "loss": 0.6674,
+      "step": 1420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.501639331670764e-05,
+      "loss": 1.2507,
+      "step": 1421
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.500836548938246e-05,
+      "loss": 0.5687,
+      "step": 1422
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.50003315411508e-05,
+      "loss": 1.3159,
+      "step": 1423
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.499229147310519e-05,
+      "loss": 1.395,
+      "step": 1424
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.498424528633908e-05,
+      "loss": 0.9869,
+      "step": 1425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.497619298194669e-05,
+      "loss": 0.3864,
+      "step": 1426
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.496813456102308e-05,
+      "loss": 1.4589,
+      "step": 1427
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.496007002466418e-05,
+      "loss": 1.1418,
+      "step": 1428
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.495199937396673e-05,
+      "loss": 1.3817,
+      "step": 1429
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.494392261002827e-05,
+      "loss": 1.3581,
+      "step": 1430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.49358397339472e-05,
+      "loss": 0.7105,
+      "step": 1431
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.492775074682276e-05,
+      "loss": 0.6363,
+      "step": 1432
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.491965564975502e-05,
+      "loss": 1.478,
+      "step": 1433
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.491155444384488e-05,
+      "loss": 1.2608,
+      "step": 1434
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.490344713019402e-05,
+      "loss": 1.2344,
+      "step": 1435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.489533370990502e-05,
+      "loss": 1.3555,
+      "step": 1436
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.488721418408127e-05,
+      "loss": 1.8575,
+      "step": 1437
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.487908855382696e-05,
+      "loss": 1.7164,
+      "step": 1438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487095682024717e-05,
+      "loss": 1.1549,
+      "step": 1439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.486281898444774e-05,
+      "loss": 0.4764,
+      "step": 1440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48546750475354e-05,
+      "loss": 0.8104,
+      "step": 1441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.484652501061766e-05,
+      "loss": 0.9989,
+      "step": 1442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.483836887480288e-05,
+      "loss": 1.3663,
+      "step": 1443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.483020664120028e-05,
+      "loss": 0.5541,
+      "step": 1444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.482203831091987e-05,
+      "loss": 1.3791,
+      "step": 1445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48138638850725e-05,
+      "loss": 1.5065,
+      "step": 1446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.480568336476985e-05,
+      "loss": 1.3544,
+      "step": 1447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.479749675112441e-05,
+      "loss": 0.6867,
+      "step": 1448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.478930404524955e-05,
+      "loss": 0.6082,
+      "step": 1449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.478110524825942e-05,
+      "loss": 1.2985,
+      "step": 1450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.477290036126903e-05,
+      "loss": 1.0988,
+      "step": 1451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.476468938539415e-05,
+      "loss": 1.2858,
+      "step": 1452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.47564723217515e-05,
+      "loss": 0.9328,
+      "step": 1453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.474824917145853e-05,
+      "loss": 1.2591,
+      "step": 1454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.474001993563351e-05,
+      "loss": 0.862,
+      "step": 1455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.473178461539563e-05,
+      "loss": 0.7267,
+      "step": 1456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.472354321186483e-05,
+      "loss": 1.2387,
+      "step": 1457
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.471529572616189e-05,
+      "loss": 1.1992,
+      "step": 1458
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470704215940844e-05,
+      "loss": 1.3719,
+      "step": 1459
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46987825127269e-05,
+      "loss": 1.5179,
+      "step": 1460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.469051678724057e-05,
+      "loss": 1.4843,
+      "step": 1461
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.468224498407353e-05,
+      "loss": 1.384,
+      "step": 1462
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.467396710435068e-05,
+      "loss": 0.7532,
+      "step": 1463
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.466568314919782e-05,
+      "loss": 1.0536,
+      "step": 1464
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.465739311974147e-05,
+      "loss": 1.2305,
+      "step": 1465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.464909701710906e-05,
+      "loss": 1.1128,
+      "step": 1466
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.464079484242884e-05,
+      "loss": 0.6408,
+      "step": 1467
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.463248659682984e-05,
+      "loss": 0.8915,
+      "step": 1468
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.462417228144193e-05,
+      "loss": 1.211,
+      "step": 1469
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.461585189739581e-05,
+      "loss": 1.1859,
+      "step": 1470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.460752544582304e-05,
+      "loss": 1.2919,
+      "step": 1471
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.459919292785595e-05,
+      "loss": 1.3392,
+      "step": 1472
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.459085434462773e-05,
+      "loss": 1.3016,
+      "step": 1473
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.458250969727238e-05,
+      "loss": 0.6577,
+      "step": 1474
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.457415898692472e-05,
+      "loss": 0.4598,
+      "step": 1475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.456580221472042e-05,
+      "loss": 1.4936,
+      "step": 1476
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.455743938179596e-05,
+      "loss": 0.9002,
+      "step": 1477
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.454907048928863e-05,
+      "loss": 1.4427,
+      "step": 1478
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.454069553833656e-05,
+      "loss": 1.4995,
+      "step": 1479
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.45323145300787e-05,
+      "loss": 1.3346,
+      "step": 1480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.452392746565482e-05,
+      "loss": 1.2736,
+      "step": 1481
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.451553434620552e-05,
+      "loss": 0.6391,
+      "step": 1482
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.450713517287221e-05,
+      "loss": 1.3605,
+      "step": 1483
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.449872994679718e-05,
+      "loss": 0.6677,
+      "step": 1484
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.449031866912346e-05,
+      "loss": 1.6748,
+      "step": 1485
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.448190134099493e-05,
+      "loss": 1.5022,
+      "step": 1486
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.447347796355632e-05,
+      "loss": 1.2775,
+      "step": 1487
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.446504853795318e-05,
+      "loss": 0.6709,
+      "step": 1488
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.445661306533185e-05,
+      "loss": 1.2621,
+      "step": 1489
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.444817154683952e-05,
+      "loss": 1.1927,
+      "step": 1490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.443972398362417e-05,
+      "loss": 0.7796,
+      "step": 1491
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.443127037683465e-05,
+      "loss": 1.4787,
+      "step": 1492
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.442281072762063e-05,
+      "loss": 1.4278,
+      "step": 1493
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.441434503713251e-05,
+      "loss": 1.2377,
+      "step": 1494
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.440587330652164e-05,
+      "loss": 1.164,
+      "step": 1495
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.43973955369401e-05,
+      "loss": 0.5477,
+      "step": 1496
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.438891172954085e-05,
+      "loss": 1.5256,
+      "step": 1497
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.438042188547762e-05,
+      "loss": 1.0149,
+      "step": 1498
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4371926005905e-05,
+      "loss": 0.6775,
+      "step": 1499
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.436342409197836e-05,
+      "loss": 0.4933,
+      "step": 1500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.435491614485396e-05,
+      "loss": 0.6305,
+      "step": 1501
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.434640216568879e-05,
+      "loss": 0.9662,
+      "step": 1502
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.433788215564076e-05,
+      "loss": 1.3147,
+      "step": 1503
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.432935611586851e-05,
+      "loss": 0.9504,
+      "step": 1504
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.432082404753154e-05,
+      "loss": 0.8983,
+      "step": 1505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.431228595179017e-05,
+      "loss": 0.6204,
+      "step": 1506
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.430374182980555e-05,
+      "loss": 1.0484,
+      "step": 1507
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.429519168273963e-05,
+      "loss": 1.0207,
+      "step": 1508
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.428663551175517e-05,
+      "loss": 1.0042,
+      "step": 1509
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.42780733180158e-05,
+      "loss": 0.6942,
+      "step": 1510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.42695051026859e-05,
+      "loss": 0.6491,
+      "step": 1511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.42609308669307e-05,
+      "loss": 0.9136,
+      "step": 1512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.425235061191631e-05,
+      "loss": 1.2512,
+      "step": 1513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.424376433880953e-05,
+      "loss": 1.4341,
+      "step": 1514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.423517204877809e-05,
+      "loss": 1.0536,
+      "step": 1515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.422657374299047e-05,
+      "loss": 0.949,
+      "step": 1516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.421796942261603e-05,
+      "loss": 1.4059,
+      "step": 1517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.420935908882488e-05,
+      "loss": 0.9719,
+      "step": 1518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.420074274278801e-05,
+      "loss": 1.0346,
+      "step": 1519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.419212038567718e-05,
+      "loss": 0.6418,
+      "step": 1520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.418349201866499e-05,
+      "loss": 1.4911,
+      "step": 1521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.417485764292487e-05,
+      "loss": 0.7396,
+      "step": 1522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4166217259631e-05,
+      "loss": 1.1844,
+      "step": 1523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.415757086995848e-05,
+      "loss": 0.5958,
+      "step": 1524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.414891847508316e-05,
+      "loss": 1.2377,
+      "step": 1525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.414026007618173e-05,
+      "loss": 1.2297,
+      "step": 1526
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.413159567443167e-05,
+      "loss": 0.896,
+      "step": 1527
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.412292527101129e-05,
+      "loss": 1.166,
+      "step": 1528
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.411424886709973e-05,
+      "loss": 1.3749,
+      "step": 1529
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.410556646387694e-05,
+      "loss": 1.1589,
+      "step": 1530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.409687806252369e-05,
+      "loss": 0.5971,
+      "step": 1531
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.408818366422154e-05,
+      "loss": 0.7785,
+      "step": 1532
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.407948327015289e-05,
+      "loss": 1.3646,
+      "step": 1533
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.407077688150096e-05,
+      "loss": 1.1006,
+      "step": 1534
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.406206449944975e-05,
+      "loss": 1.3808,
+      "step": 1535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.405334612518413e-05,
+      "loss": 1.3995,
+      "step": 1536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.404462175988972e-05,
+      "loss": 0.6672,
+      "step": 1537
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.403589140475302e-05,
+      "loss": 0.7954,
+      "step": 1538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.402715506096129e-05,
+      "loss": 0.6366,
+      "step": 1539
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.401841272970265e-05,
+      "loss": 1.1299,
+      "step": 1540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.400966441216598e-05,
+      "loss": 1.1222,
+      "step": 1541
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.400091010954104e-05,
+      "loss": 1.4085,
+      "step": 1542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.399214982301835e-05,
+      "loss": 0.8708,
+      "step": 1543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.398338355378927e-05,
+      "loss": 1.4435,
+      "step": 1544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.397461130304596e-05,
+      "loss": 0.9897,
+      "step": 1545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.396583307198143e-05,
+      "loss": 0.833,
+      "step": 1546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.395704886178945e-05,
+      "loss": 1.077,
+      "step": 1547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.394825867366462e-05,
+      "loss": 0.4635,
+      "step": 1548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.393946250880237e-05,
+      "loss": 1.589,
+      "step": 1549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.393066036839894e-05,
+      "loss": 1.3592,
+      "step": 1550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.392185225365137e-05,
+      "loss": 1.6142,
+      "step": 1551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.391303816575752e-05,
+      "loss": 1.5147,
+      "step": 1552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.390421810591606e-05,
+      "loss": 1.061,
+      "step": 1553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.389539207532646e-05,
+      "loss": 1.2823,
+      "step": 1554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.388656007518903e-05,
+      "loss": 1.5733,
+      "step": 1555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.387772210670488e-05,
+      "loss": 1.2441,
+      "step": 1556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.386887817107593e-05,
+      "loss": 1.1737,
+      "step": 1557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.386002826950489e-05,
+      "loss": 1.0643,
+      "step": 1558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.385117240319531e-05,
+      "loss": 0.6822,
+      "step": 1559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.384231057335156e-05,
+      "loss": 1.3673,
+      "step": 1560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.383344278117877e-05,
+      "loss": 1.2726,
+      "step": 1561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.382456902788295e-05,
+      "loss": 0.9631,
+      "step": 1562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.381568931467085e-05,
+      "loss": 1.0826,
+      "step": 1563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38068036427501e-05,
+      "loss": 0.9981,
+      "step": 1564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.379791201332907e-05,
+      "loss": 1.3313,
+      "step": 1565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.378901442761701e-05,
+      "loss": 0.7532,
+      "step": 1566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.378011088682392e-05,
+      "loss": 0.7888,
+      "step": 1567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.377120139216065e-05,
+      "loss": 1.1626,
+      "step": 1568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.376228594483884e-05,
+      "loss": 1.2117,
+      "step": 1569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.375336454607095e-05,
+      "loss": 1.0935,
+      "step": 1570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.374443719707024e-05,
+      "loss": 1.5217,
+      "step": 1571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.373550389905079e-05,
+      "loss": 0.6253,
+      "step": 1572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.372656465322747e-05,
+      "loss": 1.1522,
+      "step": 1573
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.371761946081599e-05,
+      "loss": 0.7389,
+      "step": 1574
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.370866832303282e-05,
+      "loss": 0.8915,
+      "step": 1575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.36997112410953e-05,
+      "loss": 1.1048,
+      "step": 1576
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.369074821622153e-05,
+      "loss": 1.003,
+      "step": 1577
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.368177924963045e-05,
+      "loss": 1.1698,
+      "step": 1578
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.367280434254177e-05,
+      "loss": 0.8689,
+      "step": 1579
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.366382349617606e-05,
+      "loss": 1.2336,
+      "step": 1580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.365483671175466e-05,
+      "loss": 1.1936,
+      "step": 1581
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.36458439904997e-05,
+      "loss": 0.7807,
+      "step": 1582
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.363684533363417e-05,
+      "loss": 1.15,
+      "step": 1583
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.362784074238184e-05,
+      "loss": 1.2575,
+      "step": 1584
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.361883021796727e-05,
+      "loss": 1.1293,
+      "step": 1585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.360981376161588e-05,
+      "loss": 0.9004,
+      "step": 1586
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.360079137455383e-05,
+      "loss": 1.2146,
+      "step": 1587
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.359176305800813e-05,
+      "loss": 0.6516,
+      "step": 1588
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.358272881320658e-05,
+      "loss": 1.206,
+      "step": 1589
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.357368864137782e-05,
+      "loss": 1.1286,
+      "step": 1590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.35646425437512e-05,
+      "loss": 1.073,
+      "step": 1591
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.3555590521557e-05,
+      "loss": 1.2945,
+      "step": 1592
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.354653257602624e-05,
+      "loss": 1.5278,
+      "step": 1593
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.353746870839073e-05,
+      "loss": 0.8902,
+      "step": 1594
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.352839891988315e-05,
+      "loss": 1.2301,
+      "step": 1595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.351932321173691e-05,
+      "loss": 0.9834,
+      "step": 1596
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.351024158518627e-05,
+      "loss": 1.1652,
+      "step": 1597
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.350115404146628e-05,
+      "loss": 1.4906,
+      "step": 1598
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.34920605818128e-05,
+      "loss": 1.243,
+      "step": 1599
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.348296120746251e-05,
+      "loss": 1.337,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.347385591965287e-05,
+      "loss": 0.8959,
+      "step": 1601
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.346474471962215e-05,
+      "loss": 1.038,
+      "step": 1602
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.34556276086094e-05,
+      "loss": 0.7287,
+      "step": 1603
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.344650458785454e-05,
+      "loss": 1.4178,
+      "step": 1604
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.343737565859824e-05,
+      "loss": 0.7017,
+      "step": 1605
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.342824082208199e-05,
+      "loss": 1.3962,
+      "step": 1606
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.341910007954808e-05,
+      "loss": 0.5986,
+      "step": 1607
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.34099534322396e-05,
+      "loss": 0.9806,
+      "step": 1608
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.340080088140044e-05,
+      "loss": 1.2048,
+      "step": 1609
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.339164242827532e-05,
+      "loss": 0.5458,
+      "step": 1610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.338247807410971e-05,
+      "loss": 1.3503,
+      "step": 1611
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.337330782014995e-05,
+      "loss": 0.8094,
+      "step": 1612
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.336413166764313e-05,
+      "loss": 1.1316,
+      "step": 1613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.335494961783718e-05,
+      "loss": 0.7933,
+      "step": 1614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.334576167198078e-05,
+      "loss": 1.28,
+      "step": 1615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.333656783132345e-05,
+      "loss": 1.225,
+      "step": 1616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.332736809711553e-05,
+      "loss": 1.0018,
+      "step": 1617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.331816247060811e-05,
+      "loss": 0.9023,
+      "step": 1618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330895095305311e-05,
+      "loss": 1.3426,
+      "step": 1619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32997335457033e-05,
+      "loss": 0.8301,
+      "step": 1620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.329051024981211e-05,
+      "loss": 1.0547,
+      "step": 1621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.328128106663392e-05,
+      "loss": 0.7006,
+      "step": 1622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.327204599742386e-05,
+      "loss": 1.0805,
+      "step": 1623
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.326280504343781e-05,
+      "loss": 1.2014,
+      "step": 1624
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.325355820593251e-05,
+      "loss": 1.2932,
+      "step": 1625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32443054861655e-05,
+      "loss": 1.0078,
+      "step": 1626
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.323504688539507e-05,
+      "loss": 1.2518,
+      "step": 1627
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.322578240488038e-05,
+      "loss": 0.8809,
+      "step": 1628
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.321651204588132e-05,
+      "loss": 1.0599,
+      "step": 1629
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.320723580965863e-05,
+      "loss": 1.4442,
+      "step": 1630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.319795369747381e-05,
+      "loss": 0.6279,
+      "step": 1631
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.318866571058922e-05,
+      "loss": 1.3262,
+      "step": 1632
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.317937185026793e-05,
+      "loss": 0.9952,
+      "step": 1633
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.317007211777389e-05,
+      "loss": 1.2261,
+      "step": 1634
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31607665143718e-05,
+      "loss": 1.002,
+      "step": 1635
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.315145504132719e-05,
+      "loss": 1.0056,
+      "step": 1636
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.314213769990634e-05,
+      "loss": 1.3758,
+      "step": 1637
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31328144913764e-05,
+      "loss": 1.1799,
+      "step": 1638
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.312348541700527e-05,
+      "loss": 1.9172,
+      "step": 1639
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.311415047806163e-05,
+      "loss": 1.2517,
+      "step": 1640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.310480967581498e-05,
+      "loss": 1.26,
+      "step": 1641
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.309546301153568e-05,
+      "loss": 1.3812,
+      "step": 1642
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.308611048649479e-05,
+      "loss": 1.44,
+      "step": 1643
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.307675210196418e-05,
+      "loss": 0.5474,
+      "step": 1644
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.306738785921658e-05,
+      "loss": 0.9506,
+      "step": 1645
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.305801775952547e-05,
+      "loss": 0.622,
+      "step": 1646
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.304864180416513e-05,
+      "loss": 0.8937,
+      "step": 1647
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.303925999441063e-05,
+      "loss": 0.7998,
+      "step": 1648
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.302987233153788e-05,
+      "loss": 0.4651,
+      "step": 1649
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.302047881682353e-05,
+      "loss": 0.5049,
+      "step": 1650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.301107945154506e-05,
+      "loss": 1.1247,
+      "step": 1651
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.300167423698071e-05,
+      "loss": 0.9933,
+      "step": 1652
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.299226317440959e-05,
+      "loss": 1.2693,
+      "step": 1653
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.29828462651115e-05,
+      "loss": 0.5007,
+      "step": 1654
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.297342351036715e-05,
+      "loss": 1.2042,
+      "step": 1655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.296399491145796e-05,
+      "loss": 1.2572,
+      "step": 1656
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.295456046966613e-05,
+      "loss": 1.1021,
+      "step": 1657
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.294512018627475e-05,
+      "loss": 0.5298,
+      "step": 1658
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.293567406256763e-05,
+      "loss": 0.9373,
+      "step": 1659
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.292622209982941e-05,
+      "loss": 1.2614,
+      "step": 1660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.291676429934548e-05,
+      "loss": 1.6055,
+      "step": 1661
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.290730066240208e-05,
+      "loss": 0.8198,
+      "step": 1662
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.289783119028618e-05,
+      "loss": 0.9322,
+      "step": 1663
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.288835588428563e-05,
+      "loss": 0.9297,
+      "step": 1664
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.287887474568897e-05,
+      "loss": 0.7874,
+      "step": 1665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.286938777578564e-05,
+      "loss": 0.3621,
+      "step": 1666
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.285989497586577e-05,
+      "loss": 1.3045,
+      "step": 1667
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.285039634722036e-05,
+      "loss": 1.4494,
+      "step": 1668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.284089189114115e-05,
+      "loss": 1.0472,
+      "step": 1669
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.283138160892074e-05,
+      "loss": 0.7939,
+      "step": 1670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.282186550185245e-05,
+      "loss": 0.5421,
+      "step": 1671
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.281234357123043e-05,
+      "loss": 1.2502,
+      "step": 1672
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.28028158183496e-05,
+      "loss": 1.528,
+      "step": 1673
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.27932822445057e-05,
+      "loss": 1.0042,
+      "step": 1674
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.278374285099523e-05,
+      "loss": 1.4421,
+      "step": 1675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.277419763911553e-05,
+      "loss": 1.0555,
+      "step": 1676
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.276464661016466e-05,
+      "loss": 1.3676,
+      "step": 1677
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.275508976544154e-05,
+      "loss": 0.5267,
+      "step": 1678
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.274552710624584e-05,
+      "loss": 1.3965,
+      "step": 1679
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.273595863387805e-05,
+      "loss": 1.1407,
+      "step": 1680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.27263843496394e-05,
+      "loss": 1.332,
+      "step": 1681
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.271680425483197e-05,
+      "loss": 1.1322,
+      "step": 1682
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.270721835075862e-05,
+      "loss": 0.6023,
+      "step": 1683
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.269762663872292e-05,
+      "loss": 0.914,
+      "step": 1684
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.268802912002936e-05,
+      "loss": 1.0517,
+      "step": 1685
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.267842579598312e-05,
+      "loss": 1.1967,
+      "step": 1686
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.26688166678902e-05,
+      "loss": 0.9814,
+      "step": 1687
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.265920173705742e-05,
+      "loss": 0.7888,
+      "step": 1688
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.264958100479233e-05,
+      "loss": 0.9467,
+      "step": 1689
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.263995447240331e-05,
+      "loss": 0.8956,
+      "step": 1690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.263032214119954e-05,
+      "loss": 1.1196,
+      "step": 1691
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.262068401249093e-05,
+      "loss": 0.7332,
+      "step": 1692
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.261104008758824e-05,
+      "loss": 0.8971,
+      "step": 1693
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.260139036780297e-05,
+      "loss": 1.5133,
+      "step": 1694
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.259173485444747e-05,
+      "loss": 1.4909,
+      "step": 1695
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.258207354883483e-05,
+      "loss": 0.4475,
+      "step": 1696
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.257240645227893e-05,
+      "loss": 1.3431,
+      "step": 1697
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.256273356609444e-05,
+      "loss": 1.3958,
+      "step": 1698
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.255305489159682e-05,
+      "loss": 1.2036,
+      "step": 1699
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.254337043010235e-05,
+      "loss": 1.1793,
+      "step": 1700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.253368018292802e-05,
+      "loss": 1.3272,
+      "step": 1701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25239841513917e-05,
+      "loss": 0.9011,
+      "step": 1702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.251428233681196e-05,
+      "loss": 0.997,
+      "step": 1703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.250457474050823e-05,
+      "loss": 0.8242,
+      "step": 1704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.249486136380068e-05,
+      "loss": 1.3406,
+      "step": 1705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.248514220801028e-05,
+      "loss": 1.1235,
+      "step": 1706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.247541727445876e-05,
+      "loss": 1.1679,
+      "step": 1707
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24656865644687e-05,
+      "loss": 1.1898,
+      "step": 1708
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.245595007936343e-05,
+      "loss": 1.1369,
+      "step": 1709
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.244620782046703e-05,
+      "loss": 1.3684,
+      "step": 1710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.243645978910441e-05,
+      "loss": 0.9936,
+      "step": 1711
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.242670598660126e-05,
+      "loss": 1.2144,
+      "step": 1712
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.241694641428404e-05,
+      "loss": 1.5452,
+      "step": 1713
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.240718107347999e-05,
+      "loss": 0.552,
+      "step": 1714
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.239740996551717e-05,
+      "loss": 0.461,
+      "step": 1715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.238763309172438e-05,
+      "loss": 1.3112,
+      "step": 1716
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.237785045343123e-05,
+      "loss": 1.3736,
+      "step": 1717
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.236806205196813e-05,
+      "loss": 1.3856,
+      "step": 1718
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.235826788866622e-05,
+      "loss": 1.0862,
+      "step": 1719
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.234846796485747e-05,
+      "loss": 0.5812,
+      "step": 1720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.233866228187463e-05,
+      "loss": 0.8738,
+      "step": 1721
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23288508410512e-05,
+      "loss": 1.2008,
+      "step": 1722
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23190336437215e-05,
+      "loss": 1.5181,
+      "step": 1723
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.230921069122061e-05,
+      "loss": 1.2981,
+      "step": 1724
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.229938198488442e-05,
+      "loss": 1.1947,
+      "step": 1725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.228954752604957e-05,
+      "loss": 0.6212,
+      "step": 1726
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.227970731605348e-05,
+      "loss": 0.5849,
+      "step": 1727
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.22698613562344e-05,
+      "loss": 1.2483,
+      "step": 1728
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.226000964793131e-05,
+      "loss": 0.7781,
+      "step": 1729
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.2250152192484e-05,
+      "loss": 0.8267,
+      "step": 1730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.224028899123303e-05,
+      "loss": 1.1712,
+      "step": 1731
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.223042004551973e-05,
+      "loss": 0.9198,
+      "step": 1732
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.222054535668625e-05,
+      "loss": 0.9033,
+      "step": 1733
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.221066492607549e-05,
+      "loss": 0.9509,
+      "step": 1734
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.220077875503113e-05,
+      "loss": 1.3733,
+      "step": 1735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.219088684489765e-05,
+      "loss": 0.5612,
+      "step": 1736
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.218098919702029e-05,
+      "loss": 1.4684,
+      "step": 1737
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.217108581274508e-05,
+      "loss": 1.5702,
+      "step": 1738
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.216117669341882e-05,
+      "loss": 1.3136,
+      "step": 1739
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.215126184038911e-05,
+      "loss": 1.0329,
+      "step": 1740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.214134125500433e-05,
+      "loss": 1.0724,
+      "step": 1741
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.213141493861361e-05,
+      "loss": 0.9584,
+      "step": 1742
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.21214828925669e-05,
+      "loss": 1.1537,
+      "step": 1743
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.211154511821489e-05,
+      "loss": 1.2946,
+      "step": 1744
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.210160161690906e-05,
+      "loss": 1.0655,
+      "step": 1745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.209165239000169e-05,
+      "loss": 1.1936,
+      "step": 1746
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.208169743884583e-05,
+      "loss": 1.051,
+      "step": 1747
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.207173676479527e-05,
+      "loss": 1.2405,
+      "step": 1748
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.206177036920464e-05,
+      "loss": 1.4922,
+      "step": 1749
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.205179825342932e-05,
+      "loss": 1.5087,
+      "step": 1750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.204182041882543e-05,
+      "loss": 1.0329,
+      "step": 1751
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.203183686674994e-05,
+      "loss": 1.3761,
+      "step": 1752
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.202184759856056e-05,
+      "loss": 0.8222,
+      "step": 1753
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.201185261561577e-05,
+      "loss": 1.1872,
+      "step": 1754
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.200185191927485e-05,
+      "loss": 1.1204,
+      "step": 1755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.199184551089781e-05,
+      "loss": 1.045,
+      "step": 1756
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.19818333918455e-05,
+      "loss": 0.7705,
+      "step": 1757
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.197181556347951e-05,
+      "loss": 1.3189,
+      "step": 1758
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.19617920271622e-05,
+      "loss": 1.1028,
+      "step": 1759
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.195176278425674e-05,
+      "loss": 0.9986,
+      "step": 1760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.194172783612704e-05,
+      "loss": 0.7939,
+      "step": 1761
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.193168718413781e-05,
+      "loss": 0.8658,
+      "step": 1762
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.192164082965454e-05,
+      "loss": 1.2558,
+      "step": 1763
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.191158877404344e-05,
+      "loss": 0.4415,
+      "step": 1764
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.190153101867158e-05,
+      "loss": 0.8195,
+      "step": 1765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.189146756490675e-05,
+      "loss": 1.415,
+      "step": 1766
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.188139841411752e-05,
+      "loss": 1.1379,
+      "step": 1767
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.187132356767325e-05,
+      "loss": 1.3,
+      "step": 1768
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.186124302694405e-05,
+      "loss": 0.4993,
+      "step": 1769
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.185115679330088e-05,
+      "loss": 0.6635,
+      "step": 1770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.184106486811535e-05,
+      "loss": 1.2158,
+      "step": 1771
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.183096725275992e-05,
+      "loss": 1.2711,
+      "step": 1772
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.182086394860784e-05,
+      "loss": 1.3179,
+      "step": 1773
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.18107549570331e-05,
+      "loss": 0.8997,
+      "step": 1774
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.180064027941046e-05,
+      "loss": 1.0842,
+      "step": 1775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.179051991711546e-05,
+      "loss": 1.7678,
+      "step": 1776
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.178039387152446e-05,
+      "loss": 1.3104,
+      "step": 1777
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.17702621440145e-05,
+      "loss": 1.2077,
+      "step": 1778
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.176012473596345e-05,
+      "loss": 1.4441,
+      "step": 1779
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.174998164874998e-05,
+      "loss": 0.6651,
+      "step": 1780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.173983288375346e-05,
+      "loss": 1.4942,
+      "step": 1781
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.172967844235407e-05,
+      "loss": 1.2201,
+      "step": 1782
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.17195183259328e-05,
+      "loss": 0.8623,
+      "step": 1783
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.170935253587133e-05,
+      "loss": 1.2748,
+      "step": 1784
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.169918107355218e-05,
+      "loss": 0.8404,
+      "step": 1785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.168900394035861e-05,
+      "loss": 0.6219,
+      "step": 1786
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.167882113767467e-05,
+      "loss": 1.1089,
+      "step": 1787
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.166863266688514e-05,
+      "loss": 1.3895,
+      "step": 1788
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.165843852937564e-05,
+      "loss": 1.4625,
+      "step": 1789
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.16482387265325e-05,
+      "loss": 0.5113,
+      "step": 1790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.163803325974281e-05,
+      "loss": 1.4916,
+      "step": 1791
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.162782213039452e-05,
+      "loss": 0.6028,
+      "step": 1792
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.161760533987623e-05,
+      "loss": 1.3891,
+      "step": 1793
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.160738288957742e-05,
+      "loss": 1.191,
+      "step": 1794
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.159715478088827e-05,
+      "loss": 1.1376,
+      "step": 1795
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.158692101519976e-05,
+      "loss": 1.5038,
+      "step": 1796
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.157668159390361e-05,
+      "loss": 1.067,
+      "step": 1797
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.156643651839235e-05,
+      "loss": 0.6911,
+      "step": 1798
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.155618579005924e-05,
+      "loss": 1.5306,
+      "step": 1799
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.154592941029835e-05,
+      "loss": 0.563,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.153566738050448e-05,
+      "loss": 0.8616,
+      "step": 1801
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.152539970207319e-05,
+      "loss": 1.1519,
+      "step": 1802
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.151512637640087e-05,
+      "loss": 0.5753,
+      "step": 1803
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.150484740488464e-05,
+      "loss": 0.6602,
+      "step": 1804
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.149456278892235e-05,
+      "loss": 0.9033,
+      "step": 1805
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.148427252991268e-05,
+      "loss": 1.3268,
+      "step": 1806
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.147397662925506e-05,
+      "loss": 0.6904,
+      "step": 1807
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.146367508834966e-05,
+      "loss": 0.9543,
+      "step": 1808
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.145336790859743e-05,
+      "loss": 0.9279,
+      "step": 1809
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.144305509140012e-05,
+      "loss": 0.6297,
+      "step": 1810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.143273663816022e-05,
+      "loss": 1.4088,
+      "step": 1811
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.142241255028097e-05,
+      "loss": 1.0781,
+      "step": 1812
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14120828291664e-05,
+      "loss": 1.3163,
+      "step": 1813
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14017474762213e-05,
+      "loss": 1.1563,
+      "step": 1814
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.139140649285121e-05,
+      "loss": 0.3986,
+      "step": 1815
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.138105988046248e-05,
+      "loss": 0.6292,
+      "step": 1816
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.137070764046217e-05,
+      "loss": 0.6296,
+      "step": 1817
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.136034977425813e-05,
+      "loss": 0.5213,
+      "step": 1818
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.134998628325901e-05,
+      "loss": 1.4974,
+      "step": 1819
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.133961716887415e-05,
+      "loss": 1.3971,
+      "step": 1820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.132924243251373e-05,
+      "loss": 0.7889,
+      "step": 1821
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.131886207558864e-05,
+      "loss": 1.1155,
+      "step": 1822
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.130847609951057e-05,
+      "loss": 0.8765,
+      "step": 1823
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.129808450569193e-05,
+      "loss": 0.6422,
+      "step": 1824
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.128768729554595e-05,
+      "loss": 1.2699,
+      "step": 1825
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.12772844704866e-05,
+      "loss": 1.4712,
+      "step": 1826
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.12668760319286e-05,
+      "loss": 1.542,
+      "step": 1827
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.125646198128745e-05,
+      "loss": 1.1364,
+      "step": 1828
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.12460423199794e-05,
+      "loss": 0.6635,
+      "step": 1829
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.123561704942148e-05,
+      "loss": 1.6251,
+      "step": 1830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.122518617103144e-05,
+      "loss": 1.075,
+      "step": 1831
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.121474968622788e-05,
+      "loss": 1.4092,
+      "step": 1832
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.120430759643007e-05,
+      "loss": 1.4485,
+      "step": 1833
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.11938599030581e-05,
+      "loss": 1.512,
+      "step": 1834
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.118340660753278e-05,
+      "loss": 1.018,
+      "step": 1835
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.117294771127574e-05,
+      "loss": 1.1203,
+      "step": 1836
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.11624832157093e-05,
+      "loss": 1.2393,
+      "step": 1837
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.11520131222566e-05,
+      "loss": 1.4323,
+      "step": 1838
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.114153743234151e-05,
+      "loss": 1.2637,
+      "step": 1839
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.113105614738866e-05,
+      "loss": 1.3221,
+      "step": 1840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.112056926882348e-05,
+      "loss": 0.8498,
+      "step": 1841
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.11100767980721e-05,
+      "loss": 1.2715,
+      "step": 1842
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.109957873656147e-05,
+      "loss": 0.6816,
+      "step": 1843
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.108907508571925e-05,
+      "loss": 1.3726,
+      "step": 1844
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.10785658469739e-05,
+      "loss": 1.2266,
+      "step": 1845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.106805102175461e-05,
+      "loss": 1.5494,
+      "step": 1846
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.105753061149134e-05,
+      "loss": 0.6278,
+      "step": 1847
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.104700461761481e-05,
+      "loss": 1.3898,
+      "step": 1848
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.103647304155654e-05,
+      "loss": 1.3187,
+      "step": 1849
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.102593588474873e-05,
+      "loss": 1.0533,
+      "step": 1850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.101539314862437e-05,
+      "loss": 0.7204,
+      "step": 1851
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.100484483461725e-05,
+      "loss": 1.0709,
+      "step": 1852
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.099429094416186e-05,
+      "loss": 0.9051,
+      "step": 1853
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.09837314786935e-05,
+      "loss": 1.6071,
+      "step": 1854
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.09731664396482e-05,
+      "loss": 0.8158,
+      "step": 1855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.096259582846271e-05,
+      "loss": 0.5344,
+      "step": 1856
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.095201964657463e-05,
+      "loss": 1.0845,
+      "step": 1857
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.094143789542224e-05,
+      "loss": 1.1923,
+      "step": 1858
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.093085057644464e-05,
+      "loss": 0.5767,
+      "step": 1859
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.092025769108158e-05,
+      "loss": 0.8268,
+      "step": 1860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.09096592407737e-05,
+      "loss": 1.3895,
+      "step": 1861
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.089905522696231e-05,
+      "loss": 0.7938,
+      "step": 1862
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.088844565108952e-05,
+      "loss": 1.4164,
+      "step": 1863
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.087783051459816e-05,
+      "loss": 1.0012,
+      "step": 1864
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.086720981893184e-05,
+      "loss": 1.2711,
+      "step": 1865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.08565835655349e-05,
+      "loss": 1.4065,
+      "step": 1866
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.084595175585249e-05,
+      "loss": 1.197,
+      "step": 1867
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.083531439133045e-05,
+      "loss": 1.1336,
+      "step": 1868
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.082467147341544e-05,
+      "loss": 1.227,
+      "step": 1869
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.081402300355482e-05,
+      "loss": 1.1428,
+      "step": 1870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.080336898319673e-05,
+      "loss": 1.1759,
+      "step": 1871
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.079270941379006e-05,
+      "loss": 1.1134,
+      "step": 1872
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.078204429678445e-05,
+      "loss": 0.9872,
+      "step": 1873
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.077137363363033e-05,
+      "loss": 1.1815,
+      "step": 1874
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.076069742577882e-05,
+      "loss": 1.0196,
+      "step": 1875
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.075001567468186e-05,
+      "loss": 1.2342,
+      "step": 1876
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.073932838179211e-05,
+      "loss": 1.6348,
+      "step": 1877
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.072863554856296e-05,
+      "loss": 1.0027,
+      "step": 1878
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.071793717644859e-05,
+      "loss": 1.4233,
+      "step": 1879
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.070723326690396e-05,
+      "loss": 0.7507,
+      "step": 1880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.069652382138469e-05,
+      "loss": 1.3795,
+      "step": 1881
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.068580884134726e-05,
+      "loss": 0.7014,
+      "step": 1882
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06750883282488e-05,
+      "loss": 1.273,
+      "step": 1883
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.066436228354729e-05,
+      "loss": 1.1365,
+      "step": 1884
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06536307087014e-05,
+      "loss": 1.2025,
+      "step": 1885
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.064289360517056e-05,
+      "loss": 1.1545,
+      "step": 1886
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.063215097441497e-05,
+      "loss": 0.9534,
+      "step": 1887
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06214028178956e-05,
+      "loss": 0.7974,
+      "step": 1888
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.061064913707409e-05,
+      "loss": 1.0264,
+      "step": 1889
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.059988993341292e-05,
+      "loss": 0.9426,
+      "step": 1890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05891252083753e-05,
+      "loss": 1.3389,
+      "step": 1891
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.057835496342512e-05,
+      "loss": 0.7243,
+      "step": 1892
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.056757920002715e-05,
+      "loss": 1.1649,
+      "step": 1893
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.055679791964677e-05,
+      "loss": 0.8468,
+      "step": 1894
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.054601112375023e-05,
+      "loss": 1.3931,
+      "step": 1895
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.053521881380446e-05,
+      "loss": 1.362,
+      "step": 1896
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.052442099127717e-05,
+      "loss": 0.7965,
+      "step": 1897
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.051361765763678e-05,
+      "loss": 1.2666,
+      "step": 1898
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.050280881435251e-05,
+      "loss": 1.3102,
+      "step": 1899
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.049199446289431e-05,
+      "loss": 1.2972,
+      "step": 1900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.048117460473287e-05,
+      "loss": 0.7632,
+      "step": 1901
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.047034924133962e-05,
+      "loss": 1.4272,
+      "step": 1902
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.045951837418678e-05,
+      "loss": 0.9429,
+      "step": 1903
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.044868200474727e-05,
+      "loss": 0.8061,
+      "step": 1904
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.043784013449478e-05,
+      "loss": 0.9701,
+      "step": 1905
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.042699276490379e-05,
+      "loss": 0.9869,
+      "step": 1906
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.041613989744944e-05,
+      "loss": 1.4681,
+      "step": 1907
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.040528153360768e-05,
+      "loss": 1.0896,
+      "step": 1908
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.03944176748552e-05,
+      "loss": 0.7928,
+      "step": 1909
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.038354832266942e-05,
+      "loss": 0.9691,
+      "step": 1910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.03726734785285e-05,
+      "loss": 1.3751,
+      "step": 1911
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.03617931439114e-05,
+      "loss": 0.9921,
+      "step": 1912
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.035090732029777e-05,
+      "loss": 0.9042,
+      "step": 1913
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.034001600916801e-05,
+      "loss": 0.6646,
+      "step": 1914
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.03291192120033e-05,
+      "loss": 1.1428,
+      "step": 1915
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.031821693028555e-05,
+      "loss": 0.832,
+      "step": 1916
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.030730916549742e-05,
+      "loss": 1.429,
+      "step": 1917
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.029639591912229e-05,
+      "loss": 1.2014,
+      "step": 1918
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.02854771926443e-05,
+      "loss": 1.6573,
+      "step": 1919
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.02745529875484e-05,
+      "loss": 1.3292,
+      "step": 1920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.026362330532014e-05,
+      "loss": 1.1667,
+      "step": 1921
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.025268814744596e-05,
+      "loss": 1.3748,
+      "step": 1922
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.024174751541298e-05,
+      "loss": 0.8086,
+      "step": 1923
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.023080141070904e-05,
+      "loss": 0.858,
+      "step": 1924
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.021984983482279e-05,
+      "loss": 0.4778,
+      "step": 1925
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.020889278924356e-05,
+      "loss": 1.1632,
+      "step": 1926
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.019793027546146e-05,
+      "loss": 1.1033,
+      "step": 1927
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.018696229496735e-05,
+      "loss": 0.6745,
+      "step": 1928
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.017598884925278e-05,
+      "loss": 1.444,
+      "step": 1929
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.016500993981014e-05,
+      "loss": 1.177,
+      "step": 1930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.015402556813245e-05,
+      "loss": 0.5783,
+      "step": 1931
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.014303573571356e-05,
+      "loss": 0.7778,
+      "step": 1932
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.013204044404804e-05,
+      "loss": 0.8957,
+      "step": 1933
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.012103969463115e-05,
+      "loss": 1.541,
+      "step": 1934
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.011003348895899e-05,
+      "loss": 1.4887,
+      "step": 1935
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.009902182852829e-05,
+      "loss": 1.2928,
+      "step": 1936
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.008800471483663e-05,
+      "loss": 1.1936,
+      "step": 1937
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.007698214938226e-05,
+      "loss": 1.5454,
+      "step": 1938
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.00659541336642e-05,
+      "loss": 1.41,
+      "step": 1939
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.005492066918217e-05,
+      "loss": 0.8574,
+      "step": 1940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.00438817574367e-05,
+      "loss": 1.2572,
+      "step": 1941
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.0032837399929e-05,
+      "loss": 0.7092,
+      "step": 1942
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.002178759816108e-05,
+      "loss": 1.0909,
+      "step": 1943
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.001073235363563e-05,
+      "loss": 1.1508,
+      "step": 1944
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.999967166785612e-05,
+      "loss": 1.2131,
+      "step": 1945
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.998860554232674e-05,
+      "loss": 1.207,
+      "step": 1946
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.997753397855242e-05,
+      "loss": 0.7121,
+      "step": 1947
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.996645697803885e-05,
+      "loss": 1.2788,
+      "step": 1948
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.995537454229242e-05,
+      "loss": 1.347,
+      "step": 1949
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.994428667282032e-05,
+      "loss": 0.5483,
+      "step": 1950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.993319337113042e-05,
+      "loss": 1.4016,
+      "step": 1951
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.992209463873135e-05,
+      "loss": 1.2586,
+      "step": 1952
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.99109904771325e-05,
+      "loss": 1.0937,
+      "step": 1953
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.989988088784395e-05,
+      "loss": 0.8814,
+      "step": 1954
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.988876587237658e-05,
+      "loss": 1.129,
+      "step": 1955
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.987764543224195e-05,
+      "loss": 1.1413,
+      "step": 1956
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.986651956895239e-05,
+      "loss": 1.6672,
+      "step": 1957
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.985538828402096e-05,
+      "loss": 1.0815,
+      "step": 1958
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.984425157896147e-05,
+      "loss": 1.3099,
+      "step": 1959
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.983310945528844e-05,
+      "loss": 1.3602,
+      "step": 1960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.982196191451712e-05,
+      "loss": 1.0579,
+      "step": 1961
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.981080895816358e-05,
+      "loss": 0.605,
+      "step": 1962
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.979965058774451e-05,
+      "loss": 1.6223,
+      "step": 1963
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.978848680477743e-05,
+      "loss": 1.1152,
+      "step": 1964
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.977731761078051e-05,
+      "loss": 1.3436,
+      "step": 1965
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.976614300727274e-05,
+      "loss": 0.5142,
+      "step": 1966
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.975496299577379e-05,
+      "loss": 1.5019,
+      "step": 1967
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.974377757780412e-05,
+      "loss": 0.5917,
+      "step": 1968
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.973258675488483e-05,
+      "loss": 1.341,
+      "step": 1969
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.972139052853786e-05,
+      "loss": 1.6742,
+      "step": 1970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.971018890028581e-05,
+      "loss": 0.4755,
+      "step": 1971
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.969898187165209e-05,
+      "loss": 1.3047,
+      "step": 1972
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.968776944416075e-05,
+      "loss": 0.8303,
+      "step": 1973
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.967655161933663e-05,
+      "loss": 0.8064,
+      "step": 1974
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.966532839870531e-05,
+      "loss": 1.3766,
+      "step": 1975
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.96540997837931e-05,
+      "loss": 1.4281,
+      "step": 1976
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.964286577612703e-05,
+      "loss": 0.8534,
+      "step": 1977
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.963162637723484e-05,
+      "loss": 1.2651,
+      "step": 1978
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.962038158864505e-05,
+      "loss": 1.5675,
+      "step": 1979
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.960913141188691e-05,
+      "loss": 1.0948,
+      "step": 1980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.959787584849035e-05,
+      "loss": 1.6266,
+      "step": 1981
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.958661489998609e-05,
+      "loss": 1.3385,
+      "step": 1982
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.957534856790556e-05,
+      "loss": 0.8288,
+      "step": 1983
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.956407685378092e-05,
+      "loss": 1.0067,
+      "step": 1984
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.955279975914508e-05,
+      "loss": 1.3719,
+      "step": 1985
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.954151728553165e-05,
+      "loss": 0.7157,
+      "step": 1986
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.953022943447498e-05,
+      "loss": 0.7888,
+      "step": 1987
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.951893620751018e-05,
+      "loss": 1.0606,
+      "step": 1988
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.950763760617306e-05,
+      "loss": 1.284,
+      "step": 1989
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.949633363200017e-05,
+      "loss": 1.2136,
+      "step": 1990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.94850242865288e-05,
+      "loss": 0.9783,
+      "step": 1991
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.947370957129697e-05,
+      "loss": 1.5097,
+      "step": 1992
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.94623894878434e-05,
+      "loss": 1.0677,
+      "step": 1993
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.94510640377076e-05,
+      "loss": 0.9588,
+      "step": 1994
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.943973322242973e-05,
+      "loss": 0.7456,
+      "step": 1995
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.942839704355075e-05,
+      "loss": 0.6653,
+      "step": 1996
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.94170555026123e-05,
+      "loss": 1.2836,
+      "step": 1997
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.94057086011568e-05,
+      "loss": 0.4227,
+      "step": 1998
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.939435634072734e-05,
+      "loss": 0.6082,
+      "step": 1999
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.93829987228678e-05,
+      "loss": 1.1528,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.937163574912275e-05,
+      "loss": 0.9719,
+      "step": 2001
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.936026742103747e-05,
+      "loss": 1.4983,
+      "step": 2002
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.934889374015802e-05,
+      "loss": 1.2155,
+      "step": 2003
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.933751470803116e-05,
+      "loss": 0.6801,
+      "step": 2004
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.932613032620437e-05,
+      "loss": 1.2835,
+      "step": 2005
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.931474059622589e-05,
+      "loss": 0.6792,
+      "step": 2006
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.930334551964462e-05,
+      "loss": 1.1057,
+      "step": 2007
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.929194509801026e-05,
+      "loss": 1.3913,
+      "step": 2008
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.928053933287324e-05,
+      "loss": 0.8555,
+      "step": 2009
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.926912822578462e-05,
+      "loss": 0.8491,
+      "step": 2010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.92577117782963e-05,
+      "loss": 1.0352,
+      "step": 2011
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.924628999196084e-05,
+      "loss": 0.7482,
+      "step": 2012
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.923486286833156e-05,
+      "loss": 1.2199,
+      "step": 2013
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.922343040896245e-05,
+      "loss": 1.0021,
+      "step": 2014
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.921199261540832e-05,
+      "loss": 1.5116,
+      "step": 2015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.920054948922463e-05,
+      "loss": 1.2482,
+      "step": 2016
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.918910103196756e-05,
+      "loss": 1.4267,
+      "step": 2017
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.917764724519406e-05,
+      "loss": 1.0679,
+      "step": 2018
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.91661881304618e-05,
+      "loss": 0.5883,
+      "step": 2019
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.915472368932916e-05,
+      "loss": 0.871,
+      "step": 2020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.914325392335521e-05,
+      "loss": 1.1459,
+      "step": 2021
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.913177883409982e-05,
+      "loss": 1.0356,
+      "step": 2022
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.912029842312352e-05,
+      "loss": 0.7129,
+      "step": 2023
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.910881269198761e-05,
+      "loss": 0.9758,
+      "step": 2024
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.909732164225407e-05,
+      "loss": 0.7006,
+      "step": 2025
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.908582527548561e-05,
+      "loss": 0.5331,
+      "step": 2026
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.90743235932457e-05,
+      "loss": 1.1346,
+      "step": 2027
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.906281659709852e-05,
+      "loss": 1.1131,
+      "step": 2028
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.905130428860893e-05,
+      "loss": 0.6576,
+      "step": 2029
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.903978666934257e-05,
+      "loss": 0.9606,
+      "step": 2030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.902826374086576e-05,
+      "loss": 0.931,
+      "step": 2031
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.901673550474558e-05,
+      "loss": 0.8156,
+      "step": 2032
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.90052019625498e-05,
+      "loss": 1.4671,
+      "step": 2033
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.899366311584692e-05,
+      "loss": 1.089,
+      "step": 2034
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.898211896620616e-05,
+      "loss": 1.2012,
+      "step": 2035
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.897056951519746e-05,
+      "loss": 0.6901,
+      "step": 2036
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.895901476439152e-05,
+      "loss": 1.1746,
+      "step": 2037
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.894745471535969e-05,
+      "loss": 1.1812,
+      "step": 2038
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.89358893696741e-05,
+      "loss": 1.3035,
+      "step": 2039
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.892431872890757e-05,
+      "loss": 1.2398,
+      "step": 2040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.891274279463361e-05,
+      "loss": 0.7703,
+      "step": 2041
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.890116156842656e-05,
+      "loss": 1.1589,
+      "step": 2042
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.888957505186136e-05,
+      "loss": 1.3137,
+      "step": 2043
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.887798324651373e-05,
+      "loss": 0.8445,
+      "step": 2044
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.886638615396008e-05,
+      "loss": 1.3044,
+      "step": 2045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.885478377577758e-05,
+      "loss": 1.3534,
+      "step": 2046
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.88431761135441e-05,
+      "loss": 1.3622,
+      "step": 2047
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.883156316883817e-05,
+      "loss": 1.5471,
+      "step": 2048
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.881994494323915e-05,
+      "loss": 0.943,
+      "step": 2049
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.880832143832703e-05,
+      "loss": 1.1599,
+      "step": 2050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.879669265568253e-05,
+      "loss": 1.3154,
+      "step": 2051
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.878505859688715e-05,
+      "loss": 0.5428,
+      "step": 2052
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.877341926352303e-05,
+      "loss": 1.0404,
+      "step": 2053
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.876177465717307e-05,
+      "loss": 1.4554,
+      "step": 2054
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.875012477942087e-05,
+      "loss": 1.4126,
+      "step": 2055
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.873846963185076e-05,
+      "loss": 1.6457,
+      "step": 2056
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.872680921604778e-05,
+      "loss": 0.8227,
+      "step": 2057
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.87151435335977e-05,
+      "loss": 1.7085,
+      "step": 2058
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.870347258608696e-05,
+      "loss": 1.3538,
+      "step": 2059
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.869179637510275e-05,
+      "loss": 0.9115,
+      "step": 2060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.868011490223302e-05,
+      "loss": 1.4281,
+      "step": 2061
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.866842816906635e-05,
+      "loss": 1.3035,
+      "step": 2062
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.865673617719208e-05,
+      "loss": 0.6373,
+      "step": 2063
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.864503892820028e-05,
+      "loss": 1.3126,
+      "step": 2064
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86333364236817e-05,
+      "loss": 1.0365,
+      "step": 2065
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.862162866522783e-05,
+      "loss": 1.0893,
+      "step": 2066
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.860991565443086e-05,
+      "loss": 1.0482,
+      "step": 2067
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.859819739288369e-05,
+      "loss": 1.0174,
+      "step": 2068
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.858647388217995e-05,
+      "loss": 0.93,
+      "step": 2069
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.8574745123914e-05,
+      "loss": 1.394,
+      "step": 2070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.856301111968085e-05,
+      "loss": 1.0268,
+      "step": 2071
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.855127187107628e-05,
+      "loss": 0.6675,
+      "step": 2072
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.85395273796968e-05,
+      "loss": 1.3386,
+      "step": 2073
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.852777764713956e-05,
+      "loss": 1.286,
+      "step": 2074
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.851602267500245e-05,
+      "loss": 1.3629,
+      "step": 2075
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.850426246488414e-05,
+      "loss": 1.1799,
+      "step": 2076
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.849249701838391e-05,
+      "loss": 1.4614,
+      "step": 2077
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.848072633710185e-05,
+      "loss": 1.1807,
+      "step": 2078
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.846895042263866e-05,
+      "loss": 1.4529,
+      "step": 2079
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.845716927659582e-05,
+      "loss": 1.1486,
+      "step": 2080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.844538290057553e-05,
+      "loss": 0.5596,
+      "step": 2081
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.843359129618065e-05,
+      "loss": 1.004,
+      "step": 2082
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.842179446501478e-05,
+      "loss": 1.226,
+      "step": 2083
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.840999240868224e-05,
+      "loss": 1.1511,
+      "step": 2084
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.839818512878805e-05,
+      "loss": 1.0335,
+      "step": 2085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.838637262693792e-05,
+      "loss": 1.668,
+      "step": 2086
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.837455490473832e-05,
+      "loss": 1.2567,
+      "step": 2087
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.836273196379637e-05,
+      "loss": 0.8532,
+      "step": 2088
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.835090380571993e-05,
+      "loss": 1.1326,
+      "step": 2089
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.83390704321176e-05,
+      "loss": 1.1832,
+      "step": 2090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.832723184459861e-05,
+      "loss": 1.3584,
+      "step": 2091
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.8315388044773e-05,
+      "loss": 1.1877,
+      "step": 2092
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.830353903425144e-05,
+      "loss": 1.0417,
+      "step": 2093
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.829168481464531e-05,
+      "loss": 1.1181,
+      "step": 2094
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.827982538756678e-05,
+      "loss": 0.6172,
+      "step": 2095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.826796075462862e-05,
+      "loss": 1.2854,
+      "step": 2096
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.82560909174444e-05,
+      "loss": 1.3279,
+      "step": 2097
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.82442158776283e-05,
+      "loss": 1.3841,
+      "step": 2098
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.823233563679534e-05,
+      "loss": 0.5515,
+      "step": 2099
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.82204501965611e-05,
+      "loss": 0.699,
+      "step": 2100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.820855955854198e-05,
+      "loss": 1.2242,
+      "step": 2101
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.819666372435504e-05,
+      "loss": 0.8598,
+      "step": 2102
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.818476269561804e-05,
+      "loss": 1.2026,
+      "step": 2103
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.817285647394948e-05,
+      "loss": 1.3922,
+      "step": 2104
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.816094506096852e-05,
+      "loss": 0.9355,
+      "step": 2105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.814902845829507e-05,
+      "loss": 1.2009,
+      "step": 2106
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.813710666754971e-05,
+      "loss": 1.2479,
+      "step": 2107
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.812517969035377e-05,
+      "loss": 1.1927,
+      "step": 2108
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.811324752832923e-05,
+      "loss": 0.8835,
+      "step": 2109
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.810131018309881e-05,
+      "loss": 0.9298,
+      "step": 2110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.808936765628594e-05,
+      "loss": 1.3505,
+      "step": 2111
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.807741994951473e-05,
+      "loss": 0.8095,
+      "step": 2112
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.806546706441e-05,
+      "loss": 1.0767,
+      "step": 2113
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.805350900259729e-05,
+      "loss": 1.1938,
+      "step": 2114
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.804154576570285e-05,
+      "loss": 1.173,
+      "step": 2115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.802957735535358e-05,
+      "loss": 1.0142,
+      "step": 2116
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.801760377317716e-05,
+      "loss": 1.2353,
+      "step": 2117
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.800562502080193e-05,
+      "loss": 1.5274,
+      "step": 2118
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.799364109985693e-05,
+      "loss": 0.636,
+      "step": 2119
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.798165201197189e-05,
+      "loss": 1.2097,
+      "step": 2120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.79696577587773e-05,
+      "loss": 1.2697,
+      "step": 2121
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.795765834190432e-05,
+      "loss": 0.6741,
+      "step": 2122
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.794565376298477e-05,
+      "loss": 1.3576,
+      "step": 2123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.793364402365123e-05,
+      "loss": 1.5012,
+      "step": 2124
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.792162912553699e-05,
+      "loss": 1.2186,
+      "step": 2125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.790960907027597e-05,
+      "loss": 1.3718,
+      "step": 2126
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.789758385950288e-05,
+      "loss": 1.1677,
+      "step": 2127
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.788555349485305e-05,
+      "loss": 0.6581,
+      "step": 2128
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.787351797796255e-05,
+      "loss": 1.2133,
+      "step": 2129
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.786147731046817e-05,
+      "loss": 1.1219,
+      "step": 2130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.784943149400737e-05,
+      "loss": 1.1877,
+      "step": 2131
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.783738053021833e-05,
+      "loss": 0.8291,
+      "step": 2132
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.78253244207399e-05,
+      "loss": 1.373,
+      "step": 2133
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.781326316721165e-05,
+      "loss": 1.3258,
+      "step": 2134
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.780119677127386e-05,
+      "loss": 0.9246,
+      "step": 2135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.778912523456748e-05,
+      "loss": 0.885,
+      "step": 2136
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.777704855873419e-05,
+      "loss": 0.9651,
+      "step": 2137
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.776496674541637e-05,
+      "loss": 1.3881,
+      "step": 2138
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.775287979625706e-05,
+      "loss": 1.4258,
+      "step": 2139
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.774078771290005e-05,
+      "loss": 1.2513,
+      "step": 2140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.772869049698976e-05,
+      "loss": 0.8357,
+      "step": 2141
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.771658815017139e-05,
+      "loss": 0.8798,
+      "step": 2142
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.770448067409077e-05,
+      "loss": 1.2983,
+      "step": 2143
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.769236807039446e-05,
+      "loss": 0.6709,
+      "step": 2144
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.768025034072973e-05,
+      "loss": 1.4633,
+      "step": 2145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.766812748674451e-05,
+      "loss": 0.858,
+      "step": 2146
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.765599951008745e-05,
+      "loss": 0.4304,
+      "step": 2147
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.76438664124079e-05,
+      "loss": 0.9316,
+      "step": 2148
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.763172819535589e-05,
+      "loss": 1.5649,
+      "step": 2149
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.761958486058215e-05,
+      "loss": 0.8756,
+      "step": 2150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.760743640973814e-05,
+      "loss": 1.5528,
+      "step": 2151
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.759528284447595e-05,
+      "loss": 1.2245,
+      "step": 2152
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.758312416644842e-05,
+      "loss": 1.2298,
+      "step": 2153
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.757096037730908e-05,
+      "loss": 0.9232,
+      "step": 2154
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.755879147871211e-05,
+      "loss": 1.195,
+      "step": 2155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.754661747231245e-05,
+      "loss": 1.4569,
+      "step": 2156
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.753443835976569e-05,
+      "loss": 1.2429,
+      "step": 2157
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.752225414272812e-05,
+      "loss": 1.491,
+      "step": 2158
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.751006482285673e-05,
+      "loss": 1.2335,
+      "step": 2159
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.749787040180923e-05,
+      "loss": 0.9566,
+      "step": 2160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.748567088124397e-05,
+      "loss": 1.3688,
+      "step": 2161
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.747346626282003e-05,
+      "loss": 1.5877,
+      "step": 2162
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.746125654819718e-05,
+      "loss": 0.622,
+      "step": 2163
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.744904173903588e-05,
+      "loss": 0.8016,
+      "step": 2164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.743682183699727e-05,
+      "loss": 0.6783,
+      "step": 2165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74245968437432e-05,
+      "loss": 1.1445,
+      "step": 2166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.741236676093623e-05,
+      "loss": 1.4508,
+      "step": 2167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.740013159023956e-05,
+      "loss": 1.4346,
+      "step": 2168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.738789133331711e-05,
+      "loss": 0.6123,
+      "step": 2169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.73756459918335e-05,
+      "loss": 0.6716,
+      "step": 2170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.736339556745402e-05,
+      "loss": 0.7265,
+      "step": 2171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.735114006184469e-05,
+      "loss": 0.9467,
+      "step": 2172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.733887947667219e-05,
+      "loss": 1.1958,
+      "step": 2173
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.732661381360388e-05,
+      "loss": 0.9894,
+      "step": 2174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.731434307430784e-05,
+      "loss": 1.3555,
+      "step": 2175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.730206726045281e-05,
+      "loss": 1.4976,
+      "step": 2176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.728978637370825e-05,
+      "loss": 1.1898,
+      "step": 2177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.727750041574432e-05,
+      "loss": 1.2838,
+      "step": 2178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.726520938823182e-05,
+      "loss": 1.0089,
+      "step": 2179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.725291329284228e-05,
+      "loss": 0.9736,
+      "step": 2180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.72406121312479e-05,
+      "loss": 0.8453,
+      "step": 2181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.722830590512157e-05,
+      "loss": 1.3131,
+      "step": 2182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.721599461613687e-05,
+      "loss": 1.4497,
+      "step": 2183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.720367826596807e-05,
+      "loss": 1.2711,
+      "step": 2184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.719135685629016e-05,
+      "loss": 0.957,
+      "step": 2185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.717903038877877e-05,
+      "loss": 1.1182,
+      "step": 2186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.716669886511022e-05,
+      "loss": 1.2084,
+      "step": 2187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.715436228696154e-05,
+      "loss": 1.4389,
+      "step": 2188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.714202065601047e-05,
+      "loss": 1.1325,
+      "step": 2189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.712967397393539e-05,
+      "loss": 1.3194,
+      "step": 2190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.711732224241538e-05,
+      "loss": 0.6805,
+      "step": 2191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.71049654631302e-05,
+      "loss": 1.3863,
+      "step": 2192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.709260363776034e-05,
+      "loss": 1.3059,
+      "step": 2193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.708023676798692e-05,
+      "loss": 1.3508,
+      "step": 2194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.706786485549179e-05,
+      "loss": 0.5999,
+      "step": 2195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.705548790195743e-05,
+      "loss": 1.0136,
+      "step": 2196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.70431059090671e-05,
+      "loss": 1.3038,
+      "step": 2197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.703071887850464e-05,
+      "loss": 1.3968,
+      "step": 2198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.701832681195464e-05,
+      "loss": 1.2446,
+      "step": 2199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.700592971110235e-05,
+      "loss": 0.9837,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.699352757763373e-05,
+      "loss": 0.612,
+      "step": 2201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.698112041323539e-05,
+      "loss": 1.2657,
+      "step": 2202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.696870821959463e-05,
+      "loss": 1.4529,
+      "step": 2203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.695629099839948e-05,
+      "loss": 1.4521,
+      "step": 2204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.694386875133858e-05,
+      "loss": 1.6137,
+      "step": 2205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.693144148010132e-05,
+      "loss": 0.822,
+      "step": 2206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.691900918637773e-05,
+      "loss": 0.8655,
+      "step": 2207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.690657187185854e-05,
+      "loss": 1.0639,
+      "step": 2208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.689412953823517e-05,
+      "loss": 0.8209,
+      "step": 2209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.68816821871997e-05,
+      "loss": 0.6359,
+      "step": 2210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.686922982044493e-05,
+      "loss": 1.3847,
+      "step": 2211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.685677243966427e-05,
+      "loss": 1.2476,
+      "step": 2212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.684431004655192e-05,
+      "loss": 0.7703,
+      "step": 2213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.683184264280264e-05,
+      "loss": 1.2113,
+      "step": 2214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.6819370230112e-05,
+      "loss": 0.9112,
+      "step": 2215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.680689281017613e-05,
+      "loss": 1.1842,
+      "step": 2216
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.679441038469191e-05,
+      "loss": 1.305,
+      "step": 2217
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.678192295535688e-05,
+      "loss": 0.5296,
+      "step": 2218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.67694305238693e-05,
+      "loss": 0.8583,
+      "step": 2219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.675693309192803e-05,
+      "loss": 0.4484,
+      "step": 2220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.67444306612327e-05,
+      "loss": 1.1536,
+      "step": 2221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.673192323348354e-05,
+      "loss": 1.5319,
+      "step": 2222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.67194108103815e-05,
+      "loss": 1.3103,
+      "step": 2223
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.670689339362825e-05,
+      "loss": 0.651,
+      "step": 2224
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.669437098492603e-05,
+      "loss": 1.1477,
+      "step": 2225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.668184358597788e-05,
+      "loss": 1.6556,
+      "step": 2226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.666931119848741e-05,
+      "loss": 1.4334,
+      "step": 2227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.6656773824159e-05,
+      "loss": 1.2866,
+      "step": 2228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.664423146469765e-05,
+      "loss": 0.7567,
+      "step": 2229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.663168412180907e-05,
+      "loss": 1.0804,
+      "step": 2230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.661913179719961e-05,
+      "loss": 1.2599,
+      "step": 2231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.660657449257634e-05,
+      "loss": 1.192,
+      "step": 2232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.6594012209647e-05,
+      "loss": 1.3948,
+      "step": 2233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.658144495011997e-05,
+      "loss": 1.6501,
+      "step": 2234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.656887271570432e-05,
+      "loss": 1.5006,
+      "step": 2235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.655629550810987e-05,
+      "loss": 1.3219,
+      "step": 2236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.654371332904701e-05,
+      "loss": 0.5759,
+      "step": 2237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.653112618022686e-05,
+      "loss": 1.3345,
+      "step": 2238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.651853406336122e-05,
+      "loss": 1.3364,
+      "step": 2239
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.650593698016251e-05,
+      "loss": 1.2092,
+      "step": 2240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.649333493234393e-05,
+      "loss": 0.9568,
+      "step": 2241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.648072792161925e-05,
+      "loss": 0.6208,
+      "step": 2242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.646811594970299e-05,
+      "loss": 1.0362,
+      "step": 2243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.645549901831027e-05,
+      "loss": 1.2366,
+      "step": 2244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.644287712915697e-05,
+      "loss": 1.2251,
+      "step": 2245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.643025028395958e-05,
+      "loss": 1.0972,
+      "step": 2246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.641761848443531e-05,
+      "loss": 1.1398,
+      "step": 2247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.640498173230199e-05,
+      "loss": 1.3824,
+      "step": 2248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.639234002927817e-05,
+      "loss": 1.26,
+      "step": 2249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.637969337708303e-05,
+      "loss": 1.0452,
+      "step": 2250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.636704177743649e-05,
+      "loss": 1.2835,
+      "step": 2251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.635438523205908e-05,
+      "loss": 0.6754,
+      "step": 2252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.634172374267203e-05,
+      "loss": 0.7654,
+      "step": 2253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.632905731099724e-05,
+      "loss": 0.9683,
+      "step": 2254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.631638593875729e-05,
+      "loss": 1.0279,
+      "step": 2255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.630370962767538e-05,
+      "loss": 1.0691,
+      "step": 2256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.629102837947546e-05,
+      "loss": 1.2089,
+      "step": 2257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.627834219588211e-05,
+      "loss": 0.9428,
+      "step": 2258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.626565107862058e-05,
+      "loss": 1.2141,
+      "step": 2259
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.625295502941681e-05,
+      "loss": 1.3826,
+      "step": 2260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.624025404999738e-05,
+      "loss": 1.0018,
+      "step": 2261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.622754814208957e-05,
+      "loss": 0.9582,
+      "step": 2262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62148373074213e-05,
+      "loss": 1.3177,
+      "step": 2263
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.620212154772119e-05,
+      "loss": 0.6267,
+      "step": 2264
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.618940086471853e-05,
+      "loss": 1.0763,
+      "step": 2265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.617667526014326e-05,
+      "loss": 0.7384,
+      "step": 2266
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.616394473572598e-05,
+      "loss": 1.3426,
+      "step": 2267
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.615120929319798e-05,
+      "loss": 0.9746,
+      "step": 2268
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.613846893429124e-05,
+      "loss": 1.0077,
+      "step": 2269
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.612572366073837e-05,
+      "loss": 0.5675,
+      "step": 2270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.611297347427265e-05,
+      "loss": 1.6069,
+      "step": 2271
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.610021837662804e-05,
+      "loss": 1.2489,
+      "step": 2272
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.608745836953919e-05,
+      "loss": 1.1848,
+      "step": 2273
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.607469345474137e-05,
+      "loss": 1.1924,
+      "step": 2274
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.606192363397056e-05,
+      "loss": 1.4438,
+      "step": 2275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.604914890896338e-05,
+      "loss": 1.1685,
+      "step": 2276
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.603636928145715e-05,
+      "loss": 1.0268,
+      "step": 2277
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.60235847531898e-05,
+      "loss": 0.8908,
+      "step": 2278
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.601079532589997e-05,
+      "loss": 0.8344,
+      "step": 2279
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.599800100132698e-05,
+      "loss": 0.7102,
+      "step": 2280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.598520178121077e-05,
+      "loss": 1.0418,
+      "step": 2281
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.597239766729197e-05,
+      "loss": 0.8422,
+      "step": 2282
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.595958866131187e-05,
+      "loss": 1.232,
+      "step": 2283
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.594677476501243e-05,
+      "loss": 0.7733,
+      "step": 2284
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.59339559801363e-05,
+      "loss": 1.3317,
+      "step": 2285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.592113230842673e-05,
+      "loss": 0.8566,
+      "step": 2286
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.590830375162771e-05,
+      "loss": 0.4456,
+      "step": 2287
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.589547031148381e-05,
+      "loss": 1.0673,
+      "step": 2288
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.588263198974035e-05,
+      "loss": 0.5458,
+      "step": 2289
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.58697887881433e-05,
+      "loss": 0.616,
+      "step": 2290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.58569407084392e-05,
+      "loss": 1.0726,
+      "step": 2291
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.584408775237538e-05,
+      "loss": 1.5763,
+      "step": 2292
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.583122992169975e-05,
+      "loss": 1.361,
+      "step": 2293
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.58183672181609e-05,
+      "loss": 1.0347,
+      "step": 2294
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.580549964350812e-05,
+      "loss": 1.1757,
+      "step": 2295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.579262719949131e-05,
+      "loss": 1.1717,
+      "step": 2296
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.577974988786108e-05,
+      "loss": 0.7837,
+      "step": 2297
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.576686771036866e-05,
+      "loss": 1.2282,
+      "step": 2298
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.575398066876596e-05,
+      "loss": 1.164,
+      "step": 2299
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.574108876480557e-05,
+      "loss": 0.93,
+      "step": 2300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.572819200024069e-05,
+      "loss": 0.9041,
+      "step": 2301
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.571529037682525e-05,
+      "loss": 0.762,
+      "step": 2302
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.570238389631377e-05,
+      "loss": 1.1418,
+      "step": 2303
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.568947256046149e-05,
+      "loss": 0.8014,
+      "step": 2304
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.567655637102427e-05,
+      "loss": 0.6523,
+      "step": 2305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.566363532975866e-05,
+      "loss": 0.8161,
+      "step": 2306
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.565070943842185e-05,
+      "loss": 1.3634,
+      "step": 2307
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.56377786987717e-05,
+      "loss": 0.9943,
+      "step": 2308
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.56248431125667e-05,
+      "loss": 1.2933,
+      "step": 2309
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.561190268156604e-05,
+      "loss": 1.3628,
+      "step": 2310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.559895740752956e-05,
+      "loss": 1.3219,
+      "step": 2311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.558600729221775e-05,
+      "loss": 0.5231,
+      "step": 2312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.557305233739175e-05,
+      "loss": 1.4725,
+      "step": 2313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.556009254481335e-05,
+      "loss": 1.4309,
+      "step": 2314
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.554712791624506e-05,
+      "loss": 1.2591,
+      "step": 2315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.553415845344996e-05,
+      "loss": 0.5718,
+      "step": 2316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.552118415819185e-05,
+      "loss": 1.2163,
+      "step": 2317
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.550820503223518e-05,
+      "loss": 0.9129,
+      "step": 2318
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.549522107734502e-05,
+      "loss": 1.3295,
+      "step": 2319
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.548223229528714e-05,
+      "loss": 0.7331,
+      "step": 2320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.546923868782793e-05,
+      "loss": 0.4686,
+      "step": 2321
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.545624025673446e-05,
+      "loss": 1.142,
+      "step": 2322
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.544323700377446e-05,
+      "loss": 0.9292,
+      "step": 2323
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.543022893071629e-05,
+      "loss": 0.9648,
+      "step": 2324
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.541721603932899e-05,
+      "loss": 1.2668,
+      "step": 2325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.540419833138224e-05,
+      "loss": 1.1326,
+      "step": 2326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.53911758086464e-05,
+      "loss": 1.1368,
+      "step": 2327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.537814847289245e-05,
+      "loss": 0.7843,
+      "step": 2328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.536511632589204e-05,
+      "loss": 1.0331,
+      "step": 2329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.53520793694175e-05,
+      "loss": 1.0295,
+      "step": 2330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.533903760524173e-05,
+      "loss": 1.2835,
+      "step": 2331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.532599103513841e-05,
+      "loss": 0.9793,
+      "step": 2332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.531293966088176e-05,
+      "loss": 1.7212,
+      "step": 2333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.529988348424674e-05,
+      "loss": 1.3509,
+      "step": 2334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.528682250700888e-05,
+      "loss": 0.5306,
+      "step": 2335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.527375673094445e-05,
+      "loss": 0.794,
+      "step": 2336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.526068615783027e-05,
+      "loss": 1.2289,
+      "step": 2337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.524761078944392e-05,
+      "loss": 1.2607,
+      "step": 2338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.523453062756357e-05,
+      "loss": 1.2245,
+      "step": 2339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.522144567396806e-05,
+      "loss": 1.3595,
+      "step": 2340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.520835593043687e-05,
+      "loss": 1.2633,
+      "step": 2341
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.519526139875014e-05,
+      "loss": 0.9997,
+      "step": 2342
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.518216208068863e-05,
+      "loss": 0.9355,
+      "step": 2343
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.516905797803383e-05,
+      "loss": 1.6109,
+      "step": 2344
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.51559490925678e-05,
+      "loss": 1.3775,
+      "step": 2345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.514283542607328e-05,
+      "loss": 1.5279,
+      "step": 2346
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.51297169803337e-05,
+      "loss": 1.7645,
+      "step": 2347
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.511659375713304e-05,
+      "loss": 1.3316,
+      "step": 2348
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.510346575825603e-05,
+      "loss": 1.1062,
+      "step": 2349
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.5090332985488e-05,
+      "loss": 1.3752,
+      "step": 2350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.507719544061496e-05,
+      "loss": 1.1562,
+      "step": 2351
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.506405312542352e-05,
+      "loss": 1.1566,
+      "step": 2352
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.505090604170096e-05,
+      "loss": 1.2851,
+      "step": 2353
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.503775419123525e-05,
+      "loss": 0.6229,
+      "step": 2354
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.502459757581495e-05,
+      "loss": 1.2733,
+      "step": 2355
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.501143619722931e-05,
+      "loss": 0.4679,
+      "step": 2356
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.499827005726819e-05,
+      "loss": 1.1453,
+      "step": 2357
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.498509915772215e-05,
+      "loss": 1.2933,
+      "step": 2358
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.497192350038232e-05,
+      "loss": 0.6543,
+      "step": 2359
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.495874308704053e-05,
+      "loss": 0.9878,
+      "step": 2360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.494555791948929e-05,
+      "loss": 0.6719,
+      "step": 2361
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.493236799952166e-05,
+      "loss": 0.6107,
+      "step": 2362
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.491917332893147e-05,
+      "loss": 0.5569,
+      "step": 2363
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.490597390951305e-05,
+      "loss": 1.2199,
+      "step": 2364
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.489276974306151e-05,
+      "loss": 1.3165,
+      "step": 2365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.487956083137251e-05,
+      "loss": 1.2752,
+      "step": 2366
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.486634717624242e-05,
+      "loss": 1.4274,
+      "step": 2367
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.485312877946823e-05,
+      "loss": 1.4408,
+      "step": 2368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.483990564284755e-05,
+      "loss": 1.3926,
+      "step": 2369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.482667776817868e-05,
+      "loss": 1.3056,
+      "step": 2370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.481344515726053e-05,
+      "loss": 1.4955,
+      "step": 2371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.480020781189268e-05,
+      "loss": 0.7362,
+      "step": 2372
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.478696573387534e-05,
+      "loss": 1.2301,
+      "step": 2373
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.477371892500935e-05,
+      "loss": 0.9747,
+      "step": 2374
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.476046738709622e-05,
+      "loss": 1.2877,
+      "step": 2375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.47472111219381e-05,
+      "loss": 0.6997,
+      "step": 2376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.473395013133775e-05,
+      "loss": 1.8155,
+      "step": 2377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.472068441709861e-05,
+      "loss": 0.8136,
+      "step": 2378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.470741398102475e-05,
+      "loss": 0.8857,
+      "step": 2379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.469413882492089e-05,
+      "loss": 0.5189,
+      "step": 2380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.468085895059236e-05,
+      "loss": 1.2587,
+      "step": 2381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.466757435984517e-05,
+      "loss": 1.1643,
+      "step": 2382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.465428505448596e-05,
+      "loss": 1.0574,
+      "step": 2383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.464099103632199e-05,
+      "loss": 0.9052,
+      "step": 2384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.462769230716118e-05,
+      "loss": 0.9888,
+      "step": 2385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.461438886881212e-05,
+      "loss": 1.2673,
+      "step": 2386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.460108072308396e-05,
+      "loss": 0.9975,
+      "step": 2387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.458776787178659e-05,
+      "loss": 1.2145,
+      "step": 2388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.457445031673046e-05,
+      "loss": 1.0738,
+      "step": 2389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.456112805972669e-05,
+      "loss": 0.8367,
+      "step": 2390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.454780110258706e-05,
+      "loss": 1.4223,
+      "step": 2391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.453446944712393e-05,
+      "loss": 0.7005,
+      "step": 2392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.452113309515038e-05,
+      "loss": 0.9889,
+      "step": 2393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.450779204848002e-05,
+      "loss": 0.564,
+      "step": 2394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.449444630892724e-05,
+      "loss": 1.056,
+      "step": 2395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.448109587830696e-05,
+      "loss": 1.261,
+      "step": 2396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.446774075843475e-05,
+      "loss": 0.5595,
+      "step": 2397
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.445438095112687e-05,
+      "loss": 1.5265,
+      "step": 2398
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.444101645820017e-05,
+      "loss": 0.973,
+      "step": 2399
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.442764728147217e-05,
+      "loss": 1.1071,
+      "step": 2400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.441427342276098e-05,
+      "loss": 0.8076,
+      "step": 2401
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.440089488388537e-05,
+      "loss": 1.4137,
+      "step": 2402
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43875116666648e-05,
+      "loss": 1.5087,
+      "step": 2403
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.437412377291927e-05,
+      "loss": 0.4953,
+      "step": 2404
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43607312044695e-05,
+      "loss": 1.2152,
+      "step": 2405
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.434733396313679e-05,
+      "loss": 1.5433,
+      "step": 2406
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43339320507431e-05,
+      "loss": 1.1626,
+      "step": 2407
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.432052546911102e-05,
+      "loss": 0.9771,
+      "step": 2408
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43071142200638e-05,
+      "loss": 1.3473,
+      "step": 2409
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.429369830542526e-05,
+      "loss": 0.5962,
+      "step": 2410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.428027772701992e-05,
+      "loss": 0.9067,
+      "step": 2411
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.42668524866729e-05,
+      "loss": 1.0437,
+      "step": 2412
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.425342258620998e-05,
+      "loss": 1.1588,
+      "step": 2413
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.423998802745754e-05,
+      "loss": 1.1129,
+      "step": 2414
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.422654881224263e-05,
+      "loss": 0.8845,
+      "step": 2415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.42131049423929e-05,
+      "loss": 0.6453,
+      "step": 2416
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.419965641973664e-05,
+      "loss": 1.0716,
+      "step": 2417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.418620324610279e-05,
+      "loss": 0.6555,
+      "step": 2418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.417274542332091e-05,
+      "loss": 0.9693,
+      "step": 2419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41592829532212e-05,
+      "loss": 1.3993,
+      "step": 2420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.414581583763448e-05,
+      "loss": 0.5874,
+      "step": 2421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.413234407839222e-05,
+      "loss": 1.0983,
+      "step": 2422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41188676773265e-05,
+      "loss": 0.8643,
+      "step": 2423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.410538663627003e-05,
+      "loss": 1.0634,
+      "step": 2424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.409190095705618e-05,
+      "loss": 0.7849,
+      "step": 2425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.407841064151896e-05,
+      "loss": 0.4918,
+      "step": 2426
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.406491569149293e-05,
+      "loss": 0.5919,
+      "step": 2427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.405141610881335e-05,
+      "loss": 1.3557,
+      "step": 2428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40379118953161e-05,
+      "loss": 1.0377,
+      "step": 2429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40244030528377e-05,
+      "loss": 1.6088,
+      "step": 2430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.401088958321524e-05,
+      "loss": 0.6447,
+      "step": 2431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.399737148828652e-05,
+      "loss": 1.3686,
+      "step": 2432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.398384876988994e-05,
+      "loss": 1.2269,
+      "step": 2433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.397032142986448e-05,
+      "loss": 1.9309,
+      "step": 2434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.39567894700498e-05,
+      "loss": 1.3472,
+      "step": 2435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.394325289228622e-05,
+      "loss": 0.4527,
+      "step": 2436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.392971169841459e-05,
+      "loss": 1.4407,
+      "step": 2437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.391616589027646e-05,
+      "loss": 1.6282,
+      "step": 2438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.390261546971402e-05,
+      "loss": 0.7889,
+      "step": 2439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.388906043857001e-05,
+      "loss": 1.5022,
+      "step": 2440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.387550079868788e-05,
+      "loss": 1.3092,
+      "step": 2441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.386193655191165e-05,
+      "loss": 1.3299,
+      "step": 2442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.384836770008601e-05,
+      "loss": 0.8005,
+      "step": 2443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.383479424505622e-05,
+      "loss": 1.103,
+      "step": 2444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.382121618866824e-05,
+      "loss": 1.2305,
+      "step": 2445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.380763353276858e-05,
+      "loss": 1.0838,
+      "step": 2446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.379404627920445e-05,
+      "loss": 1.3531,
+      "step": 2447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.378045442982359e-05,
+      "loss": 1.3284,
+      "step": 2448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.37668579864745e-05,
+      "loss": 1.3378,
+      "step": 2449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.375325695100615e-05,
+      "loss": 1.0074,
+      "step": 2450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.373965132526826e-05,
+      "loss": 1.0582,
+      "step": 2451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.37260411111111e-05,
+      "loss": 1.3831,
+      "step": 2452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.37124263103856e-05,
+      "loss": 1.0039,
+      "step": 2453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.36988069249433e-05,
+      "loss": 0.5172,
+      "step": 2454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.368518295663637e-05,
+      "loss": 0.8668,
+      "step": 2455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.36715544073176e-05,
+      "loss": 1.3683,
+      "step": 2456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.36579212788404e-05,
+      "loss": 1.2518,
+      "step": 2457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.364428357305881e-05,
+      "loss": 1.3448,
+      "step": 2458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.363064129182747e-05,
+      "loss": 1.3638,
+      "step": 2459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.361699443700168e-05,
+      "loss": 1.2998,
+      "step": 2460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.360334301043735e-05,
+      "loss": 0.984,
+      "step": 2461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.358968701399097e-05,
+      "loss": 1.2607,
+      "step": 2462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.357602644951971e-05,
+      "loss": 0.6904,
+      "step": 2463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.356236131888133e-05,
+      "loss": 0.6985,
+      "step": 2464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.354869162393422e-05,
+      "loss": 1.2771,
+      "step": 2465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.35350173665374e-05,
+      "loss": 0.3937,
+      "step": 2466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.352133854855047e-05,
+      "loss": 0.6593,
+      "step": 2467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.350765517183372e-05,
+      "loss": 0.7514,
+      "step": 2468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.349396723824796e-05,
+      "loss": 0.6964,
+      "step": 2469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.348027474965474e-05,
+      "loss": 1.1421,
+      "step": 2470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.346657770791612e-05,
+      "loss": 1.0258,
+      "step": 2471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.345287611489488e-05,
+      "loss": 1.3046,
+      "step": 2472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.34391699724543e-05,
+      "loss": 0.8924,
+      "step": 2473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.34254592824584e-05,
+      "loss": 1.571,
+      "step": 2474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.341174404677174e-05,
+      "loss": 0.5246,
+      "step": 2475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.339802426725951e-05,
+      "loss": 2.025,
+      "step": 2476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.338429994578755e-05,
+      "loss": 0.4574,
+      "step": 2477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.337057108422231e-05,
+      "loss": 1.3522,
+      "step": 2478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.33568376844308e-05,
+      "loss": 1.1829,
+      "step": 2479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.334309974828073e-05,
+      "loss": 1.0827,
+      "step": 2480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.332935727764038e-05,
+      "loss": 1.4027,
+      "step": 2481
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.331561027437864e-05,
+      "loss": 1.007,
+      "step": 2482
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.330185874036505e-05,
+      "loss": 0.7145,
+      "step": 2483
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.328810267746976e-05,
+      "loss": 1.2983,
+      "step": 2484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.327434208756348e-05,
+      "loss": 0.696,
+      "step": 2485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.326057697251761e-05,
+      "loss": 0.6998,
+      "step": 2486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.324680733420415e-05,
+      "loss": 1.2657,
+      "step": 2487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.323303317449566e-05,
+      "loss": 1.5628,
+      "step": 2488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.321925449526537e-05,
+      "loss": 1.3188,
+      "step": 2489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.320547129838714e-05,
+      "loss": 0.9242,
+      "step": 2490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.319168358573536e-05,
+      "loss": 1.3693,
+      "step": 2491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.317789135918515e-05,
+      "loss": 1.3265,
+      "step": 2492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.316409462061212e-05,
+      "loss": 0.6994,
+      "step": 2493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.315029337189259e-05,
+      "loss": 1.1428,
+      "step": 2494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.313648761490346e-05,
+      "loss": 1.2492,
+      "step": 2495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.312267735152224e-05,
+      "loss": 1.1417,
+      "step": 2496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.310886258362703e-05,
+      "loss": 1.373,
+      "step": 2497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.309504331309662e-05,
+      "loss": 1.3808,
+      "step": 2498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.30812195418103e-05,
+      "loss": 1.2683,
+      "step": 2499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.306739127164807e-05,
+      "loss": 1.1342,
+      "step": 2500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.305355850449048e-05,
+      "loss": 1.5793,
+      "step": 2501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.303972124221874e-05,
+      "loss": 1.4194,
+      "step": 2502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.302587948671463e-05,
+      "loss": 1.1132,
+      "step": 2503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.301203323986058e-05,
+      "loss": 0.6649,
+      "step": 2504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.299818250353956e-05,
+      "loss": 1.4005,
+      "step": 2505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.298432727963525e-05,
+      "loss": 1.4446,
+      "step": 2506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.297046757003186e-05,
+      "loss": 0.9864,
+      "step": 2507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.295660337661427e-05,
+      "loss": 1.2849,
+      "step": 2508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.29427347012679e-05,
+      "loss": 0.7874,
+      "step": 2509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.292886154587884e-05,
+      "loss": 1.4058,
+      "step": 2510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.291498391233377e-05,
+      "loss": 1.2552,
+      "step": 2511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.290110180251996e-05,
+      "loss": 1.3012,
+      "step": 2512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.288721521832532e-05,
+      "loss": 0.7529,
+      "step": 2513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.287332416163836e-05,
+      "loss": 1.5674,
+      "step": 2514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.285942863434816e-05,
+      "loss": 1.0868,
+      "step": 2515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.284552863834447e-05,
+      "loss": 1.4505,
+      "step": 2516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.283162417551763e-05,
+      "loss": 0.8479,
+      "step": 2517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.281771524775854e-05,
+      "loss": 1.3025,
+      "step": 2518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.280380185695878e-05,
+      "loss": 0.7263,
+      "step": 2519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.278988400501045e-05,
+      "loss": 0.7277,
+      "step": 2520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.277596169380636e-05,
+      "loss": 1.3019,
+      "step": 2521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.276203492523984e-05,
+      "loss": 0.959,
+      "step": 2522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.274810370120486e-05,
+      "loss": 1.1128,
+      "step": 2523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.273416802359601e-05,
+      "loss": 1.4485,
+      "step": 2524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.272022789430847e-05,
+      "loss": 1.0449,
+      "step": 2525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.270628331523801e-05,
+      "loss": 1.1706,
+      "step": 2526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.269233428828103e-05,
+      "loss": 0.9904,
+      "step": 2527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.267838081533453e-05,
+      "loss": 1.224,
+      "step": 2528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.266442289829611e-05,
+      "loss": 1.3055,
+      "step": 2529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.265046053906397e-05,
+      "loss": 1.288,
+      "step": 2530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.263649373953693e-05,
+      "loss": 1.437,
+      "step": 2531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.262252250161436e-05,
+      "loss": 1.3497,
+      "step": 2532
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.260854682719634e-05,
+      "loss": 1.27,
+      "step": 2533
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.259456671818345e-05,
+      "loss": 1.3602,
+      "step": 2534
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.258058217647693e-05,
+      "loss": 1.6776,
+      "step": 2535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.25665932039786e-05,
+      "loss": 1.3064,
+      "step": 2536
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.255259980259087e-05,
+      "loss": 1.3022,
+      "step": 2537
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.253860197421681e-05,
+      "loss": 0.5898,
+      "step": 2538
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.252459972076002e-05,
+      "loss": 1.1681,
+      "step": 2539
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.251059304412475e-05,
+      "loss": 1.475,
+      "step": 2540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.249658194621583e-05,
+      "loss": 1.2957,
+      "step": 2541
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.24825664289387e-05,
+      "loss": 1.117,
+      "step": 2542
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.246854649419939e-05,
+      "loss": 1.1267,
+      "step": 2543
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.245452214390456e-05,
+      "loss": 1.0522,
+      "step": 2544
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.244049337996143e-05,
+      "loss": 1.2319,
+      "step": 2545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.242646020427784e-05,
+      "loss": 1.0072,
+      "step": 2546
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.241242261876225e-05,
+      "loss": 1.4658,
+      "step": 2547
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.239838062532368e-05,
+      "loss": 0.7935,
+      "step": 2548
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.238433422587177e-05,
+      "loss": 0.9205,
+      "step": 2549
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.237028342231676e-05,
+      "loss": 1.3333,
+      "step": 2550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.23562282165695e-05,
+      "loss": 1.2082,
+      "step": 2551
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.23421686105414e-05,
+      "loss": 0.7073,
+      "step": 2552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.232810460614452e-05,
+      "loss": 1.4223,
+      "step": 2553
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.231403620529148e-05,
+      "loss": 1.0421,
+      "step": 2554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.229996340989551e-05,
+      "loss": 0.8019,
+      "step": 2555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.228588622187044e-05,
+      "loss": 0.6944,
+      "step": 2556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.227180464313069e-05,
+      "loss": 0.6866,
+      "step": 2557
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.225771867559131e-05,
+      "loss": 0.4706,
+      "step": 2558
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.224362832116787e-05,
+      "loss": 1.4264,
+      "step": 2559
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.222953358177662e-05,
+      "loss": 1.2392,
+      "step": 2560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.221543445933438e-05,
+      "loss": 0.8994,
+      "step": 2561
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.220133095575852e-05,
+      "loss": 0.9158,
+      "step": 2562
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.218722307296709e-05,
+      "loss": 1.1631,
+      "step": 2563
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.217311081287865e-05,
+      "loss": 0.6431,
+      "step": 2564
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.215899417741241e-05,
+      "loss": 1.1592,
+      "step": 2565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.214487316848817e-05,
+      "loss": 1.3177,
+      "step": 2566
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.213074778802631e-05,
+      "loss": 0.8914,
+      "step": 2567
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.211661803794782e-05,
+      "loss": 1.3886,
+      "step": 2568
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.210248392017423e-05,
+      "loss": 1.2149,
+      "step": 2569
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.208834543662774e-05,
+      "loss": 1.2407,
+      "step": 2570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.20742025892311e-05,
+      "loss": 1.3635,
+      "step": 2571
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.20600553799077e-05,
+      "loss": 1.5795,
+      "step": 2572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.204590381058143e-05,
+      "loss": 1.0256,
+      "step": 2573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.203174788317688e-05,
+      "loss": 1.5081,
+      "step": 2574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.201758759961917e-05,
+      "loss": 1.4164,
+      "step": 2575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.200342296183401e-05,
+      "loss": 1.1683,
+      "step": 2576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.198925397174773e-05,
+      "loss": 1.228,
+      "step": 2577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.197508063128722e-05,
+      "loss": 1.123,
+      "step": 2578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.196090294238001e-05,
+      "loss": 1.314,
+      "step": 2579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.194672090695419e-05,
+      "loss": 1.3538,
+      "step": 2580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.193253452693843e-05,
+      "loss": 1.4,
+      "step": 2581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.191834380426201e-05,
+      "loss": 0.5604,
+      "step": 2582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.190414874085479e-05,
+      "loss": 1.3406,
+      "step": 2583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.188994933864725e-05,
+      "loss": 1.0433,
+      "step": 2584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.187574559957039e-05,
+      "loss": 0.534,
+      "step": 2585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.18615375255559e-05,
+      "loss": 1.0993,
+      "step": 2586
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.184732511853597e-05,
+      "loss": 1.1616,
+      "step": 2587
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.183310838044343e-05,
+      "loss": 1.5706,
+      "step": 2588
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.181888731321166e-05,
+      "loss": 1.1005,
+      "step": 2589
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.180466191877467e-05,
+      "loss": 0.8525,
+      "step": 2590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.179043219906704e-05,
+      "loss": 1.3481,
+      "step": 2591
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.177619815602394e-05,
+      "loss": 1.2237,
+      "step": 2592
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.176195979158112e-05,
+      "loss": 0.65,
+      "step": 2593
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.174771710767492e-05,
+      "loss": 0.8433,
+      "step": 2594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.17334701062423e-05,
+      "loss": 0.5531,
+      "step": 2595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.171921878922076e-05,
+      "loss": 0.5562,
+      "step": 2596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.170496315854839e-05,
+      "loss": 1.0778,
+      "step": 2597
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.16907032161639e-05,
+      "loss": 1.0844,
+      "step": 2598
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.167643896400656e-05,
+      "loss": 0.8803,
+      "step": 2599
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.166217040401627e-05,
+      "loss": 1.1697,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.164789753813343e-05,
+      "loss": 0.5873,
+      "step": 2601
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.163362036829909e-05,
+      "loss": 1.2329,
+      "step": 2602
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.16193388964549e-05,
+      "loss": 1.4534,
+      "step": 2603
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.160505312454304e-05,
+      "loss": 1.2766,
+      "step": 2604
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.159076305450631e-05,
+      "loss": 1.1267,
+      "step": 2605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.157646868828808e-05,
+      "loss": 1.3842,
+      "step": 2606
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.156217002783232e-05,
+      "loss": 1.0796,
+      "step": 2607
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.154786707508358e-05,
+      "loss": 0.6807,
+      "step": 2608
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.153355983198698e-05,
+      "loss": 1.1006,
+      "step": 2609
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.15192483004882e-05,
+      "loss": 1.0328,
+      "step": 2610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.15049324825336e-05,
+      "loss": 1.1176,
+      "step": 2611
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.149061238007e-05,
+      "loss": 1.302,
+      "step": 2612
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.14762879950449e-05,
+      "loss": 1.328,
+      "step": 2613
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.14619593294063e-05,
+      "loss": 0.7238,
+      "step": 2614
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.144762638510287e-05,
+      "loss": 1.052,
+      "step": 2615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.143328916408378e-05,
+      "loss": 0.6648,
+      "step": 2616
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.141894766829885e-05,
+      "loss": 1.0255,
+      "step": 2617
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.140460189969843e-05,
+      "loss": 0.7866,
+      "step": 2618
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.139025186023348e-05,
+      "loss": 0.5926,
+      "step": 2619
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.137589755185552e-05,
+      "loss": 1.2713,
+      "step": 2620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.136153897651668e-05,
+      "loss": 1.4008,
+      "step": 2621
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.134717613616963e-05,
+      "loss": 1.1289,
+      "step": 2622
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.133280903276766e-05,
+      "loss": 0.7589,
+      "step": 2623
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.131843766826461e-05,
+      "loss": 1.544,
+      "step": 2624
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.130406204461494e-05,
+      "loss": 1.7416,
+      "step": 2625
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.128968216377363e-05,
+      "loss": 1.4902,
+      "step": 2626
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.127529802769627e-05,
+      "loss": 1.0631,
+      "step": 2627
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.126090963833905e-05,
+      "loss": 1.1352,
+      "step": 2628
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.12465169976587e-05,
+      "loss": 1.1756,
+      "step": 2629
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.123212010761255e-05,
+      "loss": 1.0254,
+      "step": 2630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.12177189701585e-05,
+      "loss": 0.5475,
+      "step": 2631
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.120331358725504e-05,
+      "loss": 1.2488,
+      "step": 2632
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.118890396086123e-05,
+      "loss": 1.27,
+      "step": 2633
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.117449009293668e-05,
+      "loss": 0.6259,
+      "step": 2634
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.116007198544163e-05,
+      "loss": 0.7645,
+      "step": 2635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.114564964033687e-05,
+      "loss": 1.3594,
+      "step": 2636
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.113122305958372e-05,
+      "loss": 0.9136,
+      "step": 2637
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.111679224514419e-05,
+      "loss": 1.0507,
+      "step": 2638
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.110235719898075e-05,
+      "loss": 0.6509,
+      "step": 2639
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.108791792305647e-05,
+      "loss": 1.3174,
+      "step": 2640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.107347441933508e-05,
+      "loss": 1.3423,
+      "step": 2641
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.105902668978081e-05,
+      "loss": 0.824,
+      "step": 2642
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.104457473635843e-05,
+      "loss": 1.7146,
+      "step": 2643
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.103011856103335e-05,
+      "loss": 1.351,
+      "step": 2644
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.101565816577158e-05,
+      "loss": 0.8542,
+      "step": 2645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.100119355253962e-05,
+      "loss": 0.9077,
+      "step": 2646
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.098672472330458e-05,
+      "loss": 1.3023,
+      "step": 2647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.097225168003414e-05,
+      "loss": 1.2549,
+      "step": 2648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.095777442469659e-05,
+      "loss": 0.5595,
+      "step": 2649
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.094329295926075e-05,
+      "loss": 1.6417,
+      "step": 2650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.092880728569602e-05,
+      "loss": 1.2315,
+      "step": 2651
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.091431740597237e-05,
+      "loss": 1.0714,
+      "step": 2652
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.089982332206039e-05,
+      "loss": 1.2925,
+      "step": 2653
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.088532503593114e-05,
+      "loss": 1.3731,
+      "step": 2654
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.087082254955634e-05,
+      "loss": 0.8094,
+      "step": 2655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.085631586490827e-05,
+      "loss": 1.2214,
+      "step": 2656
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.084180498395975e-05,
+      "loss": 1.26,
+      "step": 2657
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.082728990868418e-05,
+      "loss": 1.4112,
+      "step": 2658
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.081277064105554e-05,
+      "loss": 1.0052,
+      "step": 2659
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.079824718304837e-05,
+      "loss": 1.4503,
+      "step": 2660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.078371953663781e-05,
+      "loss": 1.4056,
+      "step": 2661
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.076918770379951e-05,
+      "loss": 0.5864,
+      "step": 2662
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.075465168650975e-05,
+      "loss": 1.3086,
+      "step": 2663
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.074011148674533e-05,
+      "loss": 1.6541,
+      "step": 2664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.072556710648369e-05,
+      "loss": 1.1637,
+      "step": 2665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.071101854770276e-05,
+      "loss": 0.7137,
+      "step": 2666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.069646581238103e-05,
+      "loss": 1.3723,
+      "step": 2667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.068190890249767e-05,
+      "loss": 1.0558,
+      "step": 2668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.06673478200323e-05,
+      "loss": 1.4983,
+      "step": 2669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.065278256696518e-05,
+      "loss": 1.3274,
+      "step": 2670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.063821314527706e-05,
+      "loss": 0.5666,
+      "step": 2671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.062363955694936e-05,
+      "loss": 1.2222,
+      "step": 2672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.060906180396399e-05,
+      "loss": 1.0029,
+      "step": 2673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.059447988830346e-05,
+      "loss": 0.8742,
+      "step": 2674
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.057989381195082e-05,
+      "loss": 0.8258,
+      "step": 2675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.056530357688971e-05,
+      "loss": 1.1982,
+      "step": 2676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.055070918510432e-05,
+      "loss": 1.5948,
+      "step": 2677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.053611063857944e-05,
+      "loss": 1.414,
+      "step": 2678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.052150793930036e-05,
+      "loss": 1.4606,
+      "step": 2679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.050690108925298e-05,
+      "loss": 1.3661,
+      "step": 2680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.049229009042378e-05,
+      "loss": 1.0711,
+      "step": 2681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.047767494479977e-05,
+      "loss": 1.336,
+      "step": 2682
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.046305565436852e-05,
+      "loss": 1.3537,
+      "step": 2683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.044843222111818e-05,
+      "loss": 0.6579,
+      "step": 2684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.043380464703748e-05,
+      "loss": 1.0727,
+      "step": 2685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.041917293411567e-05,
+      "loss": 1.0962,
+      "step": 2686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.040453708434263e-05,
+      "loss": 0.8728,
+      "step": 2687
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.038989709970872e-05,
+      "loss": 1.1754,
+      "step": 2688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.037525298220492e-05,
+      "loss": 1.2233,
+      "step": 2689
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.036060473382275e-05,
+      "loss": 1.1851,
+      "step": 2690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.034595235655429e-05,
+      "loss": 1.2604,
+      "step": 2691
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.033129585239221e-05,
+      "loss": 0.7424,
+      "step": 2692
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.03166352233297e-05,
+      "loss": 0.6116,
+      "step": 2693
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.030197047136052e-05,
+      "loss": 0.6583,
+      "step": 2694
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.028730159847903e-05,
+      "loss": 1.341,
+      "step": 2695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.02726286066801e-05,
+      "loss": 0.5429,
+      "step": 2696
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.02579514979592e-05,
+      "loss": 1.2198,
+      "step": 2697
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.024327027431233e-05,
+      "loss": 1.3275,
+      "step": 2698
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.022858493773606e-05,
+      "loss": 0.6188,
+      "step": 2699
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.02138954902275e-05,
+      "loss": 1.0225,
+      "step": 2700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.019920193378439e-05,
+      "loss": 0.9326,
+      "step": 2701
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.018450427040494e-05,
+      "loss": 0.5135,
+      "step": 2702
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.016980250208796e-05,
+      "loss": 1.7048,
+      "step": 2703
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.015509663083283e-05,
+      "loss": 1.1913,
+      "step": 2704
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.014038665863946e-05,
+      "loss": 1.3728,
+      "step": 2705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.012567258750832e-05,
+      "loss": 1.128,
+      "step": 2706
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.011095441944047e-05,
+      "loss": 1.3296,
+      "step": 2707
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.009623215643751e-05,
+      "loss": 1.6761,
+      "step": 2708
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.008150580050157e-05,
+      "loss": 1.3116,
+      "step": 2709
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.006677535363538e-05,
+      "loss": 1.6414,
+      "step": 2710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.005204081784217e-05,
+      "loss": 1.4409,
+      "step": 2711
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.003730219512582e-05,
+      "loss": 1.138,
+      "step": 2712
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.002255948749064e-05,
+      "loss": 1.4518,
+      "step": 2713
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.000781269694163e-05,
+      "loss": 1.27,
+      "step": 2714
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.99930618254842e-05,
+      "loss": 1.4912,
+      "step": 2715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.997830687512446e-05,
+      "loss": 1.0463,
+      "step": 2716
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.996354784786896e-05,
+      "loss": 1.2908,
+      "step": 2717
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.99487847457249e-05,
+      "loss": 1.0329,
+      "step": 2718
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.993401757069997e-05,
+      "loss": 0.6424,
+      "step": 2719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.991924632480238e-05,
+      "loss": 1.1412,
+      "step": 2720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.990447101004101e-05,
+      "loss": 1.1686,
+      "step": 2721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.98896916284252e-05,
+      "loss": 1.2551,
+      "step": 2722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.987490818196487e-05,
+      "loss": 1.0587,
+      "step": 2723
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.98601206726705e-05,
+      "loss": 0.9278,
+      "step": 2724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.984532910255309e-05,
+      "loss": 0.7657,
+      "step": 2725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.983053347362426e-05,
+      "loss": 1.2908,
+      "step": 2726
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.981573378789608e-05,
+      "loss": 0.8306,
+      "step": 2727
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.98009300473813e-05,
+      "loss": 1.1995,
+      "step": 2728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.978612225409312e-05,
+      "loss": 1.2109,
+      "step": 2729
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.97713104100453e-05,
+      "loss": 1.2202,
+      "step": 2730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.975649451725222e-05,
+      "loss": 1.1469,
+      "step": 2731
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.974167457772873e-05,
+      "loss": 1.254,
+      "step": 2732
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.972685059349028e-05,
+      "loss": 0.647,
+      "step": 2733
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.971202256655287e-05,
+      "loss": 0.7761,
+      "step": 2734
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.9697190498933e-05,
+      "loss": 1.0924,
+      "step": 2735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.96823543926478e-05,
+      "loss": 1.1897,
+      "step": 2736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.966751424971485e-05,
+      "loss": 1.1937,
+      "step": 2737
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.965267007215238e-05,
+      "loss": 1.5587,
+      "step": 2738
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.963782186197912e-05,
+      "loss": 1.7553,
+      "step": 2739
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.962296962121431e-05,
+      "loss": 1.0805,
+      "step": 2740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.960811335187781e-05,
+      "loss": 0.8803,
+      "step": 2741
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.959325305598999e-05,
+      "loss": 1.1027,
+      "step": 2742
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.957838873557178e-05,
+      "loss": 0.8613,
+      "step": 2743
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.956352039264465e-05,
+      "loss": 1.4275,
+      "step": 2744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.954864802923061e-05,
+      "loss": 0.5456,
+      "step": 2745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.95337716473522e-05,
+      "loss": 0.6209,
+      "step": 2746
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.95188912490326e-05,
+      "loss": 1.1084,
+      "step": 2747
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.950400683629541e-05,
+      "loss": 1.2323,
+      "step": 2748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.948911841116485e-05,
+      "loss": 1.0033,
+      "step": 2749
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.947422597566568e-05,
+      "loss": 1.1667,
+      "step": 2750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.945932953182317e-05,
+      "loss": 1.0837,
+      "step": 2751
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.944442908166318e-05,
+      "loss": 1.5271,
+      "step": 2752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.942952462721209e-05,
+      "loss": 0.8862,
+      "step": 2753
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.941461617049684e-05,
+      "loss": 1.5344,
+      "step": 2754
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.939970371354488e-05,
+      "loss": 0.5385,
+      "step": 2755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.938478725838425e-05,
+      "loss": 1.2622,
+      "step": 2756
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.936986680704349e-05,
+      "loss": 1.4811,
+      "step": 2757
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.935494236155172e-05,
+      "loss": 0.6126,
+      "step": 2758
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.934001392393859e-05,
+      "loss": 0.5301,
+      "step": 2759
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.932508149623428e-05,
+      "loss": 1.1797,
+      "step": 2760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.931014508046953e-05,
+      "loss": 1.5636,
+      "step": 2761
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.929520467867561e-05,
+      "loss": 1.0325,
+      "step": 2762
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.928026029288436e-05,
+      "loss": 1.5168,
+      "step": 2763
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.926531192512812e-05,
+      "loss": 0.6578,
+      "step": 2764
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.925035957743977e-05,
+      "loss": 1.4127,
+      "step": 2765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.923540325185279e-05,
+      "loss": 1.2462,
+      "step": 2766
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.922044295040115e-05,
+      "loss": 0.9273,
+      "step": 2767
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.920547867511936e-05,
+      "loss": 1.1102,
+      "step": 2768
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.91905104280425e-05,
+      "loss": 0.982,
+      "step": 2769
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.91755382112062e-05,
+      "loss": 1.0658,
+      "step": 2770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.916056202664654e-05,
+      "loss": 0.9914,
+      "step": 2771
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.914558187640026e-05,
+      "loss": 1.2888,
+      "step": 2772
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.913059776250456e-05,
+      "loss": 1.1552,
+      "step": 2773
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.911560968699721e-05,
+      "loss": 1.413,
+      "step": 2774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.91006176519165e-05,
+      "loss": 0.9331,
+      "step": 2775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.908562165930129e-05,
+      "loss": 0.7847,
+      "step": 2776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.907062171119095e-05,
+      "loss": 1.3746,
+      "step": 2777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.905561780962538e-05,
+      "loss": 1.5938,
+      "step": 2778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.904060995664505e-05,
+      "loss": 1.0822,
+      "step": 2779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.902559815429093e-05,
+      "loss": 0.9301,
+      "step": 2780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.90105824046046e-05,
+      "loss": 0.9354,
+      "step": 2781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.899556270962805e-05,
+      "loss": 0.554,
+      "step": 2782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.898053907140395e-05,
+      "loss": 0.6364,
+      "step": 2783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.89655114919754e-05,
+      "loss": 0.8728,
+      "step": 2784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.895047997338608e-05,
+      "loss": 1.0139,
+      "step": 2785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.89354445176802e-05,
+      "loss": 1.3873,
+      "step": 2786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.892040512690251e-05,
+      "loss": 1.3582,
+      "step": 2787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.890536180309828e-05,
+      "loss": 1.466,
+      "step": 2788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.889031454831335e-05,
+      "loss": 1.0637,
+      "step": 2789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.887526336459401e-05,
+      "loss": 1.1443,
+      "step": 2790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.886020825398722e-05,
+      "loss": 1.4623,
+      "step": 2791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.884514921854035e-05,
+      "loss": 1.3958,
+      "step": 2792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.883008626030135e-05,
+      "loss": 1.0033,
+      "step": 2793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.881501938131874e-05,
+      "loss": 1.2149,
+      "step": 2794
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.87999485836415e-05,
+      "loss": 1.2694,
+      "step": 2795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.878487386931922e-05,
+      "loss": 1.3307,
+      "step": 2796
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.876979524040194e-05,
+      "loss": 0.8987,
+      "step": 2797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.87547126989403e-05,
+      "loss": 1.2193,
+      "step": 2798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.873962624698543e-05,
+      "loss": 1.597,
+      "step": 2799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.872453588658904e-05,
+      "loss": 1.4457,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.870944161980333e-05,
+      "loss": 1.2887,
+      "step": 2801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.869434344868102e-05,
+      "loss": 0.9639,
+      "step": 2802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.867924137527542e-05,
+      "loss": 0.5048,
+      "step": 2803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.866413540164033e-05,
+      "loss": 1.3173,
+      "step": 2804
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.864902552983004e-05,
+      "loss": 0.8981,
+      "step": 2805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.863391176189946e-05,
+      "loss": 1.0365,
+      "step": 2806
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.861879409990398e-05,
+      "loss": 0.6572,
+      "step": 2807
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.860367254589953e-05,
+      "loss": 0.6817,
+      "step": 2808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.858854710194253e-05,
+      "loss": 1.4537,
+      "step": 2809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.857341777008997e-05,
+      "loss": 1.1148,
+      "step": 2810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.85582845523994e-05,
+      "loss": 0.9171,
+      "step": 2811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.854314745092882e-05,
+      "loss": 1.4711,
+      "step": 2812
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.852800646773682e-05,
+      "loss": 1.3544,
+      "step": 2813
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.851286160488248e-05,
+      "loss": 1.1805,
+      "step": 2814
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.849771286442543e-05,
+      "loss": 1.3585,
+      "step": 2815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.848256024842582e-05,
+      "loss": 0.9182,
+      "step": 2816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.846740375894432e-05,
+      "loss": 1.4733,
+      "step": 2817
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.845224339804217e-05,
+      "loss": 1.3234,
+      "step": 2818
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.843707916778106e-05,
+      "loss": 1.4409,
+      "step": 2819
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.842191107022328e-05,
+      "loss": 0.6988,
+      "step": 2820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.840673910743158e-05,
+      "loss": 0.8991,
+      "step": 2821
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.839156328146928e-05,
+      "loss": 1.2963,
+      "step": 2822
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.837638359440025e-05,
+      "loss": 0.7081,
+      "step": 2823
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.83612000482888e-05,
+      "loss": 1.2972,
+      "step": 2824
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.834601264519983e-05,
+      "loss": 1.4053,
+      "step": 2825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.833082138719877e-05,
+      "loss": 1.2971,
+      "step": 2826
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.831562627635152e-05,
+      "loss": 0.5379,
+      "step": 2827
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.830042731472454e-05,
+      "loss": 1.609,
+      "step": 2828
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.828522450438485e-05,
+      "loss": 0.7982,
+      "step": 2829
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.827001784739992e-05,
+      "loss": 1.0962,
+      "step": 2830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.825480734583778e-05,
+      "loss": 1.2205,
+      "step": 2831
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.823959300176697e-05,
+      "loss": 1.2109,
+      "step": 2832
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.82243748172566e-05,
+      "loss": 0.8626,
+      "step": 2833
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.820915279437624e-05,
+      "loss": 0.7404,
+      "step": 2834
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.8193926935196e-05,
+      "loss": 1.3675,
+      "step": 2835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.817869724178655e-05,
+      "loss": 1.4726,
+      "step": 2836
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.8163463716219e-05,
+      "loss": 1.0859,
+      "step": 2837
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.814822636056508e-05,
+      "loss": 1.3199,
+      "step": 2838
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.813298517689696e-05,
+      "loss": 1.5721,
+      "step": 2839
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.811774016728738e-05,
+      "loss": 1.2451,
+      "step": 2840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.810249133380957e-05,
+      "loss": 1.4543,
+      "step": 2841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.808723867853732e-05,
+      "loss": 0.7163,
+      "step": 2842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.807198220354486e-05,
+      "loss": 1.0005,
+      "step": 2843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.805672191090704e-05,
+      "loss": 0.474,
+      "step": 2844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.804145780269919e-05,
+      "loss": 0.8227,
+      "step": 2845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.80261898809971e-05,
+      "loss": 0.9414,
+      "step": 2846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.801091814787715e-05,
+      "loss": 1.3259,
+      "step": 2847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.801091814787715e-05,
+      "loss": 1.5345,
+      "step": 2848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.799564260541624e-05,
+      "loss": 0.8419,
+      "step": 2849
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.798036325569173e-05,
+      "loss": 0.8569,
+      "step": 2850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.796508010078159e-05,
+      "loss": 1.2941,
+      "step": 2851
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.794979314276418e-05,
+      "loss": 1.6545,
+      "step": 2852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.793450238371848e-05,
+      "loss": 1.2243,
+      "step": 2853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.791920782572397e-05,
+      "loss": 1.4236,
+      "step": 2854
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.790390947086059e-05,
+      "loss": 0.9814,
+      "step": 2855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.788860732120889e-05,
+      "loss": 0.8329,
+      "step": 2856
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.787330137884983e-05,
+      "loss": 1.5022,
+      "step": 2857
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.785799164586499e-05,
+      "loss": 1.3155,
+      "step": 2858
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.784267812433639e-05,
+      "loss": 1.4304,
+      "step": 2859
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.782736081634658e-05,
+      "loss": 1.1028,
+      "step": 2860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.781203972397866e-05,
+      "loss": 0.678,
+      "step": 2861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.77967148493162e-05,
+      "loss": 0.8514,
+      "step": 2862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.77813861944433e-05,
+      "loss": 0.6659,
+      "step": 2863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.776605376144461e-05,
+      "loss": 1.3119,
+      "step": 2864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.775071755240522e-05,
+      "loss": 1.5858,
+      "step": 2865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.773537756941082e-05,
+      "loss": 1.181,
+      "step": 2866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.772003381454754e-05,
+      "loss": 1.3072,
+      "step": 2867
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.770468628990206e-05,
+      "loss": 0.5243,
+      "step": 2868
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.768933499756155e-05,
+      "loss": 0.619,
+      "step": 2869
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.767397993961373e-05,
+      "loss": 0.4387,
+      "step": 2870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.765862111814679e-05,
+      "loss": 1.4135,
+      "step": 2871
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.764325853524947e-05,
+      "loss": 0.8898,
+      "step": 2872
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.762789219301099e-05,
+      "loss": 1.2417,
+      "step": 2873
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.761252209352109e-05,
+      "loss": 1.3116,
+      "step": 2874
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.759714823887003e-05,
+      "loss": 1.1978,
+      "step": 2875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.758177063114858e-05,
+      "loss": 1.2916,
+      "step": 2876
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.7566389272448e-05,
+      "loss": 1.4441,
+      "step": 2877
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.755100416486008e-05,
+      "loss": 1.0923,
+      "step": 2878
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.753561531047715e-05,
+      "loss": 1.2324,
+      "step": 2879
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.752022271139196e-05,
+      "loss": 1.2645,
+      "step": 2880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.750482636969787e-05,
+      "loss": 1.228,
+      "step": 2881
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.748942628748867e-05,
+      "loss": 1.1569,
+      "step": 2882
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.747402246685873e-05,
+      "loss": 0.9817,
+      "step": 2883
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.745861490990285e-05,
+      "loss": 1.0346,
+      "step": 2884
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.74432036187164e-05,
+      "loss": 0.7666,
+      "step": 2885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.742778859539523e-05,
+      "loss": 0.9983,
+      "step": 2886
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.741236984203572e-05,
+      "loss": 1.0001,
+      "step": 2887
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.73969473607347e-05,
+      "loss": 0.7251,
+      "step": 2888
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.738152115358961e-05,
+      "loss": 1.3054,
+      "step": 2889
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.736609122269828e-05,
+      "loss": 1.0759,
+      "step": 2890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.735065757015914e-05,
+      "loss": 1.1511,
+      "step": 2891
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.733522019807104e-05,
+      "loss": 0.9605,
+      "step": 2892
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.731977910853344e-05,
+      "loss": 0.9075,
+      "step": 2893
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.730433430364621e-05,
+      "loss": 1.0333,
+      "step": 2894
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.728888578550977e-05,
+      "loss": 1.1789,
+      "step": 2895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.727343355622505e-05,
+      "loss": 1.2668,
+      "step": 2896
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.725797761789348e-05,
+      "loss": 1.2542,
+      "step": 2897
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.724251797261696e-05,
+      "loss": 0.5471,
+      "step": 2898
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.722705462249796e-05,
+      "loss": 1.0323,
+      "step": 2899
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.721158756963938e-05,
+      "loss": 1.212,
+      "step": 2900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.719611681614467e-05,
+      "loss": 0.9517,
+      "step": 2901
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.718064236411779e-05,
+      "loss": 1.0353,
+      "step": 2902
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.716516421566317e-05,
+      "loss": 1.247,
+      "step": 2903
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.714968237288578e-05,
+      "loss": 0.7791,
+      "step": 2904
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.713419683789102e-05,
+      "loss": 1.4042,
+      "step": 2905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.711870761278491e-05,
+      "loss": 1.2307,
+      "step": 2906
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.710321469967385e-05,
+      "loss": 1.3266,
+      "step": 2907
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.708771810066484e-05,
+      "loss": 0.9086,
+      "step": 2908
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.707221781786532e-05,
+      "loss": 1.0651,
+      "step": 2909
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.705671385338323e-05,
+      "loss": 0.6581,
+      "step": 2910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.704120620932705e-05,
+      "loss": 1.4202,
+      "step": 2911
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.702569488780573e-05,
+      "loss": 1.5185,
+      "step": 2912
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.701017989092874e-05,
+      "loss": 0.6485,
+      "step": 2913
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.699466122080602e-05,
+      "loss": 0.7103,
+      "step": 2914
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.697913887954806e-05,
+      "loss": 0.8741,
+      "step": 2915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.69636128692658e-05,
+      "loss": 1.4162,
+      "step": 2916
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.69480831920707e-05,
+      "loss": 1.4046,
+      "step": 2917
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.693254985007472e-05,
+      "loss": 1.3307,
+      "step": 2918
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.69170128453903e-05,
+      "loss": 0.6789,
+      "step": 2919
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.690147218013045e-05,
+      "loss": 1.1288,
+      "step": 2920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.68859278564085e-05,
+      "loss": 1.1944,
+      "step": 2921
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.687037987633853e-05,
+      "loss": 0.6038,
+      "step": 2922
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.685482824203491e-05,
+      "loss": 0.5198,
+      "step": 2923
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.683927295561261e-05,
+      "loss": 0.4596,
+      "step": 2924
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.682371401918707e-05,
+      "loss": 1.5754,
+      "step": 2925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.680815143487419e-05,
+      "loss": 1.8346,
+      "step": 2926
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.679258520479045e-05,
+      "loss": 1.4211,
+      "step": 2927
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.677701533105273e-05,
+      "loss": 1.2615,
+      "step": 2928
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.67614418157785e-05,
+      "loss": 1.2323,
+      "step": 2929
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.674586466108564e-05,
+      "loss": 1.2684,
+      "step": 2930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.673028386909259e-05,
+      "loss": 0.9884,
+      "step": 2931
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.671469944191824e-05,
+      "loss": 0.5258,
+      "step": 2932
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.6699111381682e-05,
+      "loss": 1.124,
+      "step": 2933
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.668351969050376e-05,
+      "loss": 0.5411,
+      "step": 2934
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.666792437050392e-05,
+      "loss": 1.1819,
+      "step": 2935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.665232542380335e-05,
+      "loss": 0.6385,
+      "step": 2936
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.663672285252345e-05,
+      "loss": 1.6056,
+      "step": 2937
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.662111665878606e-05,
+      "loss": 0.9706,
+      "step": 2938
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.660550684471355e-05,
+      "loss": 1.4501,
+      "step": 2939
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.658989341242879e-05,
+      "loss": 1.0876,
+      "step": 2940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.657427636405514e-05,
+      "loss": 1.4747,
+      "step": 2941
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.655865570171639e-05,
+      "loss": 1.2235,
+      "step": 2942
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.65430314275369e-05,
+      "loss": 0.5638,
+      "step": 2943
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.652740354364148e-05,
+      "loss": 1.2821,
+      "step": 2944
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.651177205215546e-05,
+      "loss": 0.6673,
+      "step": 2945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.649613695520462e-05,
+      "loss": 0.8356,
+      "step": 2946
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.648049825491527e-05,
+      "loss": 1.5365,
+      "step": 2947
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.646485595341418e-05,
+      "loss": 0.822,
+      "step": 2948
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.644921005282863e-05,
+      "loss": 1.292,
+      "step": 2949
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.643356055528641e-05,
+      "loss": 0.6076,
+      "step": 2950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.641790746291571e-05,
+      "loss": 1.4684,
+      "step": 2951
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.640225077784531e-05,
+      "loss": 0.5926,
+      "step": 2952
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.638659050220442e-05,
+      "loss": 0.6104,
+      "step": 2953
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.63709266381228e-05,
+      "loss": 1.1959,
+      "step": 2954
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.63552591877306e-05,
+      "loss": 0.9444,
+      "step": 2955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.633958815315856e-05,
+      "loss": 0.6033,
+      "step": 2956
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.63239135365378e-05,
+      "loss": 0.9547,
+      "step": 2957
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.630823534000005e-05,
+      "loss": 1.1734,
+      "step": 2958
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.629255356567744e-05,
+      "loss": 0.9241,
+      "step": 2959
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.627686821570259e-05,
+      "loss": 0.6046,
+      "step": 2960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.626117929220866e-05,
+      "loss": 1.2337,
+      "step": 2961
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.624548679732925e-05,
+      "loss": 1.2491,
+      "step": 2962
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.622979073319845e-05,
+      "loss": 1.3841,
+      "step": 2963
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.621409110195085e-05,
+      "loss": 0.6673,
+      "step": 2964
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.619838790572153e-05,
+      "loss": 1.2164,
+      "step": 2965
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.618268114664604e-05,
+      "loss": 0.8906,
+      "step": 2966
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.616697082686039e-05,
+      "loss": 1.2261,
+      "step": 2967
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.615125694850114e-05,
+      "loss": 1.2008,
+      "step": 2968
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.613553951370529e-05,
+      "loss": 0.9342,
+      "step": 2969
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.611981852461032e-05,
+      "loss": 1.5033,
+      "step": 2970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.610409398335421e-05,
+      "loss": 1.2548,
+      "step": 2971
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.60883658920754e-05,
+      "loss": 0.7353,
+      "step": 2972
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.607263425291286e-05,
+      "loss": 1.114,
+      "step": 2973
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.605689906800599e-05,
+      "loss": 2.0336,
+      "step": 2974
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.60411603394947e-05,
+      "loss": 0.5963,
+      "step": 2975
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.602541806951937e-05,
+      "loss": 1.3711,
+      "step": 2976
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.600967226022089e-05,
+      "loss": 0.5293,
+      "step": 2977
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.599392291374056e-05,
+      "loss": 1.135,
+      "step": 2978
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.597817003222029e-05,
+      "loss": 1.1681,
+      "step": 2979
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.596241361780231e-05,
+      "loss": 1.5277,
+      "step": 2980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.594665367262946e-05,
+      "loss": 1.4937,
+      "step": 2981
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.593089019884499e-05,
+      "loss": 1.322,
+      "step": 2982
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.591512319859267e-05,
+      "loss": 1.5777,
+      "step": 2983
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.589935267401671e-05,
+      "loss": 1.3988,
+      "step": 2984
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.588357862726183e-05,
+      "loss": 1.0257,
+      "step": 2985
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.586780106047323e-05,
+      "loss": 0.8311,
+      "step": 2986
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.585201997579655e-05,
+      "loss": 1.1839,
+      "step": 2987
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.583623537537799e-05,
+      "loss": 1.1537,
+      "step": 2988
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.58204472613641e-05,
+      "loss": 1.5832,
+      "step": 2989
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.580465563590204e-05,
+      "loss": 1.3373,
+      "step": 2990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.578886050113935e-05,
+      "loss": 0.4877,
+      "step": 2991
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.577306185922413e-05,
+      "loss": 0.9629,
+      "step": 2992
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.57572597123049e-05,
+      "loss": 0.9389,
+      "step": 2993
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.574145406253064e-05,
+      "loss": 1.8104,
+      "step": 2994
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.572564491205087e-05,
+      "loss": 1.0273,
+      "step": 2995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.570983226301555e-05,
+      "loss": 1.295,
+      "step": 2996
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.56940161175751e-05,
+      "loss": 1.2588,
+      "step": 2997
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.567819647788046e-05,
+      "loss": 1.4218,
+      "step": 2998
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.5662373346083e-05,
+      "loss": 0.5969,
+      "step": 2999
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.56465467243346e-05,
+      "loss": 1.3902,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.563071661478758e-05,
+      "loss": 1.1849,
+      "step": 3001
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.561488301959478e-05,
+      "loss": 1.2483,
+      "step": 3002
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.559904594090945e-05,
+      "loss": 1.2132,
+      "step": 3003
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.558320538088539e-05,
+      "loss": 0.6679,
+      "step": 3004
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.55673613416768e-05,
+      "loss": 1.1381,
+      "step": 3005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.555151382543841e-05,
+      "loss": 1.3284,
+      "step": 3006
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.553566283432542e-05,
+      "loss": 1.1168,
+      "step": 3007
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.551980837049343e-05,
+      "loss": 1.4408,
+      "step": 3008
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.550395043609864e-05,
+      "loss": 0.4827,
+      "step": 3009
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.548808903329756e-05,
+      "loss": 1.1148,
+      "step": 3010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.547222416424733e-05,
+      "loss": 1.0598,
+      "step": 3011
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.545635583110544e-05,
+      "loss": 0.6475,
+      "step": 3012
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.544048403602995e-05,
+      "loss": 1.1252,
+      "step": 3013
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.542460878117928e-05,
+      "loss": 1.3297,
+      "step": 3014
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.540873006871246e-05,
+      "loss": 1.0204,
+      "step": 3015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.539284790078886e-05,
+      "loss": 0.6426,
+      "step": 3016
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.537696227956838e-05,
+      "loss": 1.0285,
+      "step": 3017
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.536107320721141e-05,
+      "loss": 0.7916,
+      "step": 3018
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.534518068587875e-05,
+      "loss": 0.6052,
+      "step": 3019
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.53292847177317e-05,
+      "loss": 0.5193,
+      "step": 3020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.531338530493207e-05,
+      "loss": 0.7241,
+      "step": 3021
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.529748244964207e-05,
+      "loss": 1.0104,
+      "step": 3022
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.528157615402439e-05,
+      "loss": 1.0685,
+      "step": 3023
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.526566642024224e-05,
+      "loss": 0.6144,
+      "step": 3024
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.524975325045924e-05,
+      "loss": 0.9709,
+      "step": 3025
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.52338366468395e-05,
+      "loss": 0.4815,
+      "step": 3026
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.52179166115476e-05,
+      "loss": 1.6873,
+      "step": 3027
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.520199314674861e-05,
+      "loss": 1.351,
+      "step": 3028
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.518606625460798e-05,
+      "loss": 1.0579,
+      "step": 3029
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.517013593729174e-05,
+      "loss": 0.916,
+      "step": 3030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.515420219696629e-05,
+      "loss": 1.0873,
+      "step": 3031
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.513826503579858e-05,
+      "loss": 1.2947,
+      "step": 3032
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.512232445595595e-05,
+      "loss": 1.3094,
+      "step": 3033
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.510638045960627e-05,
+      "loss": 1.1181,
+      "step": 3034
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.509043304891778e-05,
+      "loss": 0.434,
+      "step": 3035
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.50744822260593e-05,
+      "loss": 1.3244,
+      "step": 3036
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.505852799320004e-05,
+      "loss": 1.2987,
+      "step": 3037
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.50425703525097e-05,
+      "loss": 1.7825,
+      "step": 3038
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.502660930615843e-05,
+      "loss": 1.1778,
+      "step": 3039
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.501064485631688e-05,
+      "loss": 1.2249,
+      "step": 3040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.499467700515609e-05,
+      "loss": 0.6922,
+      "step": 3041
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.497870575484763e-05,
+      "loss": 1.3219,
+      "step": 3042
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.496273110756349e-05,
+      "loss": 0.9635,
+      "step": 3043
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.494675306547616e-05,
+      "loss": 0.6407,
+      "step": 3044
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.493077163075856e-05,
+      "loss": 1.0325,
+      "step": 3045
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.491478680558409e-05,
+      "loss": 1.4109,
+      "step": 3046
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.489879859212659e-05,
+      "loss": 1.4459,
+      "step": 3047
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.48828069925604e-05,
+      "loss": 0.5741,
+      "step": 3048
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.486681200906028e-05,
+      "loss": 1.1199,
+      "step": 3049
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.485081364380147e-05,
+      "loss": 1.1785,
+      "step": 3050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.483481189895964e-05,
+      "loss": 1.4448,
+      "step": 3051
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.481880677671098e-05,
+      "loss": 0.4935,
+      "step": 3052
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.48027982792321e-05,
+      "loss": 1.6631,
+      "step": 3053
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.478678640870007e-05,
+      "loss": 1.3515,
+      "step": 3054
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.47707711672924e-05,
+      "loss": 0.7424,
+      "step": 3055
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.475475255718711e-05,
+      "loss": 0.9785,
+      "step": 3056
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.473873058056261e-05,
+      "loss": 0.7419,
+      "step": 3057
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.472270523959787e-05,
+      "loss": 1.2519,
+      "step": 3058
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.470667653647222e-05,
+      "loss": 0.594,
+      "step": 3059
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.469064447336544e-05,
+      "loss": 1.4802,
+      "step": 3060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.467460905245786e-05,
+      "loss": 0.7422,
+      "step": 3061
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.46585702759302e-05,
+      "loss": 1.0978,
+      "step": 3062
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.464252814596367e-05,
+      "loss": 0.9142,
+      "step": 3063
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.462648266473988e-05,
+      "loss": 1.5163,
+      "step": 3064
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.461043383444098e-05,
+      "loss": 1.1713,
+      "step": 3065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.459438165724948e-05,
+      "loss": 0.7141,
+      "step": 3066
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.457832613534843e-05,
+      "loss": 1.5515,
+      "step": 3067
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.456226727092126e-05,
+      "loss": 0.8741,
+      "step": 3068
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.454620506615193e-05,
+      "loss": 1.2253,
+      "step": 3069
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.453013952322481e-05,
+      "loss": 1.9014,
+      "step": 3070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.451407064432471e-05,
+      "loss": 1.1023,
+      "step": 3071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.449799843163694e-05,
+      "loss": 1.4978,
+      "step": 3072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.448192288734724e-05,
+      "loss": 1.0389,
+      "step": 3073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.446584401364176e-05,
+      "loss": 1.0437,
+      "step": 3074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.44497618127072e-05,
+      "loss": 1.3963,
+      "step": 3075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.443367628673062e-05,
+      "loss": 1.1418,
+      "step": 3076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.441758743789958e-05,
+      "loss": 1.2399,
+      "step": 3077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.440149526840209e-05,
+      "loss": 1.383,
+      "step": 3078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.438539978042658e-05,
+      "loss": 1.1577,
+      "step": 3079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.436930097616198e-05,
+      "loss": 0.8989,
+      "step": 3080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.435319885779762e-05,
+      "loss": 1.0774,
+      "step": 3081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.433709342752333e-05,
+      "loss": 1.1472,
+      "step": 3082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.432098468752934e-05,
+      "loss": 1.3984,
+      "step": 3083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.430487264000638e-05,
+      "loss": 1.0428,
+      "step": 3084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.428875728714558e-05,
+      "loss": 0.9517,
+      "step": 3085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.427263863113858e-05,
+      "loss": 0.6176,
+      "step": 3086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.425651667417739e-05,
+      "loss": 1.3018,
+      "step": 3087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.424039141845455e-05,
+      "loss": 1.3625,
+      "step": 3088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.4224262866163e-05,
+      "loss": 1.5642,
+      "step": 3089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.420813101949613e-05,
+      "loss": 0.9007,
+      "step": 3090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.419199588064781e-05,
+      "loss": 1.0087,
+      "step": 3091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.41758574518123e-05,
+      "loss": 1.0675,
+      "step": 3092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.41597157351844e-05,
+      "loss": 0.5664,
+      "step": 3093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.414357073295925e-05,
+      "loss": 0.9356,
+      "step": 3094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.41274224473325e-05,
+      "loss": 1.3045,
+      "step": 3095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.411127088050025e-05,
+      "loss": 0.9979,
+      "step": 3096
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.409511603465901e-05,
+      "loss": 1.2835,
+      "step": 3097
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.407895791200577e-05,
+      "loss": 1.3689,
+      "step": 3098
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.406279651473795e-05,
+      "loss": 1.5225,
+      "step": 3099
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.404663184505341e-05,
+      "loss": 1.5155,
+      "step": 3100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.403046390515047e-05,
+      "loss": 0.6733,
+      "step": 3101
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.401429269722788e-05,
+      "loss": 1.2575,
+      "step": 3102
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.399811822348484e-05,
+      "loss": 1.2013,
+      "step": 3103
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.398194048612101e-05,
+      "loss": 1.3246,
+      "step": 3104
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.396575948733647e-05,
+      "loss": 1.1444,
+      "step": 3105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.394957522933173e-05,
+      "loss": 0.8138,
+      "step": 3106
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.39333877143078e-05,
+      "loss": 1.5733,
+      "step": 3107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.391719694446608e-05,
+      "loss": 0.6428,
+      "step": 3108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.390100292200843e-05,
+      "loss": 0.7273,
+      "step": 3109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.388480564913718e-05,
+      "loss": 0.3822,
+      "step": 3110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.386860512805505e-05,
+      "loss": 0.818,
+      "step": 3111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.385240136096523e-05,
+      "loss": 1.223,
+      "step": 3112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.383619435007136e-05,
+      "loss": 1.5083,
+      "step": 3113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.381998409757751e-05,
+      "loss": 1.3315,
+      "step": 3114
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.380377060568817e-05,
+      "loss": 1.4235,
+      "step": 3115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.378755387660832e-05,
+      "loss": 0.8526,
+      "step": 3116
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.377133391254332e-05,
+      "loss": 1.2794,
+      "step": 3117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.375511071569905e-05,
+      "loss": 1.1723,
+      "step": 3118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.373888428828172e-05,
+      "loss": 1.2861,
+      "step": 3119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.37226546324981e-05,
+      "loss": 0.871,
+      "step": 3120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.370642175055529e-05,
+      "loss": 1.0273,
+      "step": 3121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.369018564466092e-05,
+      "loss": 0.5651,
+      "step": 3122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.367394631702298e-05,
+      "loss": 0.7006,
+      "step": 3123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.365770376984998e-05,
+      "loss": 0.9944,
+      "step": 3124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.364145800535076e-05,
+      "loss": 1.1341,
+      "step": 3125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.362520902573472e-05,
+      "loss": 0.8941,
+      "step": 3126
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.360895683321161e-05,
+      "loss": 1.2953,
+      "step": 3127
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.359270142999165e-05,
+      "loss": 1.271,
+      "step": 3128
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.357644281828549e-05,
+      "loss": 1.0568,
+      "step": 3129
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.35601810003042e-05,
+      "loss": 0.6764,
+      "step": 3130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.354391597825934e-05,
+      "loss": 1.6355,
+      "step": 3131
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.352764775436286e-05,
+      "loss": 1.0334,
+      "step": 3132
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.351137633082713e-05,
+      "loss": 1.1994,
+      "step": 3133
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3495101709865e-05,
+      "loss": 0.6589,
+      "step": 3134
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.347882389368974e-05,
+      "loss": 0.596,
+      "step": 3135
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.346254288451503e-05,
+      "loss": 0.7933,
+      "step": 3136
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.344625868455502e-05,
+      "loss": 1.3534,
+      "step": 3137
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.342997129602427e-05,
+      "loss": 0.5008,
+      "step": 3138
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.341368072113777e-05,
+      "loss": 1.3154,
+      "step": 3139
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.339738696211098e-05,
+      "loss": 1.5557,
+      "step": 3140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.338109002115976e-05,
+      "loss": 1.2071,
+      "step": 3141
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.33647899005004e-05,
+      "loss": 1.3015,
+      "step": 3142
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.334848660234963e-05,
+      "loss": 1.1346,
+      "step": 3143
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.333218012892462e-05,
+      "loss": 0.8128,
+      "step": 3144
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.331587048244299e-05,
+      "loss": 1.3802,
+      "step": 3145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.329955766512272e-05,
+      "loss": 1.2002,
+      "step": 3146
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.328324167918232e-05,
+      "loss": 1.1954,
+      "step": 3147
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.326692252684064e-05,
+      "loss": 1.2214,
+      "step": 3148
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.325060021031703e-05,
+      "loss": 1.5004,
+      "step": 3149
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.323427473183124e-05,
+      "loss": 1.1513,
+      "step": 3150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.321794609360342e-05,
+      "loss": 0.446,
+      "step": 3151
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.32016142978542e-05,
+      "loss": 1.4123,
+      "step": 3152
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.318527934680462e-05,
+      "loss": 0.6359,
+      "step": 3153
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.316894124267615e-05,
+      "loss": 1.3647,
+      "step": 3154
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.31525999876907e-05,
+      "loss": 1.2868,
+      "step": 3155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.313625558407059e-05,
+      "loss": 0.5208,
+      "step": 3156
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.311990803403856e-05,
+      "loss": 1.1925,
+      "step": 3157
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.31035573398178e-05,
+      "loss": 0.7029,
+      "step": 3158
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.308720350363194e-05,
+      "loss": 1.5103,
+      "step": 3159
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.307084652770498e-05,
+      "loss": 1.3747,
+      "step": 3160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.305448641426143e-05,
+      "loss": 1.3747,
+      "step": 3161
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.303812316552615e-05,
+      "loss": 1.7735,
+      "step": 3162
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.302175678372445e-05,
+      "loss": 1.1933,
+      "step": 3163
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.30053872710821e-05,
+      "loss": 1.3067,
+      "step": 3164
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.298901462982525e-05,
+      "loss": 0.8967,
+      "step": 3165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.297263886218052e-05,
+      "loss": 1.3509,
+      "step": 3166
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.295625997037491e-05,
+      "loss": 1.7949,
+      "step": 3167
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.293987795663589e-05,
+      "loss": 0.6273,
+      "step": 3168
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.292349282319127e-05,
+      "loss": 0.5371,
+      "step": 3169
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.290710457226941e-05,
+      "loss": 1.169,
+      "step": 3170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.289071320609897e-05,
+      "loss": 1.1978,
+      "step": 3171
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.287431872690916e-05,
+      "loss": 0.6071,
+      "step": 3172
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.285792113692948e-05,
+      "loss": 1.0359,
+      "step": 3173
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.284152043838996e-05,
+      "loss": 0.6318,
+      "step": 3174
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.2825116633521e-05,
+      "loss": 1.1899,
+      "step": 3175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.280870972455342e-05,
+      "loss": 1.2497,
+      "step": 3176
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.279229971371849e-05,
+      "loss": 1.2189,
+      "step": 3177
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.277588660324788e-05,
+      "loss": 1.8599,
+      "step": 3178
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.27594703953737e-05,
+      "loss": 0.7316,
+      "step": 3179
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.274305109232844e-05,
+      "loss": 0.7875,
+      "step": 3180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.272662869634507e-05,
+      "loss": 1.2181,
+      "step": 3181
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.271020320965696e-05,
+      "loss": 0.6765,
+      "step": 3182
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.269377463449785e-05,
+      "loss": 0.4474,
+      "step": 3183
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.267734297310198e-05,
+      "loss": 1.4571,
+      "step": 3184
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.266090822770397e-05,
+      "loss": 0.8893,
+      "step": 3185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.264447040053884e-05,
+      "loss": 0.8344,
+      "step": 3186
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.262802949384204e-05,
+      "loss": 0.715,
+      "step": 3187
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.26115855098495e-05,
+      "loss": 1.3743,
+      "step": 3188
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.259513845079746e-05,
+      "loss": 1.236,
+      "step": 3189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.257868831892267e-05,
+      "loss": 1.1058,
+      "step": 3190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.256223511646226e-05,
+      "loss": 0.5378,
+      "step": 3191
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.254577884565377e-05,
+      "loss": 1.1931,
+      "step": 3192
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.252931950873518e-05,
+      "loss": 0.961,
+      "step": 3193
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.251285710794485e-05,
+      "loss": 1.2856,
+      "step": 3194
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.249639164552163e-05,
+      "loss": 1.0257,
+      "step": 3195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.247992312370468e-05,
+      "loss": 1.4946,
+      "step": 3196
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.246345154473368e-05,
+      "loss": 0.9685,
+      "step": 3197
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.244697691084866e-05,
+      "loss": 0.9356,
+      "step": 3198
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.243049922429009e-05,
+      "loss": 1.5299,
+      "step": 3199
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.241401848729885e-05,
+      "loss": 0.7129,
+      "step": 3200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.239753470211625e-05,
+      "loss": 1.4411,
+      "step": 3201
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.238104787098398e-05,
+      "loss": 1.4363,
+      "step": 3202
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.236455799614415e-05,
+      "loss": 1.5708,
+      "step": 3203
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.234806507983934e-05,
+      "loss": 0.6492,
+      "step": 3204
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.233156912431247e-05,
+      "loss": 1.1914,
+      "step": 3205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.231507013180691e-05,
+      "loss": 1.4534,
+      "step": 3206
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.229856810456645e-05,
+      "loss": 1.308,
+      "step": 3207
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.228206304483529e-05,
+      "loss": 1.2487,
+      "step": 3208
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2265554954858e-05,
+      "loss": 1.3497,
+      "step": 3209
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.224904383687962e-05,
+      "loss": 1.1986,
+      "step": 3210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.223252969314557e-05,
+      "loss": 1.3577,
+      "step": 3211
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.221601252590169e-05,
+      "loss": 0.5546,
+      "step": 3212
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.219949233739422e-05,
+      "loss": 1.0127,
+      "step": 3213
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.218296912986984e-05,
+      "loss": 1.1755,
+      "step": 3214
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.21664429055756e-05,
+      "loss": 1.3515,
+      "step": 3215
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.214991366675901e-05,
+      "loss": 1.307,
+      "step": 3216
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.213338141566794e-05,
+      "loss": 0.6088,
+      "step": 3217
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.211684615455068e-05,
+      "loss": 1.5647,
+      "step": 3218
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.210030788565597e-05,
+      "loss": 1.0184,
+      "step": 3219
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.208376661123292e-05,
+      "loss": 1.1976,
+      "step": 3220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.206722233353105e-05,
+      "loss": 0.718,
+      "step": 3221
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.20506750548003e-05,
+      "loss": 0.7231,
+      "step": 3222
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2034124777291e-05,
+      "loss": 1.2028,
+      "step": 3223
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.201757150325394e-05,
+      "loss": 1.0493,
+      "step": 3224
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.200101523494027e-05,
+      "loss": 1.134,
+      "step": 3225
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.198445597460152e-05,
+      "loss": 1.2711,
+      "step": 3226
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.196789372448972e-05,
+      "loss": 1.1478,
+      "step": 3227
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.195132848685718e-05,
+      "loss": 1.0014,
+      "step": 3228
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.193476026395676e-05,
+      "loss": 1.296,
+      "step": 3229
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.191818905804161e-05,
+      "loss": 1.3744,
+      "step": 3230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.190161487136536e-05,
+      "loss": 0.9796,
+      "step": 3231
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.188503770618197e-05,
+      "loss": 1.4908,
+      "step": 3232
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.18684575647459e-05,
+      "loss": 1.1112,
+      "step": 3233
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.185187444931191e-05,
+      "loss": 1.806,
+      "step": 3234
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.183528836213527e-05,
+      "loss": 1.2737,
+      "step": 3235
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.181869930547157e-05,
+      "loss": 1.6024,
+      "step": 3236
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.180210728157685e-05,
+      "loss": 0.6458,
+      "step": 3237
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.178551229270754e-05,
+      "loss": 1.3586,
+      "step": 3238
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.176891434112046e-05,
+      "loss": 1.4168,
+      "step": 3239
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.175231342907284e-05,
+      "loss": 1.3509,
+      "step": 3240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.173570955882236e-05,
+      "loss": 0.5593,
+      "step": 3241
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.171910273262703e-05,
+      "loss": 1.2463,
+      "step": 3242
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.17024929527453e-05,
+      "loss": 0.6259,
+      "step": 3243
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.168588022143602e-05,
+      "loss": 1.0148,
+      "step": 3244
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.166926454095841e-05,
+      "loss": 1.4337,
+      "step": 3245
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.165264591357215e-05,
+      "loss": 1.2693,
+      "step": 3246
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.163602434153728e-05,
+      "loss": 0.9974,
+      "step": 3247
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.161939982711423e-05,
+      "loss": 1.2787,
+      "step": 3248
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.160277237256387e-05,
+      "loss": 1.1432,
+      "step": 3249
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.158614198014744e-05,
+      "loss": 1.2263,
+      "step": 3250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.156950865212657e-05,
+      "loss": 0.5975,
+      "step": 3251
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.155287239076336e-05,
+      "loss": 0.826,
+      "step": 3252
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.153623319832019e-05,
+      "loss": 0.9029,
+      "step": 3253
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.151959107705995e-05,
+      "loss": 0.9449,
+      "step": 3254
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.150294602924586e-05,
+      "loss": 1.0366,
+      "step": 3255
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.148629805714159e-05,
+      "loss": 1.5235,
+      "step": 3256
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.146964716301113e-05,
+      "loss": 1.291,
+      "step": 3257
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.145299334911896e-05,
+      "loss": 1.1577,
+      "step": 3258
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.14363366177299e-05,
+      "loss": 1.1713,
+      "step": 3259
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.141967697110919e-05,
+      "loss": 1.6282,
+      "step": 3260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.140301441152241e-05,
+      "loss": 0.5759,
+      "step": 3261
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.138634894123563e-05,
+      "loss": 0.6241,
+      "step": 3262
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.136968056251528e-05,
+      "loss": 0.6533,
+      "step": 3263
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.135300927762811e-05,
+      "loss": 1.2192,
+      "step": 3264
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.133633508884138e-05,
+      "loss": 0.6758,
+      "step": 3265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.13196579984227e-05,
+      "loss": 1.5645,
+      "step": 3266
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.130297800864005e-05,
+      "loss": 0.9648,
+      "step": 3267
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.128629512176181e-05,
+      "loss": 1.2356,
+      "step": 3268
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.126960934005678e-05,
+      "loss": 0.5797,
+      "step": 3269
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.125292066579415e-05,
+      "loss": 0.6703,
+      "step": 3270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.123622910124346e-05,
+      "loss": 1.2406,
+      "step": 3271
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.121953464867472e-05,
+      "loss": 1.4095,
+      "step": 3272
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.120283731035828e-05,
+      "loss": 1.4976,
+      "step": 3273
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.118613708856487e-05,
+      "loss": 1.3356,
+      "step": 3274
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.116943398556566e-05,
+      "loss": 0.6738,
+      "step": 3275
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.115272800363218e-05,
+      "loss": 1.2495,
+      "step": 3276
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.113601914503634e-05,
+      "loss": 0.9958,
+      "step": 3277
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.111930741205047e-05,
+      "loss": 1.1796,
+      "step": 3278
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.110259280694729e-05,
+      "loss": 1.4889,
+      "step": 3279
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.108587533199988e-05,
+      "loss": 0.7245,
+      "step": 3280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.106915498948175e-05,
+      "loss": 1.046,
+      "step": 3281
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.105243178166676e-05,
+      "loss": 0.8051,
+      "step": 3282
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.10357057108292e-05,
+      "loss": 1.3317,
+      "step": 3283
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.101897677924374e-05,
+      "loss": 1.537,
+      "step": 3284
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.10022449891854e-05,
+      "loss": 0.8529,
+      "step": 3285
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.098551034292963e-05,
+      "loss": 1.4873,
+      "step": 3286
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.096877284275224e-05,
+      "loss": 0.6528,
+      "step": 3287
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.095203249092948e-05,
+      "loss": 0.9494,
+      "step": 3288
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.093528928973791e-05,
+      "loss": 1.3867,
+      "step": 3289
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.091854324145457e-05,
+      "loss": 1.7095,
+      "step": 3290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.09017943483568e-05,
+      "loss": 1.2567,
+      "step": 3291
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.088504261272235e-05,
+      "loss": 0.5657,
+      "step": 3292
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.08682880368294e-05,
+      "loss": 0.6163,
+      "step": 3293
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.085153062295648e-05,
+      "loss": 1.6726,
+      "step": 3294
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.083477037338251e-05,
+      "loss": 0.8764,
+      "step": 3295
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.081800729038679e-05,
+      "loss": 1.0475,
+      "step": 3296
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.080124137624903e-05,
+      "loss": 0.716,
+      "step": 3297
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.078447263324929e-05,
+      "loss": 1.3982,
+      "step": 3298
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.076770106366804e-05,
+      "loss": 1.3417,
+      "step": 3299
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.075092666978614e-05,
+      "loss": 1.3454,
+      "step": 3300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.073414945388479e-05,
+      "loss": 0.4495,
+      "step": 3301
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.071736941824564e-05,
+      "loss": 0.8526,
+      "step": 3302
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.070058656515066e-05,
+      "loss": 0.9401,
+      "step": 3303
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.068380089688226e-05,
+      "loss": 1.0118,
+      "step": 3304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.066701241572317e-05,
+      "loss": 1.3775,
+      "step": 3305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.065022112395658e-05,
+      "loss": 1.1957,
+      "step": 3306
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0633427023866e-05,
+      "loss": 1.0389,
+      "step": 3307
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.061663011773532e-05,
+      "loss": 1.1333,
+      "step": 3308
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.059983040784885e-05,
+      "loss": 1.112,
+      "step": 3309
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.058302789649127e-05,
+      "loss": 1.1652,
+      "step": 3310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.056622258594765e-05,
+      "loss": 1.2703,
+      "step": 3311
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.054941447850339e-05,
+      "loss": 1.3874,
+      "step": 3312
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.053260357644433e-05,
+      "loss": 1.1753,
+      "step": 3313
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.051578988205662e-05,
+      "loss": 1.1414,
+      "step": 3314
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.049897339762692e-05,
+      "loss": 1.3282,
+      "step": 3315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.048215412544213e-05,
+      "loss": 1.309,
+      "step": 3316
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.046533206778959e-05,
+      "loss": 0.8568,
+      "step": 3317
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.044850722695701e-05,
+      "loss": 1.2191,
+      "step": 3318
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.04316796052325e-05,
+      "loss": 0.9907,
+      "step": 3319
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.041484920490454e-05,
+      "loss": 1.4339,
+      "step": 3320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.039801602826193e-05,
+      "loss": 0.988,
+      "step": 3321
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.038118007759393e-05,
+      "loss": 1.5195,
+      "step": 3322
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.036434135519013e-05,
+      "loss": 0.5834,
+      "step": 3323
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.034749986334052e-05,
+      "loss": 1.2053,
+      "step": 3324
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.033065560433547e-05,
+      "loss": 1.369,
+      "step": 3325
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.031380858046567e-05,
+      "loss": 0.5493,
+      "step": 3326
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.029695879402228e-05,
+      "loss": 1.3783,
+      "step": 3327
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.028010624729675e-05,
+      "loss": 0.5595,
+      "step": 3328
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.026325094258095e-05,
+      "loss": 1.6087,
+      "step": 3329
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.02463928821671e-05,
+      "loss": 0.548,
+      "step": 3330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.022953206834783e-05,
+      "loss": 1.37,
+      "step": 3331
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.021266850341613e-05,
+      "loss": 0.6297,
+      "step": 3332
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.019580218966532e-05,
+      "loss": 1.2312,
+      "step": 3333
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.017893312938918e-05,
+      "loss": 1.3474,
+      "step": 3334
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.016206132488179e-05,
+      "loss": 0.5713,
+      "step": 3335
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.014518677843763e-05,
+      "loss": 1.1132,
+      "step": 3336
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.012830949235154e-05,
+      "loss": 1.5466,
+      "step": 3337
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.011142946891878e-05,
+      "loss": 1.3432,
+      "step": 3338
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.00945467104349e-05,
+      "loss": 1.168,
+      "step": 3339
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.00776612191959e-05,
+      "loss": 1.0361,
+      "step": 3340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.006077299749811e-05,
+      "loss": 0.9632,
+      "step": 3341
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.004388204763826e-05,
+      "loss": 0.9162,
+      "step": 3342
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.002698837191342e-05,
+      "loss": 1.0303,
+      "step": 3343
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.001009197262102e-05,
+      "loss": 1.0108,
+      "step": 3344
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.999319285205891e-05,
+      "loss": 1.2961,
+      "step": 3345
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.997629101252526e-05,
+      "loss": 0.8649,
+      "step": 3346
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.995938645631867e-05,
+      "loss": 1.0692,
+      "step": 3347
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.994247918573805e-05,
+      "loss": 0.9916,
+      "step": 3348
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.99255692030827e-05,
+      "loss": 0.7646,
+      "step": 3349
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.99086565106523e-05,
+      "loss": 1.5901,
+      "step": 3350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.989174111074688e-05,
+      "loss": 0.6702,
+      "step": 3351
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.987482300566686e-05,
+      "loss": 0.8874,
+      "step": 3352
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.9857902197713e-05,
+      "loss": 1.1278,
+      "step": 3353
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.984097868918644e-05,
+      "loss": 1.2255,
+      "step": 3354
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.982405248238873e-05,
+      "loss": 1.4521,
+      "step": 3355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.980712357962169e-05,
+      "loss": 1.4699,
+      "step": 3356
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.97901919831876e-05,
+      "loss": 1.0846,
+      "step": 3357
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.977325769538906e-05,
+      "loss": 1.1882,
+      "step": 3358
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.975632071852906e-05,
+      "loss": 1.4489,
+      "step": 3359
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.973938105491091e-05,
+      "loss": 1.0911,
+      "step": 3360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.972243870683836e-05,
+      "loss": 1.6976,
+      "step": 3361
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.970549367661544e-05,
+      "loss": 0.8817,
+      "step": 3362
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.968854596654663e-05,
+      "loss": 0.6651,
+      "step": 3363
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.967159557893668e-05,
+      "loss": 1.3432,
+      "step": 3364
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.965464251609081e-05,
+      "loss": 0.6291,
+      "step": 3365
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.963768678031449e-05,
+      "loss": 1.2319,
+      "step": 3366
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.962072837391367e-05,
+      "loss": 1.1898,
+      "step": 3367
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.960376729919458e-05,
+      "loss": 1.2721,
+      "step": 3368
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.958680355846384e-05,
+      "loss": 1.6333,
+      "step": 3369
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.956983715402843e-05,
+      "loss": 1.2212,
+      "step": 3370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.955286808819569e-05,
+      "loss": 1.1142,
+      "step": 3371
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.953589636327335e-05,
+      "loss": 0.8226,
+      "step": 3372
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.951892198156944e-05,
+      "loss": 1.2176,
+      "step": 3373
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.950194494539241e-05,
+      "loss": 0.7255,
+      "step": 3374
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.948496525705107e-05,
+      "loss": 0.9868,
+      "step": 3375
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.946798291885453e-05,
+      "loss": 0.5085,
+      "step": 3376
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.945099793311233e-05,
+      "loss": 0.7534,
+      "step": 3377
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.943401030213432e-05,
+      "loss": 1.3418,
+      "step": 3378
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.941702002823075e-05,
+      "loss": 1.4463,
+      "step": 3379
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.940002711371219e-05,
+      "loss": 1.6614,
+      "step": 3380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.938303156088961e-05,
+      "loss": 0.3428,
+      "step": 3381
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.93660333720743e-05,
+      "loss": 1.3481,
+      "step": 3382
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.934903254957793e-05,
+      "loss": 1.4063,
+      "step": 3383
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.933202909571254e-05,
+      "loss": 1.3402,
+      "step": 3384
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.93150230127905e-05,
+      "loss": 1.0636,
+      "step": 3385
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.929801430312455e-05,
+      "loss": 1.1145,
+      "step": 3386
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.928100296902778e-05,
+      "loss": 1.3094,
+      "step": 3387
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.926398901281365e-05,
+      "loss": 0.9258,
+      "step": 3388
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.924697243679598e-05,
+      "loss": 1.0947,
+      "step": 3389
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.922995324328893e-05,
+      "loss": 1.0551,
+      "step": 3390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.921293143460703e-05,
+      "loss": 1.3713,
+      "step": 3391
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.919590701306514e-05,
+      "loss": 0.8793,
+      "step": 3392
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.917887998097851e-05,
+      "loss": 1.5704,
+      "step": 3393
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.916185034066271e-05,
+      "loss": 1.3212,
+      "step": 3394
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.914481809443372e-05,
+      "loss": 1.4089,
+      "step": 3395
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.912778324460781e-05,
+      "loss": 0.7506,
+      "step": 3396
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.911074579350163e-05,
+      "loss": 0.6544,
+      "step": 3397
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.90937057434322e-05,
+      "loss": 1.0507,
+      "step": 3398
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.907666309671686e-05,
+      "loss": 0.6765,
+      "step": 3399
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.905961785567334e-05,
+      "loss": 1.3252,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.904257002261972e-05,
+      "loss": 1.4288,
+      "step": 3401
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.902551959987438e-05,
+      "loss": 0.819,
+      "step": 3402
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.90084665897561e-05,
+      "loss": 0.6455,
+      "step": 3403
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.899141099458402e-05,
+      "loss": 1.1489,
+      "step": 3404
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.89743528166776e-05,
+      "loss": 1.4602,
+      "step": 3405
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.895729205835668e-05,
+      "loss": 1.2331,
+      "step": 3406
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.89402287219414e-05,
+      "loss": 0.759,
+      "step": 3407
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.892316280975232e-05,
+      "loss": 0.8601,
+      "step": 3408
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.890609432411031e-05,
+      "loss": 1.1267,
+      "step": 3409
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.88890232673366e-05,
+      "loss": 1.0843,
+      "step": 3410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.887194964175274e-05,
+      "loss": 0.4706,
+      "step": 3411
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.885487344968067e-05,
+      "loss": 0.9299,
+      "step": 3412
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.883779469344268e-05,
+      "loss": 0.6947,
+      "step": 3413
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.882071337536138e-05,
+      "loss": 0.8218,
+      "step": 3414
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.880362949775973e-05,
+      "loss": 0.6349,
+      "step": 3415
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.878654306296108e-05,
+      "loss": 1.4186,
+      "step": 3416
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.876945407328908e-05,
+      "loss": 1.7237,
+      "step": 3417
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.875236253106773e-05,
+      "loss": 0.6169,
+      "step": 3418
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.873526843862141e-05,
+      "loss": 1.2525,
+      "step": 3419
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.871817179827483e-05,
+      "loss": 1.3631,
+      "step": 3420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.870107261235304e-05,
+      "loss": 0.6291,
+      "step": 3421
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.868397088318143e-05,
+      "loss": 1.0827,
+      "step": 3422
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.866686661308577e-05,
+      "loss": 1.1817,
+      "step": 3423
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.864975980439215e-05,
+      "loss": 1.254,
+      "step": 3424
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.863265045942698e-05,
+      "loss": 0.9951,
+      "step": 3425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.861553858051706e-05,
+      "loss": 1.5576,
+      "step": 3426
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.859842416998953e-05,
+      "loss": 1.1938,
+      "step": 3427
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.858130723017187e-05,
+      "loss": 1.2416,
+      "step": 3428
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.856418776339186e-05,
+      "loss": 1.3404,
+      "step": 3429
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.854706577197767e-05,
+      "loss": 1.3059,
+      "step": 3430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.852994125825783e-05,
+      "loss": 0.9771,
+      "step": 3431
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.851281422456114e-05,
+      "loss": 0.4978,
+      "step": 3432
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.849568467321684e-05,
+      "loss": 1.0671,
+      "step": 3433
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.847855260655442e-05,
+      "loss": 1.2421,
+      "step": 3434
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.846141802690379e-05,
+      "loss": 1.469,
+      "step": 3435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.844428093659512e-05,
+      "loss": 1.2017,
+      "step": 3436
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.842714133795901e-05,
+      "loss": 1.126,
+      "step": 3437
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.840999923332633e-05,
+      "loss": 1.0894,
+      "step": 3438
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.839285462502833e-05,
+      "loss": 1.2244,
+      "step": 3439
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.837570751539657e-05,
+      "loss": 1.4609,
+      "step": 3440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.8358557906763e-05,
+      "loss": 1.28,
+      "step": 3441
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.834140580145985e-05,
+      "loss": 1.5654,
+      "step": 3442
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.832425120181974e-05,
+      "loss": 1.2655,
+      "step": 3443
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.830709411017558e-05,
+      "loss": 1.266,
+      "step": 3444
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.828993452886069e-05,
+      "loss": 1.0759,
+      "step": 3445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.827277246020863e-05,
+      "loss": 0.7493,
+      "step": 3446
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.825560790655341e-05,
+      "loss": 1.5502,
+      "step": 3447
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.823844087022927e-05,
+      "loss": 1.2176,
+      "step": 3448
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.822127135357088e-05,
+      "loss": 1.1014,
+      "step": 3449
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.820409935891316e-05,
+      "loss": 1.498,
+      "step": 3450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.818692488859148e-05,
+      "loss": 1.6156,
+      "step": 3451
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.816974794494142e-05,
+      "loss": 1.2711,
+      "step": 3452
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.815256853029896e-05,
+      "loss": 1.3744,
+      "step": 3453
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.813538664700047e-05,
+      "loss": 1.1179,
+      "step": 3454
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.811820229738251e-05,
+      "loss": 0.6687,
+      "step": 3455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.810101548378215e-05,
+      "loss": 1.2543,
+      "step": 3456
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.808382620853664e-05,
+      "loss": 1.1884,
+      "step": 3457
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.80666344739837e-05,
+      "loss": 1.1864,
+      "step": 3458
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.804944028246124e-05,
+      "loss": 1.3797,
+      "step": 3459
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.803224363630764e-05,
+      "loss": 0.8043,
+      "step": 3460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.801504453786152e-05,
+      "loss": 0.797,
+      "step": 3461
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.79978429894619e-05,
+      "loss": 0.5085,
+      "step": 3462
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.798063899344809e-05,
+      "loss": 0.7195,
+      "step": 3463
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.796343255215974e-05,
+      "loss": 1.4712,
+      "step": 3464
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.794622366793682e-05,
+      "loss": 0.8345,
+      "step": 3465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.79290123431197e-05,
+      "loss": 1.3137,
+      "step": 3466
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.791179858004899e-05,
+      "loss": 1.3709,
+      "step": 3467
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.789458238106568e-05,
+      "loss": 1.2914,
+      "step": 3468
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.787736374851109e-05,
+      "loss": 1.4664,
+      "step": 3469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.786014268472689e-05,
+      "loss": 1.1494,
+      "step": 3470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7842919192055e-05,
+      "loss": 1.3915,
+      "step": 3471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.782569327283777e-05,
+      "loss": 1.8558,
+      "step": 3472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.780846492941781e-05,
+      "loss": 1.5528,
+      "step": 3473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.779123416413813e-05,
+      "loss": 0.7425,
+      "step": 3474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.777400097934198e-05,
+      "loss": 1.3532,
+      "step": 3475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7756765377373e-05,
+      "loss": 1.4894,
+      "step": 3476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.773952736057517e-05,
+      "loss": 1.3508,
+      "step": 3477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.77222869312927e-05,
+      "loss": 0.9704,
+      "step": 3478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.770504409187028e-05,
+      "loss": 1.3348,
+      "step": 3479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.76877988446528e-05,
+      "loss": 1.2205,
+      "step": 3480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.767055119198555e-05,
+      "loss": 1.0603,
+      "step": 3481
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.76533011362141e-05,
+      "loss": 0.8506,
+      "step": 3482
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.763604867968439e-05,
+      "loss": 1.244,
+      "step": 3483
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.761879382474265e-05,
+      "loss": 1.0281,
+      "step": 3484
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.760153657373545e-05,
+      "loss": 1.2552,
+      "step": 3485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.758427692900971e-05,
+      "loss": 1.1479,
+      "step": 3486
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.756701489291263e-05,
+      "loss": 0.7112,
+      "step": 3487
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.754975046779176e-05,
+      "loss": 1.0492,
+      "step": 3488
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.753248365599498e-05,
+      "loss": 1.3426,
+      "step": 3489
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.75152144598705e-05,
+      "loss": 0.5123,
+      "step": 3490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.749794288176681e-05,
+      "loss": 1.1426,
+      "step": 3491
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.748066892403278e-05,
+      "loss": 0.6936,
+      "step": 3492
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.746339258901758e-05,
+      "loss": 0.8425,
+      "step": 3493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.744611387907069e-05,
+      "loss": 1.1594,
+      "step": 3494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.742883279654192e-05,
+      "loss": 1.5117,
+      "step": 3495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.741154934378145e-05,
+      "loss": 1.5134,
+      "step": 3496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.73942635231397e-05,
+      "loss": 1.2358,
+      "step": 3497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.737697533696747e-05,
+      "loss": 1.4966,
+      "step": 3498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.735968478761586e-05,
+      "loss": 1.1464,
+      "step": 3499
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.734239187743627e-05,
+      "loss": 0.5401,
+      "step": 3500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.732509660878049e-05,
+      "loss": 1.7965,
+      "step": 3501
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.730779898400057e-05,
+      "loss": 1.2173,
+      "step": 3502
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.729049900544889e-05,
+      "loss": 1.2044,
+      "step": 3503
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.727319667547818e-05,
+      "loss": 1.0037,
+      "step": 3504
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.725589199644145e-05,
+      "loss": 1.5987,
+      "step": 3505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.723858497069204e-05,
+      "loss": 1.0229,
+      "step": 3506
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.722127560058363e-05,
+      "loss": 1.2911,
+      "step": 3507
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.720396388847022e-05,
+      "loss": 0.6163,
+      "step": 3508
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.718664983670607e-05,
+      "loss": 1.247,
+      "step": 3509
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.716933344764587e-05,
+      "loss": 1.3384,
+      "step": 3510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.715201472364449e-05,
+      "loss": 1.4348,
+      "step": 3511
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.713469366705721e-05,
+      "loss": 1.2933,
+      "step": 3512
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.711737028023963e-05,
+      "loss": 1.1776,
+      "step": 3513
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.710004456554761e-05,
+      "loss": 1.4235,
+      "step": 3514
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.708271652533739e-05,
+      "loss": 0.9318,
+      "step": 3515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.706538616196545e-05,
+      "loss": 0.7174,
+      "step": 3516
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.704805347778867e-05,
+      "loss": 1.6334,
+      "step": 3517
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.70307184751642e-05,
+      "loss": 1.2415,
+      "step": 3518
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.701338115644949e-05,
+      "loss": 1.3464,
+      "step": 3519
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.699604152400236e-05,
+      "loss": 0.8399,
+      "step": 3520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.697869958018086e-05,
+      "loss": 0.9336,
+      "step": 3521
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.696135532734346e-05,
+      "loss": 1.1022,
+      "step": 3522
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.694400876784885e-05,
+      "loss": 1.1881,
+      "step": 3523
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.69266599040561e-05,
+      "loss": 0.8706,
+      "step": 3524
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.690930873832454e-05,
+      "loss": 1.095,
+      "step": 3525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.689195527301387e-05,
+      "loss": 1.1235,
+      "step": 3526
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.687459951048406e-05,
+      "loss": 0.6926,
+      "step": 3527
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.685724145309539e-05,
+      "loss": 1.3216,
+      "step": 3528
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.683988110320849e-05,
+      "loss": 1.0405,
+      "step": 3529
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.682251846318424e-05,
+      "loss": 0.9889,
+      "step": 3530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.680515353538392e-05,
+      "loss": 1.4109,
+      "step": 3531
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.678778632216904e-05,
+      "loss": 1.2036,
+      "step": 3532
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.677041682590146e-05,
+      "loss": 1.8105,
+      "step": 3533
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.675304504894332e-05,
+      "loss": 1.5686,
+      "step": 3534
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.673567099365714e-05,
+      "loss": 1.4253,
+      "step": 3535
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.671829466240566e-05,
+      "loss": 1.3311,
+      "step": 3536
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.670091605755199e-05,
+      "loss": 1.2538,
+      "step": 3537
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.668353518145954e-05,
+      "loss": 0.5439,
+      "step": 3538
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.666615203649199e-05,
+      "loss": 1.3154,
+      "step": 3539
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.664876662501339e-05,
+      "loss": 1.0574,
+      "step": 3540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.663137894938804e-05,
+      "loss": 0.9654,
+      "step": 3541
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.661398901198061e-05,
+      "loss": 1.0602,
+      "step": 3542
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6596596815156e-05,
+      "loss": 1.2666,
+      "step": 3543
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.65792023612795e-05,
+      "loss": 1.3576,
+      "step": 3544
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.656180565271661e-05,
+      "loss": 1.2113,
+      "step": 3545
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.654440669183326e-05,
+      "loss": 1.3312,
+      "step": 3546
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.652700548099558e-05,
+      "loss": 1.2927,
+      "step": 3547
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.650960202257004e-05,
+      "loss": 0.6075,
+      "step": 3548
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.649219631892345e-05,
+      "loss": 0.9344,
+      "step": 3549
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.647478837242287e-05,
+      "loss": 1.4222,
+      "step": 3550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.645737818543572e-05,
+      "loss": 0.8514,
+      "step": 3551
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.643996576032966e-05,
+      "loss": 0.5049,
+      "step": 3552
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.64225510994727e-05,
+      "loss": 1.4248,
+      "step": 3553
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.640513420523317e-05,
+      "loss": 1.4136,
+      "step": 3554
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.638771507997965e-05,
+      "loss": 0.7795,
+      "step": 3555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.637029372608109e-05,
+      "loss": 1.4229,
+      "step": 3556
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.635287014590664e-05,
+      "loss": 1.3852,
+      "step": 3557
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.633544434182587e-05,
+      "loss": 0.7702,
+      "step": 3558
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.631801631620858e-05,
+      "loss": 1.4278,
+      "step": 3559
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.630058607142491e-05,
+      "loss": 0.4695,
+      "step": 3560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.628315360984526e-05,
+      "loss": 1.0418,
+      "step": 3561
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.626571893384035e-05,
+      "loss": 1.3721,
+      "step": 3562
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.624828204578125e-05,
+      "loss": 1.643,
+      "step": 3563
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.623084294803923e-05,
+      "loss": 1.0311,
+      "step": 3564
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.621340164298597e-05,
+      "loss": 1.7141,
+      "step": 3565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.619595813299337e-05,
+      "loss": 1.3006,
+      "step": 3566
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.617851242043366e-05,
+      "loss": 1.1954,
+      "step": 3567
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.616106450767936e-05,
+      "loss": 1.2928,
+      "step": 3568
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.614361439710331e-05,
+      "loss": 0.8372,
+      "step": 3569
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.612616209107863e-05,
+      "loss": 1.1921,
+      "step": 3570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.610870759197876e-05,
+      "loss": 1.3403,
+      "step": 3571
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.60912509021774e-05,
+      "loss": 1.2012,
+      "step": 3572
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.607379202404856e-05,
+      "loss": 1.4994,
+      "step": 3573
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.605633095996659e-05,
+      "loss": 1.2144,
+      "step": 3574
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.603886771230608e-05,
+      "loss": 0.4834,
+      "step": 3575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.602140228344197e-05,
+      "loss": 0.606,
+      "step": 3576
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.600393467574943e-05,
+      "loss": 1.1585,
+      "step": 3577
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.5986464891604e-05,
+      "loss": 1.3672,
+      "step": 3578
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.596899293338147e-05,
+      "loss": 0.939,
+      "step": 3579
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.595151880345792e-05,
+      "loss": 0.5547,
+      "step": 3580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.593404250420977e-05,
+      "loss": 1.2162,
+      "step": 3581
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.591656403801368e-05,
+      "loss": 1.424,
+      "step": 3582
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.589908340724666e-05,
+      "loss": 1.0355,
+      "step": 3583
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.588160061428594e-05,
+      "loss": 1.5091,
+      "step": 3584
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.586411566150914e-05,
+      "loss": 1.3427,
+      "step": 3585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.584662855129411e-05,
+      "loss": 1.4456,
+      "step": 3586
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.5829139286019e-05,
+      "loss": 0.4602,
+      "step": 3587
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.581164786806224e-05,
+      "loss": 0.6329,
+      "step": 3588
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.57941542998026e-05,
+      "loss": 0.8481,
+      "step": 3589
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.577665858361913e-05,
+      "loss": 1.3006,
+      "step": 3590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.575916072189111e-05,
+      "loss": 1.0379,
+      "step": 3591
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.574166071699821e-05,
+      "loss": 1.3764,
+      "step": 3592
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.572415857132031e-05,
+      "loss": 1.2363,
+      "step": 3593
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.570665428723761e-05,
+      "loss": 1.041,
+      "step": 3594
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.568914786713061e-05,
+      "loss": 0.9034,
+      "step": 3595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.56716393133801e-05,
+      "loss": 1.1339,
+      "step": 3596
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.565412862836715e-05,
+      "loss": 1.3317,
+      "step": 3597
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.563661581447311e-05,
+      "loss": 1.1422,
+      "step": 3598
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.561910087407965e-05,
+      "loss": 1.2712,
+      "step": 3599
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.56015838095687e-05,
+      "loss": 1.2097,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.558406462332252e-05,
+      "loss": 1.3927,
+      "step": 3601
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.556654331772359e-05,
+      "loss": 0.621,
+      "step": 3602
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.554901989515474e-05,
+      "loss": 0.9147,
+      "step": 3603
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.553149435799906e-05,
+      "loss": 1.2028,
+      "step": 3604
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.551396670863992e-05,
+      "loss": 1.2275,
+      "step": 3605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.5496436949461e-05,
+      "loss": 1.1411,
+      "step": 3606
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.547890508284627e-05,
+      "loss": 0.9327,
+      "step": 3607
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.546137111117996e-05,
+      "loss": 1.1795,
+      "step": 3608
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.544383503684661e-05,
+      "loss": 1.5794,
+      "step": 3609
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.542629686223103e-05,
+      "loss": 1.1093,
+      "step": 3610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.54087565897183e-05,
+      "loss": 1.1836,
+      "step": 3611
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.539121422169385e-05,
+      "loss": 0.5287,
+      "step": 3612
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.537366976054333e-05,
+      "loss": 1.2109,
+      "step": 3613
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.535612320865267e-05,
+      "loss": 1.6292,
+      "step": 3614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.533857456840817e-05,
+      "loss": 1.1356,
+      "step": 3615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.53210238421963e-05,
+      "loss": 1.6447,
+      "step": 3616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.530347103240391e-05,
+      "loss": 1.0916,
+      "step": 3617
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.528591614141805e-05,
+      "loss": 1.1844,
+      "step": 3618
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.526835917162614e-05,
+      "loss": 0.8847,
+      "step": 3619
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.525080012541582e-05,
+      "loss": 0.8077,
+      "step": 3620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.523323900517502e-05,
+      "loss": 1.2381,
+      "step": 3621
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.521567581329196e-05,
+      "loss": 1.0836,
+      "step": 3622
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.519811055215514e-05,
+      "loss": 1.1582,
+      "step": 3623
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.518054322415337e-05,
+      "loss": 0.8779,
+      "step": 3624
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.51629738316757e-05,
+      "loss": 1.3921,
+      "step": 3625
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.514540237711149e-05,
+      "loss": 1.2732,
+      "step": 3626
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.512782886285036e-05,
+      "loss": 1.3758,
+      "step": 3627
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.51102532912822e-05,
+      "loss": 1.061,
+      "step": 3628
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.509267566479722e-05,
+      "loss": 1.381,
+      "step": 3629
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.507509598578588e-05,
+      "loss": 0.6391,
+      "step": 3630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.505751425663892e-05,
+      "loss": 1.1059,
+      "step": 3631
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.50399304797474e-05,
+      "loss": 0.7963,
+      "step": 3632
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.502234465750256e-05,
+      "loss": 1.3754,
+      "step": 3633
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.500475679229603e-05,
+      "loss": 1.2565,
+      "step": 3634
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.498716688651965e-05,
+      "loss": 0.9277,
+      "step": 3635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.496957494256557e-05,
+      "loss": 0.9306,
+      "step": 3636
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.495198096282618e-05,
+      "loss": 0.6162,
+      "step": 3637
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.493438494969422e-05,
+      "loss": 1.2088,
+      "step": 3638
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.491678690556259e-05,
+      "loss": 1.292,
+      "step": 3639
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.489918683282459e-05,
+      "loss": 0.8741,
+      "step": 3640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.488158473387368e-05,
+      "loss": 0.9259,
+      "step": 3641
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.486398061110374e-05,
+      "loss": 1.2719,
+      "step": 3642
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.484637446690877e-05,
+      "loss": 0.8021,
+      "step": 3643
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.482876630368313e-05,
+      "loss": 1.3335,
+      "step": 3644
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.481115612382147e-05,
+      "loss": 0.9545,
+      "step": 3645
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.479354392971864e-05,
+      "loss": 0.965,
+      "step": 3646
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.477592972376984e-05,
+      "loss": 1.5416,
+      "step": 3647
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.475831350837049e-05,
+      "loss": 1.1605,
+      "step": 3648
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.474069528591635e-05,
+      "loss": 0.9921,
+      "step": 3649
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.472307505880333e-05,
+      "loss": 0.6017,
+      "step": 3650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.470545282942778e-05,
+      "loss": 1.2325,
+      "step": 3651
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.468782860018614e-05,
+      "loss": 1.1896,
+      "step": 3652
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.467020237347528e-05,
+      "loss": 0.703,
+      "step": 3653
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.465257415169226e-05,
+      "loss": 1.5269,
+      "step": 3654
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.463494393723443e-05,
+      "loss": 1.0448,
+      "step": 3655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.46173117324994e-05,
+      "loss": 1.3122,
+      "step": 3656
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.459967753988505e-05,
+      "loss": 1.4644,
+      "step": 3657
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.458204136178958e-05,
+      "loss": 1.1557,
+      "step": 3658
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.456440320061137e-05,
+      "loss": 0.6097,
+      "step": 3659
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.454676305874914e-05,
+      "loss": 1.2199,
+      "step": 3660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.452912093860186e-05,
+      "loss": 1.3366,
+      "step": 3661
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.451147684256877e-05,
+      "loss": 1.65,
+      "step": 3662
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.44938307730494e-05,
+      "loss": 0.9324,
+      "step": 3663
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.447618273244347e-05,
+      "loss": 1.4047,
+      "step": 3664
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.445853272315106e-05,
+      "loss": 1.4457,
+      "step": 3665
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.444088074757246e-05,
+      "loss": 1.3154,
+      "step": 3666
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.442322680810828e-05,
+      "loss": 0.9484,
+      "step": 3667
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.440557090715933e-05,
+      "loss": 1.4584,
+      "step": 3668
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.438791304712676e-05,
+      "loss": 1.3992,
+      "step": 3669
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.437025323041192e-05,
+      "loss": 0.6416,
+      "step": 3670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.435259145941645e-05,
+      "loss": 1.4349,
+      "step": 3671
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.433492773654228e-05,
+      "loss": 1.2064,
+      "step": 3672
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.431726206419156e-05,
+      "loss": 0.485,
+      "step": 3673
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.429959444476677e-05,
+      "loss": 1.3742,
+      "step": 3674
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.428192488067055e-05,
+      "loss": 0.8781,
+      "step": 3675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.426425337430595e-05,
+      "loss": 0.6159,
+      "step": 3676
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.424657992807616e-05,
+      "loss": 1.1675,
+      "step": 3677
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.422890454438469e-05,
+      "loss": 0.9867,
+      "step": 3678
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.421122722563528e-05,
+      "loss": 1.4736,
+      "step": 3679
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.419354797423198e-05,
+      "loss": 1.2836,
+      "step": 3680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.417586679257905e-05,
+      "loss": 1.3555,
+      "step": 3681
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.415818368308107e-05,
+      "loss": 1.0547,
+      "step": 3682
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.414049864814283e-05,
+      "loss": 0.7915,
+      "step": 3683
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.412281169016943e-05,
+      "loss": 1.335,
+      "step": 3684
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.410512281156617e-05,
+      "loss": 0.979,
+      "step": 3685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.408743201473868e-05,
+      "loss": 1.2407,
+      "step": 3686
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.40697393020928e-05,
+      "loss": 0.8149,
+      "step": 3687
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.405204467603464e-05,
+      "loss": 0.8671,
+      "step": 3688
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.40343481389706e-05,
+      "loss": 1.4423,
+      "step": 3689
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.401664969330731e-05,
+      "loss": 0.5336,
+      "step": 3690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.399894934145165e-05,
+      "loss": 0.9231,
+      "step": 3691
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.39812470858108e-05,
+      "loss": 1.466,
+      "step": 3692
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.396354292879217e-05,
+      "loss": 0.904,
+      "step": 3693
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.394583687280344e-05,
+      "loss": 1.2309,
+      "step": 3694
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.392812892025253e-05,
+      "loss": 1.5439,
+      "step": 3695
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.391041907354762e-05,
+      "loss": 0.9272,
+      "step": 3696
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.38927073350972e-05,
+      "loss": 1.0222,
+      "step": 3697
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.387499370730991e-05,
+      "loss": 1.2334,
+      "step": 3698
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.385727819259478e-05,
+      "loss": 1.2959,
+      "step": 3699
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.383956079336099e-05,
+      "loss": 0.8451,
+      "step": 3700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.382184151201804e-05,
+      "loss": 1.0824,
+      "step": 3701
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.380412035097562e-05,
+      "loss": 0.4379,
+      "step": 3702
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.378639731264377e-05,
+      "loss": 1.2988,
+      "step": 3703
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.37686723994327e-05,
+      "loss": 1.0147,
+      "step": 3704
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.37509456137529e-05,
+      "loss": 0.7997,
+      "step": 3705
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.373321695801515e-05,
+      "loss": 1.1747,
+      "step": 3706
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.37154864346304e-05,
+      "loss": 0.7153,
+      "step": 3707
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.369775404601001e-05,
+      "loss": 1.2522,
+      "step": 3708
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.368001979456539e-05,
+      "loss": 1.0493,
+      "step": 3709
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.366228368270837e-05,
+      "loss": 0.8771,
+      "step": 3710
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.364454571285097e-05,
+      "loss": 0.7026,
+      "step": 3711
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.362680588740541e-05,
+      "loss": 1.2735,
+      "step": 3712
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.360906420878428e-05,
+      "loss": 1.377,
+      "step": 3713
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.35913206794003e-05,
+      "loss": 1.1269,
+      "step": 3714
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.357357530166654e-05,
+      "loss": 0.9695,
+      "step": 3715
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.355582807799627e-05,
+      "loss": 1.3754,
+      "step": 3716
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.353807901080301e-05,
+      "loss": 1.3155,
+      "step": 3717
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.352032810250055e-05,
+      "loss": 1.0808,
+      "step": 3718
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.350257535550292e-05,
+      "loss": 1.0456,
+      "step": 3719
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.34848207722244e-05,
+      "loss": 1.2476,
+      "step": 3720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.346706435507954e-05,
+      "loss": 0.6419,
+      "step": 3721
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.34493061064831e-05,
+      "loss": 1.3022,
+      "step": 3722
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.343154602885011e-05,
+      "loss": 1.4833,
+      "step": 3723
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.341378412459589e-05,
+      "loss": 1.1403,
+      "step": 3724
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.339602039613589e-05,
+      "loss": 1.2167,
+      "step": 3725
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.337825484588596e-05,
+      "loss": 0.3692,
+      "step": 3726
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.336048747626209e-05,
+      "loss": 1.1158,
+      "step": 3727
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.334271828968055e-05,
+      "loss": 0.888,
+      "step": 3728
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.332494728855786e-05,
+      "loss": 0.6367,
+      "step": 3729
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.33071744753108e-05,
+      "loss": 1.1193,
+      "step": 3730
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.328939985235637e-05,
+      "loss": 1.374,
+      "step": 3731
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.32716234221118e-05,
+      "loss": 0.8917,
+      "step": 3732
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.325384518699466e-05,
+      "loss": 0.8728,
+      "step": 3733
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.323606514942263e-05,
+      "loss": 1.3755,
+      "step": 3734
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.321828331181375e-05,
+      "loss": 0.9967,
+      "step": 3735
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.320049967658623e-05,
+      "loss": 1.1405,
+      "step": 3736
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.318271424615857e-05,
+      "loss": 1.0068,
+      "step": 3737
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.31649270229495e-05,
+      "loss": 0.5779,
+      "step": 3738
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.314713800937797e-05,
+      "loss": 0.5206,
+      "step": 3739
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.312934720786321e-05,
+      "loss": 1.4471,
+      "step": 3740
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.311155462082468e-05,
+      "loss": 1.42,
+      "step": 3741
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.309376025068207e-05,
+      "loss": 0.8607,
+      "step": 3742
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.307596409985532e-05,
+      "loss": 1.4657,
+      "step": 3743
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.305816617076463e-05,
+      "loss": 0.6801,
+      "step": 3744
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.304036646583043e-05,
+      "loss": 1.5747,
+      "step": 3745
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.302256498747336e-05,
+      "loss": 0.9879,
+      "step": 3746
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.300476173811438e-05,
+      "loss": 1.0335,
+      "step": 3747
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.298695672017457e-05,
+      "loss": 1.6611,
+      "step": 3748
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.296914993607538e-05,
+      "loss": 1.1357,
+      "step": 3749
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.295134138823841e-05,
+      "loss": 0.7943,
+      "step": 3750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.293353107908555e-05,
+      "loss": 1.2654,
+      "step": 3751
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.29157190110389e-05,
+      "loss": 1.5126,
+      "step": 3752
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.289790518652082e-05,
+      "loss": 0.9674,
+      "step": 3753
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.288008960795389e-05,
+      "loss": 0.8517,
+      "step": 3754
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.286227227776092e-05,
+      "loss": 1.0875,
+      "step": 3755
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.284445319836502e-05,
+      "loss": 1.3951,
+      "step": 3756
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.282663237218944e-05,
+      "loss": 0.866,
+      "step": 3757
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.280880980165778e-05,
+      "loss": 1.501,
+      "step": 3758
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.279098548919375e-05,
+      "loss": 1.0551,
+      "step": 3759
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.277315943722142e-05,
+      "loss": 1.3576,
+      "step": 3760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.275533164816502e-05,
+      "loss": 0.4911,
+      "step": 3761
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.273750212444904e-05,
+      "loss": 0.9511,
+      "step": 3762
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.27196708684982e-05,
+      "loss": 1.1877,
+      "step": 3763
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.270183788273746e-05,
+      "loss": 1.1181,
+      "step": 3764
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.268400316959201e-05,
+      "loss": 1.5606,
+      "step": 3765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.266616673148731e-05,
+      "loss": 0.9748,
+      "step": 3766
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.264832857084898e-05,
+      "loss": 1.2936,
+      "step": 3767
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.263048869010294e-05,
+      "loss": 0.7833,
+      "step": 3768
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.261264709167533e-05,
+      "loss": 1.0552,
+      "step": 3769
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.259480377799249e-05,
+      "loss": 1.2338,
+      "step": 3770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.257695875148106e-05,
+      "loss": 1.081,
+      "step": 3771
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.255911201456782e-05,
+      "loss": 1.1272,
+      "step": 3772
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.254126356967987e-05,
+      "loss": 0.5743,
+      "step": 3773
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.25234134192445e-05,
+      "loss": 1.1591,
+      "step": 3774
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.250556156568925e-05,
+      "loss": 0.5218,
+      "step": 3775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.248770801144186e-05,
+      "loss": 0.6589,
+      "step": 3776
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.246985275893032e-05,
+      "loss": 0.7325,
+      "step": 3777
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.245199581058288e-05,
+      "loss": 1.2181,
+      "step": 3778
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.243413716882796e-05,
+      "loss": 0.9881,
+      "step": 3779
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.241627683609426e-05,
+      "loss": 0.8056,
+      "step": 3780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.23984148148107e-05,
+      "loss": 1.6322,
+      "step": 3781
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.238055110740641e-05,
+      "loss": 1.1037,
+      "step": 3782
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.236268571631077e-05,
+      "loss": 0.799,
+      "step": 3783
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.234481864395338e-05,
+      "loss": 1.3489,
+      "step": 3784
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.232694989276407e-05,
+      "loss": 1.1102,
+      "step": 3785
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.230907946517289e-05,
+      "loss": 0.4686,
+      "step": 3786
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.229120736361015e-05,
+      "loss": 0.8684,
+      "step": 3787
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.227333359050634e-05,
+      "loss": 0.678,
+      "step": 3788
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.22554581482922e-05,
+      "loss": 1.3516,
+      "step": 3789
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.223758103939872e-05,
+      "loss": 1.3109,
+      "step": 3790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.221970226625708e-05,
+      "loss": 0.853,
+      "step": 3791
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.220182183129869e-05,
+      "loss": 1.5319,
+      "step": 3792
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.218393973695521e-05,
+      "loss": 1.2123,
+      "step": 3793
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.216605598565855e-05,
+      "loss": 0.7199,
+      "step": 3794
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.214817057984074e-05,
+      "loss": 0.8872,
+      "step": 3795
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.213028352193414e-05,
+      "loss": 1.2939,
+      "step": 3796
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.21123948143713e-05,
+      "loss": 1.1907,
+      "step": 3797
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.209450445958498e-05,
+      "loss": 0.722,
+      "step": 3798
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.207661246000819e-05,
+      "loss": 1.3425,
+      "step": 3799
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.205871881807415e-05,
+      "loss": 1.1343,
+      "step": 3800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.204082353621628e-05,
+      "loss": 0.7346,
+      "step": 3801
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.202292661686826e-05,
+      "loss": 1.7295,
+      "step": 3802
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2005028062464e-05,
+      "loss": 0.5046,
+      "step": 3803
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.198712787543758e-05,
+      "loss": 1.1777,
+      "step": 3804
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.196922605822334e-05,
+      "loss": 1.4533,
+      "step": 3805
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.195132261325587e-05,
+      "loss": 1.3102,
+      "step": 3806
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.193341754296989e-05,
+      "loss": 1.0539,
+      "step": 3807
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.191551084980044e-05,
+      "loss": 0.9651,
+      "step": 3808
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.189760253618271e-05,
+      "loss": 1.0515,
+      "step": 3809
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.187969260455217e-05,
+      "loss": 0.9832,
+      "step": 3810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.186178105734443e-05,
+      "loss": 0.8654,
+      "step": 3811
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.184386789699544e-05,
+      "loss": 0.7107,
+      "step": 3812
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.182595312594123e-05,
+      "loss": 1.0658,
+      "step": 3813
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.180803674661816e-05,
+      "loss": 1.2203,
+      "step": 3814
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.179011876146274e-05,
+      "loss": 1.115,
+      "step": 3815
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.177219917291174e-05,
+      "loss": 0.8916,
+      "step": 3816
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.175427798340213e-05,
+      "loss": 1.4689,
+      "step": 3817
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.173635519537107e-05,
+      "loss": 1.072,
+      "step": 3818
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.171843081125603e-05,
+      "loss": 0.7158,
+      "step": 3819
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.170050483349458e-05,
+      "loss": 1.1348,
+      "step": 3820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.168257726452458e-05,
+      "loss": 1.2233,
+      "step": 3821
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.166464810678407e-05,
+      "loss": 0.8212,
+      "step": 3822
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.164671736271136e-05,
+      "loss": 1.3011,
+      "step": 3823
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.16287850347449e-05,
+      "loss": 1.5057,
+      "step": 3824
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.161085112532342e-05,
+      "loss": 0.9132,
+      "step": 3825
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.159291563688583e-05,
+      "loss": 1.3502,
+      "step": 3826
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.157497857187127e-05,
+      "loss": 0.3783,
+      "step": 3827
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.155703993271909e-05,
+      "loss": 1.2123,
+      "step": 3828
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.153909972186883e-05,
+      "loss": 0.5588,
+      "step": 3829
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.152115794176029e-05,
+      "loss": 0.8917,
+      "step": 3830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.150321459483346e-05,
+      "loss": 1.6774,
+      "step": 3831
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.148526968352852e-05,
+      "loss": 1.4163,
+      "step": 3832
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.146732321028593e-05,
+      "loss": 1.0836,
+      "step": 3833
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.144937517754627e-05,
+      "loss": 0.5588,
+      "step": 3834
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.14314255877504e-05,
+      "loss": 0.6963,
+      "step": 3835
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.141347444333938e-05,
+      "loss": 1.4544,
+      "step": 3836
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.139552174675447e-05,
+      "loss": 0.9208,
+      "step": 3837
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.137756750043714e-05,
+      "loss": 1.5127,
+      "step": 3838
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.135961170682906e-05,
+      "loss": 1.5118,
+      "step": 3839
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.134165436837217e-05,
+      "loss": 1.3266,
+      "step": 3840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.132369548750851e-05,
+      "loss": 0.8124,
+      "step": 3841
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.130573506668047e-05,
+      "loss": 1.0073,
+      "step": 3842
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.128777310833051e-05,
+      "loss": 0.5491,
+      "step": 3843
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.126980961490142e-05,
+      "loss": 0.6395,
+      "step": 3844
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.125184458883611e-05,
+      "loss": 0.4672,
+      "step": 3845
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.123387803257772e-05,
+      "loss": 1.2378,
+      "step": 3846
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.121590994856962e-05,
+      "loss": 1.4917,
+      "step": 3847
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.119794033925541e-05,
+      "loss": 1.1245,
+      "step": 3848
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.117996920707882e-05,
+      "loss": 0.9675,
+      "step": 3849
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.116199655448385e-05,
+      "loss": 1.0927,
+      "step": 3850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.114402238391469e-05,
+      "loss": 0.9914,
+      "step": 3851
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.112604669781572e-05,
+      "loss": 0.8532,
+      "step": 3852
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.110806949863157e-05,
+      "loss": 1.3881,
+      "step": 3853
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.109009078880701e-05,
+      "loss": 1.2725,
+      "step": 3854
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.10721105707871e-05,
+      "loss": 0.6957,
+      "step": 3855
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.105412884701701e-05,
+      "loss": 1.3775,
+      "step": 3856
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.103614561994219e-05,
+      "loss": 1.5023,
+      "step": 3857
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.101816089200826e-05,
+      "loss": 1.3779,
+      "step": 3858
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1000174665661036e-05,
+      "loss": 1.3241,
+      "step": 3859
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.098218694334657e-05,
+      "loss": 1.5441,
+      "step": 3860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0964197727511096e-05,
+      "loss": 0.5974,
+      "step": 3861
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0946207020601056e-05,
+      "loss": 0.7375,
+      "step": 3862
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0928214825063104e-05,
+      "loss": 0.8657,
+      "step": 3863
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0910221143344045e-05,
+      "loss": 1.0289,
+      "step": 3864
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.089222597789098e-05,
+      "loss": 1.2397,
+      "step": 3865
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.087422933115112e-05,
+      "loss": 1.3073,
+      "step": 3866
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.085623120557195e-05,
+      "loss": 0.776,
+      "step": 3867
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.083823160360108e-05,
+      "loss": 0.8506,
+      "step": 3868
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.082023052768641e-05,
+      "loss": 1.4772,
+      "step": 3869
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.080222798027596e-05,
+      "loss": 0.6135,
+      "step": 3870
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.078422396381799e-05,
+      "loss": 1.3013,
+      "step": 3871
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.076621848076096e-05,
+      "loss": 1.2285,
+      "step": 3872
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0748211533553533e-05,
+      "loss": 1.1394,
+      "step": 3873
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.073020312464455e-05,
+      "loss": 1.3554,
+      "step": 3874
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.071219325648305e-05,
+      "loss": 1.0402,
+      "step": 3875
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0694181931518304e-05,
+      "loss": 1.4511,
+      "step": 3876
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.067616915219975e-05,
+      "loss": 0.7074,
+      "step": 3877
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.065815492097703e-05,
+      "loss": 1.2223,
+      "step": 3878
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.06401392403e-05,
+      "loss": 1.377,
+      "step": 3879
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0622122112618686e-05,
+      "loss": 1.2843,
+      "step": 3880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.060410354038333e-05,
+      "loss": 1.0015,
+      "step": 3881
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.058608352604437e-05,
+      "loss": 1.3163,
+      "step": 3882
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.056806207205242e-05,
+      "loss": 0.7669,
+      "step": 3883
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.055003918085832e-05,
+      "loss": 0.5657,
+      "step": 3884
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0532014854913085e-05,
+      "loss": 1.2269,
+      "step": 3885
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.051398909666795e-05,
+      "loss": 1.0828,
+      "step": 3886
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0495961908574294e-05,
+      "loss": 1.2983,
+      "step": 3887
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.047793329308374e-05,
+      "loss": 1.0397,
+      "step": 3888
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0459903252648077e-05,
+      "loss": 1.3078,
+      "step": 3889
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0441871789719305e-05,
+      "loss": 0.4687,
+      "step": 3890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.042383890674961e-05,
+      "loss": 1.7211,
+      "step": 3891
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0405804606191375e-05,
+      "loss": 1.0595,
+      "step": 3892
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.038776889049715e-05,
+      "loss": 0.686,
+      "step": 3893
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.036973176211974e-05,
+      "loss": 1.2904,
+      "step": 3894
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0351693223512074e-05,
+      "loss": 0.7302,
+      "step": 3895
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.03336532771273e-05,
+      "loss": 0.9681,
+      "step": 3896
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.031561192541877e-05,
+      "loss": 1.2722,
+      "step": 3897
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.029756917084e-05,
+      "loss": 0.8191,
+      "step": 3898
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0279525015844727e-05,
+      "loss": 1.1256,
+      "step": 3899
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.026147946288685e-05,
+      "loss": 1.2974,
+      "step": 3900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.02434325144205e-05,
+      "loss": 0.688,
+      "step": 3901
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.022538417289991e-05,
+      "loss": 1.8695,
+      "step": 3902
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.020733444077964e-05,
+      "loss": 1.2189,
+      "step": 3903
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.01892833205143e-05,
+      "loss": 0.8389,
+      "step": 3904
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0171230814558774e-05,
+      "loss": 1.1583,
+      "step": 3905
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.015317692536811e-05,
+      "loss": 1.1843,
+      "step": 3906
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.013512165539754e-05,
+      "loss": 1.3255,
+      "step": 3907
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.011706500710249e-05,
+      "loss": 0.6558,
+      "step": 3908
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.009900698293857e-05,
+      "loss": 1.126,
+      "step": 3909
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.008094758536157e-05,
+      "loss": 1.3328,
+      "step": 3910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0062886816827494e-05,
+      "loss": 1.7767,
+      "step": 3911
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.004482467979249e-05,
+      "loss": 1.3444,
+      "step": 3912
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.002676117671294e-05,
+      "loss": 0.508,
+      "step": 3913
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0008696310045375e-05,
+      "loss": 1.162,
+      "step": 3914
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9990630082246537e-05,
+      "loss": 1.1389,
+      "step": 3915
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.997256249577331e-05,
+      "loss": 1.0358,
+      "step": 3916
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9954493553082825e-05,
+      "loss": 1.2939,
+      "step": 3917
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.993642325663235e-05,
+      "loss": 0.9399,
+      "step": 3918
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.991835160887934e-05,
+      "loss": 1.0009,
+      "step": 3919
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.990027861228148e-05,
+      "loss": 0.9911,
+      "step": 3920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.988220426929658e-05,
+      "loss": 1.3792,
+      "step": 3921
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.986412858238267e-05,
+      "loss": 1.4836,
+      "step": 3922
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.984605155399794e-05,
+      "loss": 0.7654,
+      "step": 3923
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9827973186600786e-05,
+      "loss": 1.2994,
+      "step": 3924
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.980989348264974e-05,
+      "loss": 1.2951,
+      "step": 3925
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.97918124446036e-05,
+      "loss": 1.1745,
+      "step": 3926
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.977373007492124e-05,
+      "loss": 1.4431,
+      "step": 3927
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.975564637606183e-05,
+      "loss": 1.3017,
+      "step": 3928
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.973756135048462e-05,
+      "loss": 1.1972,
+      "step": 3929
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.971947500064907e-05,
+      "loss": 0.8688,
+      "step": 3930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.970138732901486e-05,
+      "loss": 0.8785,
+      "step": 3931
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.96832983380418e-05,
+      "loss": 1.0436,
+      "step": 3932
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.966520803018992e-05,
+      "loss": 0.6154,
+      "step": 3933
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.964711640791939e-05,
+      "loss": 1.3387,
+      "step": 3934
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.962902347369059e-05,
+      "loss": 1.2209,
+      "step": 3935
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9610929229964066e-05,
+      "loss": 1.067,
+      "step": 3936
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9592833679200524e-05,
+      "loss": 0.7081,
+      "step": 3937
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9574736823860886e-05,
+      "loss": 1.6532,
+      "step": 3938
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.955663866640623e-05,
+      "loss": 0.9858,
+      "step": 3939
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9538539209297815e-05,
+      "loss": 1.4106,
+      "step": 3940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.952043845499704e-05,
+      "loss": 1.0863,
+      "step": 3941
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9502336405965564e-05,
+      "loss": 0.4896,
+      "step": 3942
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.948423306466514e-05,
+      "loss": 1.0402,
+      "step": 3943
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9466128433557744e-05,
+      "loss": 1.3765,
+      "step": 3944
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.944802251510551e-05,
+      "loss": 1.0916,
+      "step": 3945
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9429915311770737e-05,
+      "loss": 1.35,
+      "step": 3946
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9411806826015934e-05,
+      "loss": 0.8887,
+      "step": 3947
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9393697060303744e-05,
+      "loss": 1.2059,
+      "step": 3948
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.937558601709702e-05,
+      "loss": 1.3266,
+      "step": 3949
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.935747369885873e-05,
+      "loss": 1.0477,
+      "step": 3950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9339360108052124e-05,
+      "loss": 1.0769,
+      "step": 3951
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.93212452471405e-05,
+      "loss": 1.2738,
+      "step": 3952
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.930312911858742e-05,
+      "loss": 1.4217,
+      "step": 3953
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.928501172485655e-05,
+      "loss": 1.2339,
+      "step": 3954
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.926689306841179e-05,
+      "loss": 1.5005,
+      "step": 3955
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9248773151717174e-05,
+      "loss": 1.1265,
+      "step": 3956
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.923065197723693e-05,
+      "loss": 1.2575,
+      "step": 3957
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9212529547435424e-05,
+      "loss": 0.9943,
+      "step": 3958
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.919440586477723e-05,
+      "loss": 1.4348,
+      "step": 3959
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.917628093172706e-05,
+      "loss": 0.7179,
+      "step": 3960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.915815475074983e-05,
+      "loss": 1.4383,
+      "step": 3961
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.91400273243106e-05,
+      "loss": 0.5699,
+      "step": 3962
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9121898654874594e-05,
+      "loss": 1.257,
+      "step": 3963
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.910376874490724e-05,
+      "loss": 1.3107,
+      "step": 3964
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.908563759687411e-05,
+      "loss": 0.5199,
+      "step": 3965
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.906750521324093e-05,
+      "loss": 1.1364,
+      "step": 3966
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.904937159647361e-05,
+      "loss": 1.1496,
+      "step": 3967
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.903123674903825e-05,
+      "loss": 1.0756,
+      "step": 3968
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.901310067340109e-05,
+      "loss": 0.7485,
+      "step": 3969
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.899496337202853e-05,
+      "loss": 1.3732,
+      "step": 3970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.897682484738717e-05,
+      "loss": 1.452,
+      "step": 3971
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.895868510194374e-05,
+      "loss": 1.4047,
+      "step": 3972
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8940544138165144e-05,
+      "loss": 1.1905,
+      "step": 3973
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8922401958518495e-05,
+      "loss": 0.7412,
+      "step": 3974
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.890425856547099e-05,
+      "loss": 0.8673,
+      "step": 3975
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8886113961490084e-05,
+      "loss": 1.0652,
+      "step": 3976
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.886796814904329e-05,
+      "loss": 1.4678,
+      "step": 3977
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8849821130598424e-05,
+      "loss": 0.7634,
+      "step": 3978
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8831672908623326e-05,
+      "loss": 1.4137,
+      "step": 3979
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8813523485586084e-05,
+      "loss": 1.281,
+      "step": 3980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8795372863954924e-05,
+      "loss": 0.8537,
+      "step": 3981
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.877722104619823e-05,
+      "loss": 1.0265,
+      "step": 3982
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.875906803478457e-05,
+      "loss": 1.0849,
+      "step": 3983
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.874091383218263e-05,
+      "loss": 0.695,
+      "step": 3984
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.872275844086135e-05,
+      "loss": 1.35,
+      "step": 3985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.87046018632897e-05,
+      "loss": 0.5217,
+      "step": 3986
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.868644410193692e-05,
+      "loss": 1.4349,
+      "step": 3987
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.866828515927236e-05,
+      "loss": 1.2855,
+      "step": 3988
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.865012503776556e-05,
+      "loss": 0.6102,
+      "step": 3989
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8631963739886175e-05,
+      "loss": 1.0841,
+      "step": 3990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8613801268104075e-05,
+      "loss": 1.3404,
+      "step": 3991
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8595637624889246e-05,
+      "loss": 1.2858,
+      "step": 3992
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.857747281271185e-05,
+      "loss": 1.3757,
+      "step": 3993
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.855930683404223e-05,
+      "loss": 1.4828,
+      "step": 3994
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8541139691350834e-05,
+      "loss": 0.4066,
+      "step": 3995
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.852297138710832e-05,
+      "loss": 1.3705,
+      "step": 3996
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.850480192378549e-05,
+      "loss": 1.2678,
+      "step": 3997
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8486631303853264e-05,
+      "loss": 1.1879,
+      "step": 3998
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.846845952978279e-05,
+      "loss": 1.1097,
+      "step": 3999
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8450286604045324e-05,
+      "loss": 1.1409,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8432112529112295e-05,
+      "loss": 0.8454,
+      "step": 4001
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8413937307455255e-05,
+      "loss": 0.6461,
+      "step": 4002
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8395760941545986e-05,
+      "loss": 1.0712,
+      "step": 4003
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8377583433856356e-05,
+      "loss": 0.9694,
+      "step": 4004
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8359404786858385e-05,
+      "loss": 0.6846,
+      "step": 4005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8341225003024346e-05,
+      "loss": 0.9473,
+      "step": 4006
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.832304408482652e-05,
+      "loss": 1.6092,
+      "step": 4007
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8304862034737486e-05,
+      "loss": 1.0516,
+      "step": 4008
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8286678855229846e-05,
+      "loss": 0.5294,
+      "step": 4009
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8268494548776475e-05,
+      "loss": 0.7682,
+      "step": 4010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.82503091178503e-05,
+      "loss": 1.028,
+      "step": 4011
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8232122564924485e-05,
+      "loss": 0.6694,
+      "step": 4012
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8213934892472275e-05,
+      "loss": 1.3889,
+      "step": 4013
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.819574610296713e-05,
+      "loss": 1.1843,
+      "step": 4014
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.81775561988826e-05,
+      "loss": 1.4789,
+      "step": 4015
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.815936518269245e-05,
+      "loss": 0.7437,
+      "step": 4016
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8141173056870535e-05,
+      "loss": 1.5014,
+      "step": 4017
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.812297982389091e-05,
+      "loss": 1.1423,
+      "step": 4018
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8104785486227756e-05,
+      "loss": 0.9804,
+      "step": 4019
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8086590046355416e-05,
+      "loss": 0.4383,
+      "step": 4020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8068393506748366e-05,
+      "loss": 0.7629,
+      "step": 4021
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.805019586988124e-05,
+      "loss": 1.3452,
+      "step": 4022
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.803199713822883e-05,
+      "loss": 0.5653,
+      "step": 4023
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.801379731426609e-05,
+      "loss": 0.7078,
+      "step": 4024
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7995596400468057e-05,
+      "loss": 1.3052,
+      "step": 4025
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7977394399309995e-05,
+      "loss": 0.6602,
+      "step": 4026
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.795919131326727e-05,
+      "loss": 0.9357,
+      "step": 4027
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.794098714481542e-05,
+      "loss": 1.0341,
+      "step": 4028
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7922781896430115e-05,
+      "loss": 1.5785,
+      "step": 4029
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.790457557058716e-05,
+      "loss": 1.5973,
+      "step": 4030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.788636816976255e-05,
+      "loss": 0.8804,
+      "step": 4031
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.786815969643236e-05,
+      "loss": 0.9703,
+      "step": 4032
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.784995015307289e-05,
+      "loss": 1.1518,
+      "step": 4033
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.783173954216049e-05,
+      "loss": 1.4358,
+      "step": 4034
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.781352786617178e-05,
+      "loss": 1.1829,
+      "step": 4035
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7795315127583396e-05,
+      "loss": 1.1838,
+      "step": 4036
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.777710132887222e-05,
+      "loss": 1.4604,
+      "step": 4037
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.77588864725152e-05,
+      "loss": 0.7955,
+      "step": 4038
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.774067056098947e-05,
+      "loss": 1.1979,
+      "step": 4039
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.772245359677232e-05,
+      "loss": 1.6064,
+      "step": 4040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.770423558234114e-05,
+      "loss": 1.2268,
+      "step": 4041
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7686016520173504e-05,
+      "loss": 0.9058,
+      "step": 4042
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.76677964127471e-05,
+      "loss": 1.2837,
+      "step": 4043
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7649575262539776e-05,
+      "loss": 1.2398,
+      "step": 4044
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.763135307202952e-05,
+      "loss": 1.4129,
+      "step": 4045
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7613129843694436e-05,
+      "loss": 1.0578,
+      "step": 4046
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7594905580012804e-05,
+      "loss": 1.0904,
+      "step": 4047
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7576680283463024e-05,
+      "loss": 1.1772,
+      "step": 4048
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.755845395652365e-05,
+      "loss": 1.3358,
+      "step": 4049
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7540226601673374e-05,
+      "loss": 0.9274,
+      "step": 4050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7521998221391004e-05,
+      "loss": 1.1503,
+      "step": 4051
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.750376881815552e-05,
+      "loss": 1.3534,
+      "step": 4052
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7485538394446026e-05,
+      "loss": 0.564,
+      "step": 4053
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.746730695274176e-05,
+      "loss": 1.5027,
+      "step": 4054
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.74490744955221e-05,
+      "loss": 0.4322,
+      "step": 4055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7430841025266604e-05,
+      "loss": 1.0999,
+      "step": 4056
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.741260654445486e-05,
+      "loss": 1.2561,
+      "step": 4057
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.739437105556673e-05,
+      "loss": 0.645,
+      "step": 4058
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7376134561082106e-05,
+      "loss": 1.5604,
+      "step": 4059
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7357897063481095e-05,
+      "loss": 1.281,
+      "step": 4060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.733965856524386e-05,
+      "loss": 0.7217,
+      "step": 4061
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.732141906885078e-05,
+      "loss": 1.0922,
+      "step": 4062
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.73031785767823e-05,
+      "loss": 1.3617,
+      "step": 4063
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7284937091519056e-05,
+      "loss": 1.3572,
+      "step": 4064
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.726669461554179e-05,
+      "loss": 1.1365,
+      "step": 4065
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7248451151331386e-05,
+      "loss": 0.7769,
+      "step": 4066
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7230206701368864e-05,
+      "loss": 0.9152,
+      "step": 4067
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.721196126813535e-05,
+      "loss": 1.0436,
+      "step": 4068
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.719371485411218e-05,
+      "loss": 0.9067,
+      "step": 4069
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.717546746178073e-05,
+      "loss": 1.1137,
+      "step": 4070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.715721909362257e-05,
+      "loss": 1.1643,
+      "step": 4071
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.713896975211937e-05,
+      "loss": 1.2307,
+      "step": 4072
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7120719439752966e-05,
+      "loss": 0.6857,
+      "step": 4073
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.710246815900529e-05,
+      "loss": 1.2592,
+      "step": 4074
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.708421591235844e-05,
+      "loss": 0.5583,
+      "step": 4075
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7065962702294615e-05,
+      "loss": 1.0871,
+      "step": 4076
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.704770853129615e-05,
+      "loss": 1.0184,
+      "step": 4077
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.702945340184555e-05,
+      "loss": 1.3489,
+      "step": 4078
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.701119731642537e-05,
+      "loss": 0.5409,
+      "step": 4079
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.699294027751839e-05,
+      "loss": 1.4624,
+      "step": 4080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.697468228760746e-05,
+      "loss": 0.9842,
+      "step": 4081
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.695642334917555e-05,
+      "loss": 1.5054,
+      "step": 4082
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6938163464705804e-05,
+      "loss": 1.0516,
+      "step": 4083
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6919902636681465e-05,
+      "loss": 1.023,
+      "step": 4084
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6901640867585924e-05,
+      "loss": 1.0785,
+      "step": 4085
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.688337815990267e-05,
+      "loss": 0.7229,
+      "step": 4086
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.686511451611536e-05,
+      "loss": 0.799,
+      "step": 4087
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.684684993870772e-05,
+      "loss": 0.8351,
+      "step": 4088
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6828584430163666e-05,
+      "loss": 0.4925,
+      "step": 4089
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6810317992967224e-05,
+      "loss": 1.4556,
+      "step": 4090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.67920506296025e-05,
+      "loss": 1.8893,
+      "step": 4091
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.677378234255379e-05,
+      "loss": 1.3177,
+      "step": 4092
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6755513134305474e-05,
+      "loss": 1.2973,
+      "step": 4093
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.673724300734209e-05,
+      "loss": 1.1897,
+      "step": 4094
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.671897196414826e-05,
+      "loss": 0.7978,
+      "step": 4095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.670070000720875e-05,
+      "loss": 1.2083,
+      "step": 4096
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6682427139008466e-05,
+      "loss": 1.1885,
+      "step": 4097
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6664153362032415e-05,
+      "loss": 1.2217,
+      "step": 4098
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.664587867876573e-05,
+      "loss": 1.2295,
+      "step": 4099
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.662760309169369e-05,
+      "loss": 1.1626,
+      "step": 4100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.660932660330168e-05,
+      "loss": 1.1352,
+      "step": 4101
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6591049216075206e-05,
+      "loss": 1.1615,
+      "step": 4102
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.65727709324999e-05,
+      "loss": 1.0629,
+      "step": 4103
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.65544917550615e-05,
+      "loss": 0.6489,
+      "step": 4104
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.65362116862459e-05,
+      "loss": 1.0476,
+      "step": 4105
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6517930728539094e-05,
+      "loss": 1.3535,
+      "step": 4106
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.649964888442717e-05,
+      "loss": 1.1186,
+      "step": 4107
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.648136615639641e-05,
+      "loss": 1.1909,
+      "step": 4108
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.646308254693312e-05,
+      "loss": 1.5548,
+      "step": 4109
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.644479805852382e-05,
+      "loss": 1.5039,
+      "step": 4110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6426512693655086e-05,
+      "loss": 1.627,
+      "step": 4111
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6408226454813654e-05,
+      "loss": 0.6439,
+      "step": 4112
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.638993934448632e-05,
+      "loss": 1.256,
+      "step": 4113
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.637165136516007e-05,
+      "loss": 1.1792,
+      "step": 4114
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.635336251932198e-05,
+      "loss": 1.4251,
+      "step": 4115
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.63350728094592e-05,
+      "loss": 1.3608,
+      "step": 4116
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6316782238059084e-05,
+      "loss": 1.3423,
+      "step": 4117
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.629849080760901e-05,
+      "loss": 1.6513,
+      "step": 4118
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.628019852059656e-05,
+      "loss": 1.6813,
+      "step": 4119
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.626190537950935e-05,
+      "loss": 0.4367,
+      "step": 4120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6243611386835205e-05,
+      "loss": 1.2986,
+      "step": 4121
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.622531654506197e-05,
+      "loss": 0.6903,
+      "step": 4122
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6207020856677665e-05,
+      "loss": 0.7729,
+      "step": 4123
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6188724324170404e-05,
+      "loss": 1.4418,
+      "step": 4124
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.617042695002843e-05,
+      "loss": 0.557,
+      "step": 4125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6152128736740086e-05,
+      "loss": 0.4436,
+      "step": 4126
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6133829686793837e-05,
+      "loss": 0.9336,
+      "step": 4127
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.611552980267827e-05,
+      "loss": 1.096,
+      "step": 4128
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6097229086882055e-05,
+      "loss": 1.2115,
+      "step": 4129
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6078927541894e-05,
+      "loss": 0.9703,
+      "step": 4130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.606062517020303e-05,
+      "loss": 1.5324,
+      "step": 4131
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.604232197429817e-05,
+      "loss": 1.4755,
+      "step": 4132
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.602401795666857e-05,
+      "loss": 0.4034,
+      "step": 4133
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.600571311980347e-05,
+      "loss": 0.5804,
+      "step": 4134
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5987407466192234e-05,
+      "loss": 1.3834,
+      "step": 4135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.596910099832434e-05,
+      "loss": 0.6365,
+      "step": 4136
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5950793718689364e-05,
+      "loss": 1.3017,
+      "step": 4137
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.593248562977702e-05,
+      "loss": 1.1469,
+      "step": 4138
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.591417673407711e-05,
+      "loss": 1.5989,
+      "step": 4139
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5895867034079555e-05,
+      "loss": 1.2868,
+      "step": 4140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.587755653227435e-05,
+      "loss": 1.1704,
+      "step": 4141
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.585924523115166e-05,
+      "loss": 1.3799,
+      "step": 4142
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5840933133201704e-05,
+      "loss": 0.4926,
+      "step": 4143
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.582262024091486e-05,
+      "loss": 0.8137,
+      "step": 4144
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.580430655678156e-05,
+      "loss": 0.7451,
+      "step": 4145
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5785992083292404e-05,
+      "loss": 1.0055,
+      "step": 4146
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5767676822938044e-05,
+      "loss": 1.0761,
+      "step": 4147
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5749360778209246e-05,
+      "loss": 1.1389,
+      "step": 4148
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.573104395159693e-05,
+      "loss": 1.3838,
+      "step": 4149
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.571272634559208e-05,
+      "loss": 0.7941,
+      "step": 4150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.569440796268578e-05,
+      "loss": 0.7053,
+      "step": 4151
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.567608880536926e-05,
+      "loss": 0.7693,
+      "step": 4152
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.565776887613382e-05,
+      "loss": 0.6101,
+      "step": 4153
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5639448177470864e-05,
+      "loss": 1.1494,
+      "step": 4154
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.562112671187194e-05,
+      "loss": 1.2267,
+      "step": 4155
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5602804481828654e-05,
+      "loss": 0.5363,
+      "step": 4156
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.558448148983274e-05,
+      "loss": 1.3106,
+      "step": 4157
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5566157738376034e-05,
+      "loss": 1.3847,
+      "step": 4158
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5547833229950465e-05,
+      "loss": 1.5057,
+      "step": 4159
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5529507967048066e-05,
+      "loss": 1.2014,
+      "step": 4160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5511181952161004e-05,
+      "loss": 0.6549,
+      "step": 4161
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.54928551877815e-05,
+      "loss": 1.4417,
+      "step": 4162
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5474527676401913e-05,
+      "loss": 1.308,
+      "step": 4163
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5456199420514674e-05,
+      "loss": 1.4005,
+      "step": 4164
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.543787042261236e-05,
+      "loss": 0.6754,
+      "step": 4165
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.541954068518757e-05,
+      "loss": 0.6666,
+      "step": 4166
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5401210210733125e-05,
+      "loss": 1.2823,
+      "step": 4167
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.538287900174181e-05,
+      "loss": 1.3969,
+      "step": 4168
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.536454706070662e-05,
+      "loss": 1.2878,
+      "step": 4169
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5346214390120574e-05,
+      "loss": 0.7653,
+      "step": 4170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.532788099247685e-05,
+      "loss": 0.7612,
+      "step": 4171
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.530954687026867e-05,
+      "loss": 1.0778,
+      "step": 4172
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.529121202598939e-05,
+      "loss": 1.2541,
+      "step": 4173
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.527287646213245e-05,
+      "loss": 1.1528,
+      "step": 4174
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.525454018119139e-05,
+      "loss": 0.6861,
+      "step": 4175
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5236203185659884e-05,
+      "loss": 1.4098,
+      "step": 4176
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.521786547803162e-05,
+      "loss": 1.298,
+      "step": 4177
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.519952706080047e-05,
+      "loss": 1.3147,
+      "step": 4178
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.518118793646035e-05,
+      "loss": 1.2339,
+      "step": 4179
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5162848107505284e-05,
+      "loss": 0.5056,
+      "step": 4180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.514450757642939e-05,
+      "loss": 1.3468,
+      "step": 4181
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5126166345726893e-05,
+      "loss": 0.8003,
+      "step": 4182
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.510782441789211e-05,
+      "loss": 0.9015,
+      "step": 4183
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.508948179541945e-05,
+      "loss": 0.9966,
+      "step": 4184
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.50711384808034e-05,
+      "loss": 1.0107,
+      "step": 4185
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5052794476538574e-05,
+      "loss": 1.0732,
+      "step": 4186
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.503444978511965e-05,
+      "loss": 1.2003,
+      "step": 4187
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.501610440904141e-05,
+      "loss": 1.7181,
+      "step": 4188
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.499775835079876e-05,
+      "loss": 1.1453,
+      "step": 4189
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.497941161288664e-05,
+      "loss": 0.5613,
+      "step": 4190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.49610641978001e-05,
+      "loss": 0.8324,
+      "step": 4191
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4942716108034345e-05,
+      "loss": 0.6103,
+      "step": 4192
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.492436734608456e-05,
+      "loss": 1.2325,
+      "step": 4193
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.490601791444614e-05,
+      "loss": 1.1379,
+      "step": 4194
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.488766781561445e-05,
+      "loss": 1.005,
+      "step": 4195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4869317052085065e-05,
+      "loss": 1.0647,
+      "step": 4196
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.485096562635359e-05,
+      "loss": 0.4293,
+      "step": 4197
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.483261354091569e-05,
+      "loss": 1.3993,
+      "step": 4198
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.481426079826719e-05,
+      "loss": 1.4178,
+      "step": 4199
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.479590740090392e-05,
+      "loss": 1.1751,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.477755335132192e-05,
+      "loss": 1.8518,
+      "step": 4201
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4759198652017174e-05,
+      "loss": 1.3781,
+      "step": 4202
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4740843305485875e-05,
+      "loss": 1.1,
+      "step": 4203
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.472248731422425e-05,
+      "loss": 1.0778,
+      "step": 4204
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4704130680728584e-05,
+      "loss": 1.305,
+      "step": 4205
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4685773407495323e-05,
+      "loss": 0.7726,
+      "step": 4206
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.466741549702095e-05,
+      "loss": 0.886,
+      "step": 4207
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4649056951802044e-05,
+      "loss": 1.2404,
+      "step": 4208
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.463069777433527e-05,
+      "loss": 1.5396,
+      "step": 4209
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.46123379671174e-05,
+      "loss": 0.9392,
+      "step": 4210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.459397753264525e-05,
+      "loss": 1.3043,
+      "step": 4211
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.457561647341576e-05,
+      "loss": 1.3939,
+      "step": 4212
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.455725479192593e-05,
+      "loss": 1.5532,
+      "step": 4213
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4538892490672856e-05,
+      "loss": 1.2003,
+      "step": 4214
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.452052957215372e-05,
+      "loss": 1.5192,
+      "step": 4215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.450216603886579e-05,
+      "loss": 0.3824,
+      "step": 4216
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4483801893306396e-05,
+      "loss": 0.571,
+      "step": 4217
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.446543713797298e-05,
+      "loss": 0.8425,
+      "step": 4218
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.444707177536305e-05,
+      "loss": 1.2322,
+      "step": 4219
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.442870580797422e-05,
+      "loss": 0.9863,
+      "step": 4220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4410339238304144e-05,
+      "loss": 1.1981,
+      "step": 4221
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4391972068850586e-05,
+      "loss": 1.3418,
+      "step": 4222
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.437360430211138e-05,
+      "loss": 1.312,
+      "step": 4223
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.435523594058447e-05,
+      "loss": 1.0584,
+      "step": 4224
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.433686698676782e-05,
+      "loss": 1.3663,
+      "step": 4225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4318497443159564e-05,
+      "loss": 1.3504,
+      "step": 4226
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.430012731225782e-05,
+      "loss": 0.5793,
+      "step": 4227
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.428175659656087e-05,
+      "loss": 1.5253,
+      "step": 4228
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4263385298567e-05,
+      "loss": 0.9502,
+      "step": 4229
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.424501342077465e-05,
+      "loss": 0.7976,
+      "step": 4230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4226640965682276e-05,
+      "loss": 1.2514,
+      "step": 4231
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4208267935788424e-05,
+      "loss": 1.1413,
+      "step": 4232
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4189894333591774e-05,
+      "loss": 1.2089,
+      "step": 4233
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4171520161591006e-05,
+      "loss": 1.0839,
+      "step": 4234
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4153145422284926e-05,
+      "loss": 0.7856,
+      "step": 4235
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.413477011817239e-05,
+      "loss": 0.5781,
+      "step": 4236
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.411639425175238e-05,
+      "loss": 1.0614,
+      "step": 4237
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.40980178255239e-05,
+      "loss": 1.2026,
+      "step": 4238
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4079640841986026e-05,
+      "loss": 0.9275,
+      "step": 4239
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.406126330363796e-05,
+      "loss": 1.5206,
+      "step": 4240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4042885212978956e-05,
+      "loss": 1.0993,
+      "step": 4241
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.402450657250833e-05,
+      "loss": 1.2439,
+      "step": 4242
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.400612738472548e-05,
+      "loss": 0.4205,
+      "step": 4243
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.398774765212988e-05,
+      "loss": 0.9312,
+      "step": 4244
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.396936737722109e-05,
+      "loss": 1.6536,
+      "step": 4245
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.395098656249873e-05,
+      "loss": 1.1739,
+      "step": 4246
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.393260521046249e-05,
+      "loss": 1.148,
+      "step": 4247
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3914223323612145e-05,
+      "loss": 1.8876,
+      "step": 4248
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.389584090444755e-05,
+      "loss": 1.0086,
+      "step": 4249
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.387745795546857e-05,
+      "loss": 1.7735,
+      "step": 4250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.385907447917526e-05,
+      "loss": 0.8893,
+      "step": 4251
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3840690478067625e-05,
+      "loss": 1.1159,
+      "step": 4252
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.382230595464582e-05,
+      "loss": 1.297,
+      "step": 4253
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3803920911410036e-05,
+      "loss": 1.3372,
+      "step": 4254
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.378553535086056e-05,
+      "loss": 1.1413,
+      "step": 4255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.376714927549772e-05,
+      "loss": 1.4157,
+      "step": 4256
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3748762687821916e-05,
+      "loss": 0.6985,
+      "step": 4257
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.373037559033366e-05,
+      "loss": 1.1781,
+      "step": 4258
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.371198798553346e-05,
+      "loss": 1.4145,
+      "step": 4259
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3693599875921996e-05,
+      "loss": 0.726,
+      "step": 4260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.36752112639999e-05,
+      "loss": 1.2156,
+      "step": 4261
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.365682215226797e-05,
+      "loss": 1.2914,
+      "step": 4262
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3638432543227e-05,
+      "loss": 1.6235,
+      "step": 4263
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3620042439377904e-05,
+      "loss": 1.3946,
+      "step": 4264
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.360165184322162e-05,
+      "loss": 0.8733,
+      "step": 4265
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.358326075725919e-05,
+      "loss": 0.5039,
+      "step": 4266
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3564869183991705e-05,
+      "loss": 1.3698,
+      "step": 4267
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.354647712592032e-05,
+      "loss": 1.0619,
+      "step": 4268
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3528084585546276e-05,
+      "loss": 1.7112,
+      "step": 4269
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3509691565370846e-05,
+      "loss": 1.1633,
+      "step": 4270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.349129806789539e-05,
+      "loss": 1.5726,
+      "step": 4271
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.347290409562134e-05,
+      "loss": 1.429,
+      "step": 4272
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3454509651050156e-05,
+      "loss": 1.2936,
+      "step": 4273
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3436114736683426e-05,
+      "loss": 1.3966,
+      "step": 4274
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.341771935502272e-05,
+      "loss": 1.2553,
+      "step": 4275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.339932350856977e-05,
+      "loss": 1.1444,
+      "step": 4276
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.338092719982625e-05,
+      "loss": 0.9948,
+      "step": 4277
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.336253043129402e-05,
+      "loss": 0.9109,
+      "step": 4278
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.334413320547491e-05,
+      "loss": 1.1058,
+      "step": 4279
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.332573552487087e-05,
+      "loss": 1.486,
+      "step": 4280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.330733739198389e-05,
+      "loss": 1.2349,
+      "step": 4281
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3288938809316e-05,
+      "loss": 0.7045,
+      "step": 4282
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3270539779369344e-05,
+      "loss": 1.2811,
+      "step": 4283
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.325214030464606e-05,
+      "loss": 1.1562,
+      "step": 4284
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3233740387648424e-05,
+      "loss": 0.9774,
+      "step": 4285
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.321534003087868e-05,
+      "loss": 0.8603,
+      "step": 4286
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.319693923683924e-05,
+      "loss": 0.9699,
+      "step": 4287
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3178538008032464e-05,
+      "loss": 1.349,
+      "step": 4288
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3160136346960855e-05,
+      "loss": 1.3821,
+      "step": 4289
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3141734256126927e-05,
+      "loss": 1.3293,
+      "step": 4290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.312333173803329e-05,
+      "loss": 1.613,
+      "step": 4291
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3104928795182574e-05,
+      "loss": 1.4119,
+      "step": 4292
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.308652543007749e-05,
+      "loss": 0.671,
+      "step": 4293
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3068121645220814e-05,
+      "loss": 0.7253,
+      "step": 4294
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.304971744311534e-05,
+      "loss": 1.4939,
+      "step": 4295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.303131282626398e-05,
+      "loss": 1.0307,
+      "step": 4296
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.301290779716963e-05,
+      "loss": 1.3444,
+      "step": 4297
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.29945023583353e-05,
+      "loss": 0.8124,
+      "step": 4298
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.297609651226405e-05,
+      "loss": 1.0673,
+      "step": 4299
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2957690261458934e-05,
+      "loss": 0.7972,
+      "step": 4300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.293928360842316e-05,
+      "loss": 1.3359,
+      "step": 4301
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.292087655565989e-05,
+      "loss": 0.8788,
+      "step": 4302
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.290246910567243e-05,
+      "loss": 1.0403,
+      "step": 4303
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2884061260964065e-05,
+      "loss": 1.4447,
+      "step": 4304
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2865653024038194e-05,
+      "loss": 1.4053,
+      "step": 4305
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.284724439739825e-05,
+      "loss": 0.7525,
+      "step": 4306
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.282883538354767e-05,
+      "loss": 1.1179,
+      "step": 4307
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.281042598499003e-05,
+      "loss": 1.3728,
+      "step": 4308
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.279201620422887e-05,
+      "loss": 0.6926,
+      "step": 4309
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2773606043767875e-05,
+      "loss": 1.1346,
+      "step": 4310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2755195506110676e-05,
+      "loss": 1.4339,
+      "step": 4311
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2736784593761066e-05,
+      "loss": 0.7001,
+      "step": 4312
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2718373309222794e-05,
+      "loss": 1.0079,
+      "step": 4313
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.269996165499973e-05,
+      "loss": 0.937,
+      "step": 4314
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2681549633595736e-05,
+      "loss": 1.6542,
+      "step": 4315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.266313724751478e-05,
+      "loss": 1.5941,
+      "step": 4316
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2644724499260836e-05,
+      "loss": 1.1579,
+      "step": 4317
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2626311391337934e-05,
+      "loss": 1.0762,
+      "step": 4318
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.260789792625019e-05,
+      "loss": 1.4365,
+      "step": 4319
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.258948410650172e-05,
+      "loss": 1.2789,
+      "step": 4320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2571069934596703e-05,
+      "loss": 1.0209,
+      "step": 4321
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.25526554130394e-05,
+      "loss": 1.4799,
+      "step": 4322
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2534240544334053e-05,
+      "loss": 0.4681,
+      "step": 4323
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.251582533098503e-05,
+      "loss": 0.6666,
+      "step": 4324
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.249740977549667e-05,
+      "loss": 1.0932,
+      "step": 4325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2478993880373415e-05,
+      "loss": 1.2451,
+      "step": 4326
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2460577648119736e-05,
+      "loss": 1.2161,
+      "step": 4327
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.244216108124013e-05,
+      "loss": 1.2915,
+      "step": 4328
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.242374418223916e-05,
+      "loss": 1.3501,
+      "step": 4329
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.240532695362143e-05,
+      "loss": 1.0466,
+      "step": 4330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2386909397891595e-05,
+      "loss": 1.4977,
+      "step": 4331
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.236849151755434e-05,
+      "loss": 1.3523,
+      "step": 4332
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.235007331511442e-05,
+      "loss": 1.5087,
+      "step": 4333
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2331654793076576e-05,
+      "loss": 1.2564,
+      "step": 4334
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2313235953945685e-05,
+      "loss": 1.3871,
+      "step": 4335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.229481680022657e-05,
+      "loss": 1.3049,
+      "step": 4336
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2276397334424164e-05,
+      "loss": 1.4427,
+      "step": 4337
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.22579775590434e-05,
+      "loss": 1.3069,
+      "step": 4338
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2239557476589304e-05,
+      "loss": 1.4195,
+      "step": 4339
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.22211370895669e-05,
+      "loss": 1.1032,
+      "step": 4340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.220271640048124e-05,
+      "loss": 1.1928,
+      "step": 4341
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.218429541183748e-05,
+      "loss": 1.1311,
+      "step": 4342
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.216587412614077e-05,
+      "loss": 0.9546,
+      "step": 4343
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.214745254589629e-05,
+      "loss": 1.0978,
+      "step": 4344
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.212903067360929e-05,
+      "loss": 0.8793,
+      "step": 4345
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.211060851178508e-05,
+      "loss": 0.9194,
+      "step": 4346
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.209218606292894e-05,
+      "loss": 1.4842,
+      "step": 4347
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.207376332954624e-05,
+      "loss": 1.3601,
+      "step": 4348
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.205534031414239e-05,
+      "loss": 1.3863,
+      "step": 4349
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2036917019222806e-05,
+      "loss": 1.261,
+      "step": 4350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2018493447292984e-05,
+      "loss": 0.6119,
+      "step": 4351
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.200006960085841e-05,
+      "loss": 1.2624,
+      "step": 4352
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.198164548242465e-05,
+      "loss": 1.5364,
+      "step": 4353
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.196322109449728e-05,
+      "loss": 1.892,
+      "step": 4354
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1944796439581924e-05,
+      "loss": 1.1985,
+      "step": 4355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.192637152018425e-05,
+      "loss": 1.6326,
+      "step": 4356
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1907946338809944e-05,
+      "loss": 1.2099,
+      "step": 4357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.188952089796474e-05,
+      "loss": 1.5465,
+      "step": 4358
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1871095200154385e-05,
+      "loss": 0.5412,
+      "step": 4359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.185266924788471e-05,
+      "loss": 1.5135,
+      "step": 4360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.183424304366152e-05,
+      "loss": 1.1735,
+      "step": 4361
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.181581658999072e-05,
+      "loss": 1.5438,
+      "step": 4362
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.179738988937817e-05,
+      "loss": 1.3461,
+      "step": 4363
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.177896294432984e-05,
+      "loss": 1.2447,
+      "step": 4364
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.176053575735169e-05,
+      "loss": 0.8981,
+      "step": 4365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.17421083309497e-05,
+      "loss": 1.3334,
+      "step": 4366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.172368066762996e-05,
+      "loss": 0.8226,
+      "step": 4367
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.170525276989847e-05,
+      "loss": 0.7918,
+      "step": 4368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.168682464026139e-05,
+      "loss": 1.3202,
+      "step": 4369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1668396281224795e-05,
+      "loss": 0.6162,
+      "step": 4370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.164996769529491e-05,
+      "loss": 1.1577,
+      "step": 4371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.163153888497788e-05,
+      "loss": 1.5082,
+      "step": 4372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.161310985277994e-05,
+      "loss": 1.1712,
+      "step": 4373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1594680601207345e-05,
+      "loss": 1.3545,
+      "step": 4374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1576251132766386e-05,
+      "loss": 1.219,
+      "step": 4375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.155782144996336e-05,
+      "loss": 1.2527,
+      "step": 4376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.153939155530463e-05,
+      "loss": 1.23,
+      "step": 4377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1520961451296556e-05,
+      "loss": 1.2621,
+      "step": 4378
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.150253114044553e-05,
+      "loss": 0.6542,
+      "step": 4379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1484100625257994e-05,
+      "loss": 1.0384,
+      "step": 4380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.146566990824039e-05,
+      "loss": 1.3545,
+      "step": 4381
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.144723899189922e-05,
+      "loss": 0.8788,
+      "step": 4382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.142880787874098e-05,
+      "loss": 1.0034,
+      "step": 4383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.141037657127219e-05,
+      "loss": 1.3553,
+      "step": 4384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.139194507199945e-05,
+      "loss": 1.2145,
+      "step": 4385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.137351338342932e-05,
+      "loss": 1.6032,
+      "step": 4386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.135508150806843e-05,
+      "loss": 1.1053,
+      "step": 4387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.13366494484234e-05,
+      "loss": 1.2768,
+      "step": 4388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1318217207000926e-05,
+      "loss": 0.7329,
+      "step": 4389
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.12997847863077e-05,
+      "loss": 0.782,
+      "step": 4390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1281352188850395e-05,
+      "loss": 0.4962,
+      "step": 4391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1262919417135793e-05,
+      "loss": 1.4,
+      "step": 4392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.124448647367062e-05,
+      "loss": 1.0431,
+      "step": 4393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1226053360961687e-05,
+      "loss": 0.6191,
+      "step": 4394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.12076200815158e-05,
+      "loss": 1.2754,
+      "step": 4395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.118918663783978e-05,
+      "loss": 0.9001,
+      "step": 4396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.117075303244049e-05,
+      "loss": 1.2377,
+      "step": 4397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.11523192678248e-05,
+      "loss": 0.6008,
+      "step": 4398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.11338853464996e-05,
+      "loss": 1.0013,
+      "step": 4399
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1115451270971825e-05,
+      "loss": 1.5596,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.109701704374841e-05,
+      "loss": 0.6754,
+      "step": 4401
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.107858266733632e-05,
+      "loss": 0.9452,
+      "step": 4402
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1060148144242515e-05,
+      "loss": 1.2668,
+      "step": 4403
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.104171347697401e-05,
+      "loss": 0.4141,
+      "step": 4404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.102327866803782e-05,
+      "loss": 0.9535,
+      "step": 4405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.100484371994099e-05,
+      "loss": 0.9889,
+      "step": 4406
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0986408635190575e-05,
+      "loss": 0.596,
+      "step": 4407
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.096797341629366e-05,
+      "loss": 1.2556,
+      "step": 4408
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.094953806575733e-05,
+      "loss": 1.4423,
+      "step": 4409
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0931102586088696e-05,
+      "loss": 0.9938,
+      "step": 4410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.09126669797949e-05,
+      "loss": 0.9529,
+      "step": 4411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.089423124938307e-05,
+      "loss": 0.895,
+      "step": 4412
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0875795397360406e-05,
+      "loss": 1.3162,
+      "step": 4413
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.085735942623405e-05,
+      "loss": 1.4269,
+      "step": 4414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0838923338511224e-05,
+      "loss": 0.7019,
+      "step": 4415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.082048713669913e-05,
+      "loss": 1.2185,
+      "step": 4416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0802050823305e-05,
+      "loss": 0.5878,
+      "step": 4417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.078361440083608e-05,
+      "loss": 1.2029,
+      "step": 4418
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.076517787179963e-05,
+      "loss": 0.6428,
+      "step": 4419
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.074674123870291e-05,
+      "loss": 1.448,
+      "step": 4420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0728304504053225e-05,
+      "loss": 1.2422,
+      "step": 4421
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0709867670357856e-05,
+      "loss": 1.3782,
+      "step": 4422
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.069143074012416e-05,
+      "loss": 1.3155,
+      "step": 4423
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.067299371585941e-05,
+      "loss": 1.589,
+      "step": 4424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.065455660007098e-05,
+      "loss": 1.3421,
+      "step": 4425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.063611939526621e-05,
+      "loss": 1.3463,
+      "step": 4426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0617682103952477e-05,
+      "loss": 1.424,
+      "step": 4427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0599244728637144e-05,
+      "loss": 1.3963,
+      "step": 4428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.05808072718276e-05,
+      "loss": 1.21,
+      "step": 4429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.056236973603127e-05,
+      "loss": 0.7482,
+      "step": 4430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0543932123755535e-05,
+      "loss": 1.1825,
+      "step": 4431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0525494437507824e-05,
+      "loss": 1.3653,
+      "step": 4432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.050705667979557e-05,
+      "loss": 1.0192,
+      "step": 4433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.048861885312621e-05,
+      "loss": 1.3382,
+      "step": 4434
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0470180960007195e-05,
+      "loss": 1.4264,
+      "step": 4435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0451743002945995e-05,
+      "loss": 1.1517,
+      "step": 4436
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.043330498445006e-05,
+      "loss": 1.0161,
+      "step": 4437
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.041486690702687e-05,
+      "loss": 1.1603,
+      "step": 4438
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.039642877318391e-05,
+      "loss": 1.1415,
+      "step": 4439
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.037799058542868e-05,
+      "loss": 1.1409,
+      "step": 4440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.035955234626867e-05,
+      "loss": 1.5422,
+      "step": 4441
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.03411140582114e-05,
+      "loss": 1.4309,
+      "step": 4442
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.032267572376435e-05,
+      "loss": 1.4977,
+      "step": 4443
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.030423734543508e-05,
+      "loss": 0.5748,
+      "step": 4444
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.028579892573106e-05,
+      "loss": 0.953,
+      "step": 4445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.02673604671599e-05,
+      "loss": 1.6698,
+      "step": 4446
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.024892197222905e-05,
+      "loss": 1.0848,
+      "step": 4447
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.023048344344612e-05,
+      "loss": 1.0056,
+      "step": 4448
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.021204488331861e-05,
+      "loss": 1.3358,
+      "step": 4449
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.019360629435407e-05,
+      "loss": 1.0215,
+      "step": 4450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.01751676790601e-05,
+      "loss": 1.2202,
+      "step": 4451
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0156729039944196e-05,
+      "loss": 1.2096,
+      "step": 4452
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.013829037951397e-05,
+      "loss": 1.3227,
+      "step": 4453
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0119851700276954e-05,
+      "loss": 1.0806,
+      "step": 4454
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0101413004740725e-05,
+      "loss": 1.096,
+      "step": 4455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0082974295412854e-05,
+      "loss": 0.545,
+      "step": 4456
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.00645355748009e-05,
+      "loss": 1.4569,
+      "step": 4457
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0046096845412435e-05,
+      "loss": 1.0936,
+      "step": 4458
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.002765810975504e-05,
+      "loss": 1.447,
+      "step": 4459
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0009219370336277e-05,
+      "loss": 0.9071,
+      "step": 4460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.999078062966373e-05,
+      "loss": 1.4761,
+      "step": 4461
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.997234189024499e-05,
+      "loss": 1.0883,
+      "step": 4462
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9953903154587576e-05,
+      "loss": 0.872,
+      "step": 4463
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.993546442519911e-05,
+      "loss": 1.1013,
+      "step": 4464
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9917025704587164e-05,
+      "loss": 1.7916,
+      "step": 4465
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9898586995259294e-05,
+      "loss": 0.7818,
+      "step": 4466
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.988014829972305e-05,
+      "loss": 1.638,
+      "step": 4467
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.986170962048604e-05,
+      "loss": 1.5953,
+      "step": 4468
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9843270960055815e-05,
+      "loss": 0.7231,
+      "step": 4469
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.982483232093992e-05,
+      "loss": 1.1393,
+      "step": 4470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.980639370564593e-05,
+      "loss": 0.6202,
+      "step": 4471
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9787955116681406e-05,
+      "loss": 0.9964,
+      "step": 4472
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.976951655655391e-05,
+      "loss": 1.2248,
+      "step": 4473
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9751078027770956e-05,
+      "loss": 1.3607,
+      "step": 4474
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.973263953284012e-05,
+      "loss": 0.5331,
+      "step": 4475
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.971420107426895e-05,
+      "loss": 0.5006,
+      "step": 4476
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9695762654564946e-05,
+      "loss": 1.1454,
+      "step": 4477
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.967732427623566e-05,
+      "loss": 0.5421,
+      "step": 4478
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.965888594178861e-05,
+      "loss": 0.5724,
+      "step": 4479
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.964044765373134e-05,
+      "loss": 1.2678,
+      "step": 4480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.962200941457133e-05,
+      "loss": 0.9564,
+      "step": 4481
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9603571226816085e-05,
+      "loss": 1.244,
+      "step": 4482
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.958513309297314e-05,
+      "loss": 1.2588,
+      "step": 4483
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9566695015549953e-05,
+      "loss": 1.2076,
+      "step": 4484
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.954825699705403e-05,
+      "loss": 0.3648,
+      "step": 4485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.95298190399928e-05,
+      "loss": 0.9946,
+      "step": 4486
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9511381146873795e-05,
+      "loss": 0.7094,
+      "step": 4487
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.949294332020444e-05,
+      "loss": 1.1788,
+      "step": 4488
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9474505562492194e-05,
+      "loss": 1.1761,
+      "step": 4489
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9456067876244476e-05,
+      "loss": 1.2573,
+      "step": 4490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.943763026396874e-05,
+      "loss": 1.2397,
+      "step": 4491
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9419192728172406e-05,
+      "loss": 1.0864,
+      "step": 4492
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9400755271362854e-05,
+      "loss": 1.1902,
+      "step": 4493
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9382317896047535e-05,
+      "loss": 1.5228,
+      "step": 4494
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9363880604733794e-05,
+      "loss": 1.0756,
+      "step": 4495
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.934544339992904e-05,
+      "loss": 0.9592,
+      "step": 4496
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.93270062841406e-05,
+      "loss": 1.4686,
+      "step": 4497
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9308569259875854e-05,
+      "loss": 0.9839,
+      "step": 4498
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9290132329642155e-05,
+      "loss": 1.2315,
+      "step": 4499
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927169549594679e-05,
+      "loss": 0.4985,
+      "step": 4500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9253258761297096e-05,
+      "loss": 1.5876,
+      "step": 4501
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.923482212820038e-05,
+      "loss": 0.9975,
+      "step": 4502
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.921638559916394e-05,
+      "loss": 0.6402,
+      "step": 4503
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9197949176695016e-05,
+      "loss": 1.6052,
+      "step": 4504
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.917951286330088e-05,
+      "loss": 0.9496,
+      "step": 4505
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9161076661488795e-05,
+      "loss": 0.9418,
+      "step": 4506
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9142640573765965e-05,
+      "loss": 1.3721,
+      "step": 4507
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.912420460263961e-05,
+      "loss": 0.9422,
+      "step": 4508
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.910576875061692e-05,
+      "loss": 1.4047,
+      "step": 4509
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9087333020205114e-05,
+      "loss": 1.0491,
+      "step": 4510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9068897413911315e-05,
+      "loss": 1.6852,
+      "step": 4511
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9050461934242694e-05,
+      "loss": 0.5652,
+      "step": 4512
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.903202658370634e-05,
+      "loss": 0.635,
+      "step": 4513
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.901359136480943e-05,
+      "loss": 0.4757,
+      "step": 4514
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.899515628005902e-05,
+      "loss": 0.5117,
+      "step": 4515
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8976721331962186e-05,
+      "loss": 1.3941,
+      "step": 4516
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8958286523026e-05,
+      "loss": 0.5963,
+      "step": 4517
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.89398518557575e-05,
+      "loss": 1.2443,
+      "step": 4518
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.892141733266371e-05,
+      "loss": 0.6124,
+      "step": 4519
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8902982956251584e-05,
+      "loss": 1.0886,
+      "step": 4520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.888454872902818e-05,
+      "loss": 0.6622,
+      "step": 4521
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.886611465350041e-05,
+      "loss": 1.5248,
+      "step": 4522
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.884768073217523e-05,
+      "loss": 0.6079,
+      "step": 4523
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8829246967559524e-05,
+      "loss": 1.2285,
+      "step": 4524
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.881081336216023e-05,
+      "loss": 0.8373,
+      "step": 4525
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.879237991848422e-05,
+      "loss": 0.6893,
+      "step": 4526
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.877394663903833e-05,
+      "loss": 0.6708,
+      "step": 4527
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.875551352632939e-05,
+      "loss": 1.1211,
+      "step": 4528
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.873708058286422e-05,
+      "loss": 1.2452,
+      "step": 4529
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.871864781114962e-05,
+      "loss": 0.9233,
+      "step": 4530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8700215213692324e-05,
+      "loss": 0.6008,
+      "step": 4531
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.868178279299907e-05,
+      "loss": 1.0471,
+      "step": 4532
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.86633505515766e-05,
+      "loss": 1.2455,
+      "step": 4533
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.864491849193159e-05,
+      "loss": 1.4659,
+      "step": 4534
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8626486616570686e-05,
+      "loss": 0.5603,
+      "step": 4535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.860805492800056e-05,
+      "loss": 1.1873,
+      "step": 4536
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.858962342872783e-05,
+      "loss": 0.9782,
+      "step": 4537
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8571192121259046e-05,
+      "loss": 0.5562,
+      "step": 4538
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8552761008100786e-05,
+      "loss": 1.4867,
+      "step": 4539
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8534330091759624e-05,
+      "loss": 1.5261,
+      "step": 4540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.851589937474202e-05,
+      "loss": 1.5698,
+      "step": 4541
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.849746885955448e-05,
+      "loss": 1.2559,
+      "step": 4542
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.847903854870345e-05,
+      "loss": 1.2867,
+      "step": 4543
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8460608444695384e-05,
+      "loss": 0.9648,
+      "step": 4544
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.844217855003664e-05,
+      "loss": 1.3282,
+      "step": 4545
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.842374886723364e-05,
+      "loss": 0.6052,
+      "step": 4546
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.840531939879267e-05,
+      "loss": 0.7903,
+      "step": 4547
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.838689014722007e-05,
+      "loss": 0.9284,
+      "step": 4548
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.836846111502214e-05,
+      "loss": 1.2115,
+      "step": 4549
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.835003230470511e-05,
+      "loss": 1.3596,
+      "step": 4550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8331603718775196e-05,
+      "loss": 1.4352,
+      "step": 4551
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.831317535973862e-05,
+      "loss": 1.3805,
+      "step": 4552
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.829474723010154e-05,
+      "loss": 0.5772,
+      "step": 4553
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8276319332370066e-05,
+      "loss": 1.1557,
+      "step": 4554
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.82578916690503e-05,
+      "loss": 1.3345,
+      "step": 4555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.823946424264832e-05,
+      "loss": 0.4445,
+      "step": 4556
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8221037055670176e-05,
+      "loss": 1.2253,
+      "step": 4557
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.820261011062183e-05,
+      "loss": 0.8764,
+      "step": 4558
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.818418341000929e-05,
+      "loss": 1.1067,
+      "step": 4559
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.816575695633849e-05,
+      "loss": 1.3794,
+      "step": 4560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8147330752115304e-05,
+      "loss": 1.4152,
+      "step": 4561
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.812890479984561e-05,
+      "loss": 1.3793,
+      "step": 4562
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.811047910203527e-05,
+      "loss": 1.2042,
+      "step": 4563
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8092053661190074e-05,
+      "loss": 1.4815,
+      "step": 4564
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.807362847981576e-05,
+      "loss": 1.0139,
+      "step": 4565
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8055203560418074e-05,
+      "loss": 1.3541,
+      "step": 4566
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.803677890550273e-05,
+      "loss": 0.2943,
+      "step": 4567
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8018354517575364e-05,
+      "loss": 0.7746,
+      "step": 4568
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.799993039914161e-05,
+      "loss": 0.7636,
+      "step": 4569
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.798150655270702e-05,
+      "loss": 1.1706,
+      "step": 4570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.79630829807772e-05,
+      "loss": 1.0558,
+      "step": 4571
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.794465968585762e-05,
+      "loss": 1.2154,
+      "step": 4572
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.792623667045377e-05,
+      "loss": 0.9197,
+      "step": 4573
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.790781393707106e-05,
+      "loss": 1.5652,
+      "step": 4574
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.788939148821493e-05,
+      "loss": 1.1347,
+      "step": 4575
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.787096932639072e-05,
+      "loss": 1.0263,
+      "step": 4576
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.785254745410373e-05,
+      "loss": 0.581,
+      "step": 4577
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7834125873859245e-05,
+      "loss": 1.2209,
+      "step": 4578
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.781570458816252e-05,
+      "loss": 1.0116,
+      "step": 4579
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.779728359951877e-05,
+      "loss": 1.2102,
+      "step": 4580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.777886291043311e-05,
+      "loss": 1.0904,
+      "step": 4581
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.77604425234107e-05,
+      "loss": 0.6339,
+      "step": 4582
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7742022440956616e-05,
+      "loss": 1.3728,
+      "step": 4583
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.772360266557586e-05,
+      "loss": 0.7525,
+      "step": 4584
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.770518319977344e-05,
+      "loss": 1.4852,
+      "step": 4585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.768676404605433e-05,
+      "loss": 1.509,
+      "step": 4586
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.766834520692343e-05,
+      "loss": 1.7831,
+      "step": 4587
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.76499266848856e-05,
+      "loss": 1.2857,
+      "step": 4588
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7631508482445655e-05,
+      "loss": 1.4581,
+      "step": 4589
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7613090602108416e-05,
+      "loss": 1.2812,
+      "step": 4590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.759467304637858e-05,
+      "loss": 1.5473,
+      "step": 4591
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.757625581776085e-05,
+      "loss": 1.5028,
+      "step": 4592
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.755783891875987e-05,
+      "loss": 1.5571,
+      "step": 4593
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.753942235188028e-05,
+      "loss": 0.4806,
+      "step": 4594
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7521006119626596e-05,
+      "loss": 1.2998,
+      "step": 4595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.750259022450335e-05,
+      "loss": 1.2657,
+      "step": 4596
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7484174669014985e-05,
+      "loss": 1.0926,
+      "step": 4597
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.746575945566595e-05,
+      "loss": 1.3954,
+      "step": 4598
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.744734458696062e-05,
+      "loss": 0.7091,
+      "step": 4599
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7428930065403295e-05,
+      "loss": 0.8691,
+      "step": 4600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.74105158934983e-05,
+      "loss": 1.0198,
+      "step": 4601
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.739210207374982e-05,
+      "loss": 1.6154,
+      "step": 4602
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7373688608662084e-05,
+      "loss": 1.3302,
+      "step": 4603
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7355275500739175e-05,
+      "loss": 1.1139,
+      "step": 4604
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7336862752485226e-05,
+      "loss": 1.4816,
+      "step": 4605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.731845036640427e-05,
+      "loss": 0.6317,
+      "step": 4606
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.730003834500029e-05,
+      "loss": 1.282,
+      "step": 4607
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.728162669077721e-05,
+      "loss": 1.2667,
+      "step": 4608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.726321540623894e-05,
+      "loss": 1.1197,
+      "step": 4609
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7244804493889336e-05,
+      "loss": 1.1499,
+      "step": 4610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.722639395623215e-05,
+      "loss": 1.5868,
+      "step": 4611
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.720798379577113e-05,
+      "loss": 1.0851,
+      "step": 4612
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.718957401500998e-05,
+      "loss": 0.6711,
+      "step": 4613
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.717116461645234e-05,
+      "loss": 1.3239,
+      "step": 4614
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.715275560260177e-05,
+      "loss": 0.612,
+      "step": 4615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.71343469759618e-05,
+      "loss": 1.0526,
+      "step": 4616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.711593873903594e-05,
+      "loss": 1.3313,
+      "step": 4617
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7097530894327584e-05,
+      "loss": 1.6282,
+      "step": 4618
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.707912344434013e-05,
+      "loss": 1.095,
+      "step": 4619
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.706071639157685e-05,
+      "loss": 0.9343,
+      "step": 4620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.704230973854108e-05,
+      "loss": 1.4672,
+      "step": 4621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7023903487735974e-05,
+      "loss": 1.4905,
+      "step": 4622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.70054976416647e-05,
+      "loss": 1.0433,
+      "step": 4623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.698709220283038e-05,
+      "loss": 1.1826,
+      "step": 4624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.696868717373604e-05,
+      "loss": 1.6076,
+      "step": 4625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6950282556884665e-05,
+      "loss": 0.6381,
+      "step": 4626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6931878354779184e-05,
+      "loss": 1.1542,
+      "step": 4627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.691347456992251e-05,
+      "loss": 1.3021,
+      "step": 4628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.689507120481744e-05,
+      "loss": 1.0027,
+      "step": 4629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6876668261966735e-05,
+      "loss": 1.0036,
+      "step": 4630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.685826574387308e-05,
+      "loss": 0.4359,
+      "step": 4631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.683986365303916e-05,
+      "loss": 1.4797,
+      "step": 4632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.682146199196755e-05,
+      "loss": 0.5533,
+      "step": 4633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6803060763160785e-05,
+      "loss": 0.5133,
+      "step": 4634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6784659969121316e-05,
+      "loss": 1.1771,
+      "step": 4635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.676625961235159e-05,
+      "loss": 1.6779,
+      "step": 4636
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6747859695353946e-05,
+      "loss": 1.3218,
+      "step": 4637
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6729460220630674e-05,
+      "loss": 0.6779,
+      "step": 4638
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6711061190684e-05,
+      "loss": 0.4807,
+      "step": 4639
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6692662608016117e-05,
+      "loss": 1.5044,
+      "step": 4640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6674264475129135e-05,
+      "loss": 0.9879,
+      "step": 4641
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6655866794525085e-05,
+      "loss": 1.4551,
+      "step": 4642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.663746956870599e-05,
+      "loss": 0.7152,
+      "step": 4643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.661907280017376e-05,
+      "loss": 0.7047,
+      "step": 4644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.660067649143026e-05,
+      "loss": 1.0039,
+      "step": 4645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.658228064497728e-05,
+      "loss": 1.1549,
+      "step": 4646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.656388526331658e-05,
+      "loss": 1.2357,
+      "step": 4647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.654549034894985e-05,
+      "loss": 1.2911,
+      "step": 4648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6527095904378674e-05,
+      "loss": 1.3109,
+      "step": 4649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6508701932104607e-05,
+      "loss": 1.3796,
+      "step": 4650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6490308434629166e-05,
+      "loss": 1.3372,
+      "step": 4651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.647191541445373e-05,
+      "loss": 1.3505,
+      "step": 4652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.645352287407969e-05,
+      "loss": 1.1664,
+      "step": 4653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.643513081600829e-05,
+      "loss": 1.7176,
+      "step": 4654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.641673924274082e-05,
+      "loss": 1.2166,
+      "step": 4655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6398348156778396e-05,
+      "loss": 1.6903,
+      "step": 4656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.637995756062212e-05,
+      "loss": 1.4075,
+      "step": 4657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.636156745677301e-05,
+      "loss": 0.9775,
+      "step": 4658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.634317784773204e-05,
+      "loss": 1.2824,
+      "step": 4659
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6324788736000116e-05,
+      "loss": 1.1984,
+      "step": 4660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.630640012407802e-05,
+      "loss": 0.882,
+      "step": 4661
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.628801201446653e-05,
+      "loss": 0.576,
+      "step": 4662
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6269624409666354e-05,
+      "loss": 1.2849,
+      "step": 4663
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.62512373121781e-05,
+      "loss": 0.626,
+      "step": 4664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.62328507245023e-05,
+      "loss": 0.9534,
+      "step": 4665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.621446464913945e-05,
+      "loss": 1.3356,
+      "step": 4666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6196079088589975e-05,
+      "loss": 1.1856,
+      "step": 4667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6177694045354197e-05,
+      "loss": 1.4835,
+      "step": 4668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.615930952193238e-05,
+      "loss": 0.5588,
+      "step": 4669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.614092552082475e-05,
+      "loss": 0.9403,
+      "step": 4670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6122542044531435e-05,
+      "loss": 0.9275,
+      "step": 4671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.610415909555247e-05,
+      "loss": 1.7078,
+      "step": 4672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.608577667638785e-05,
+      "loss": 1.0832,
+      "step": 4673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.606739478953752e-05,
+      "loss": 1.2931,
+      "step": 4674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.604901343750128e-05,
+      "loss": 1.2234,
+      "step": 4675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.603063262277892e-05,
+      "loss": 1.0209,
+      "step": 4676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.601225234787012e-05,
+      "loss": 1.0355,
+      "step": 4677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.599387261527453e-05,
+      "loss": 0.6098,
+      "step": 4678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.597549342749168e-05,
+      "loss": 1.1452,
+      "step": 4679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.595711478702107e-05,
+      "loss": 1.0616,
+      "step": 4680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.593873669636204e-05,
+      "loss": 1.3451,
+      "step": 4681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.592035915801398e-05,
+      "loss": 1.3402,
+      "step": 4682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.590198217447612e-05,
+      "loss": 1.2537,
+      "step": 4683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.588360574824762e-05,
+      "loss": 0.9816,
+      "step": 4684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.586522988182761e-05,
+      "loss": 1.4191,
+      "step": 4685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5846854577715085e-05,
+      "loss": 0.7714,
+      "step": 4686
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.582847983840902e-05,
+      "loss": 0.5447,
+      "step": 4687
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5810105666408244e-05,
+      "loss": 1.5216,
+      "step": 4688
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.579173206421158e-05,
+      "loss": 1.3079,
+      "step": 4689
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.577335903431774e-05,
+      "loss": 1.3172,
+      "step": 4690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5754986579225375e-05,
+      "loss": 1.1811,
+      "step": 4691
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5736614701433e-05,
+      "loss": 1.2502,
+      "step": 4692
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.571824340343914e-05,
+      "loss": 1.0889,
+      "step": 4693
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.569987268774219e-05,
+      "loss": 1.2428,
+      "step": 4694
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5681502556840454e-05,
+      "loss": 1.3174,
+      "step": 4695
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5663133013232176e-05,
+      "loss": 1.4703,
+      "step": 4696
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.564476405941555e-05,
+      "loss": 1.4416,
+      "step": 4697
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.562639569788863e-05,
+      "loss": 1.0402,
+      "step": 4698
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.560802793114943e-05,
+      "loss": 1.3961,
+      "step": 4699
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.558966076169586e-05,
+      "loss": 1.3489,
+      "step": 4700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.557129419202579e-05,
+      "loss": 0.49,
+      "step": 4701
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.555292822463695e-05,
+      "loss": 0.3201,
+      "step": 4702
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5534562862027035e-05,
+      "loss": 0.9634,
+      "step": 4703
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.55161981066936e-05,
+      "loss": 0.9192,
+      "step": 4704
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.549783396113423e-05,
+      "loss": 1.2976,
+      "step": 4705
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.547947042784629e-05,
+      "loss": 1.2236,
+      "step": 4706
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.546110750932715e-05,
+      "loss": 0.6571,
+      "step": 4707
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.544274520807409e-05,
+      "loss": 1.0475,
+      "step": 4708
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.542438352658426e-05,
+      "loss": 1.372,
+      "step": 4709
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.540602246735478e-05,
+      "loss": 1.2457,
+      "step": 4710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.538766203288261e-05,
+      "loss": 0.8248,
+      "step": 4711
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5369302225664734e-05,
+      "loss": 1.3551,
+      "step": 4712
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.535094304819797e-05,
+      "loss": 1.1349,
+      "step": 4713
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.533258450297907e-05,
+      "loss": 1.4555,
+      "step": 4714
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.531422659250469e-05,
+      "loss": 1.2605,
+      "step": 4715
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.529586931927142e-05,
+      "loss": 0.9794,
+      "step": 4716
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.527751268577577e-05,
+      "loss": 1.5789,
+      "step": 4717
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.525915669451414e-05,
+      "loss": 1.3216,
+      "step": 4718
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.524080134798283e-05,
+      "loss": 0.5068,
+      "step": 4719
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5222446648678094e-05,
+      "loss": 1.0162,
+      "step": 4720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5204092599096085e-05,
+      "loss": 1.2199,
+      "step": 4721
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.518573920173283e-05,
+      "loss": 1.0988,
+      "step": 4722
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.516738645908431e-05,
+      "loss": 0.7283,
+      "step": 4723
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.514903437364642e-05,
+      "loss": 1.361,
+      "step": 4724
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.513068294791494e-05,
+      "loss": 0.6278,
+      "step": 4725
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.511233218438555e-05,
+      "loss": 1.0623,
+      "step": 4726
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5093982085553875e-05,
+      "loss": 1.5955,
+      "step": 4727
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.507563265391545e-05,
+      "loss": 1.604,
+      "step": 4728
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.505728389196568e-05,
+      "loss": 0.6588,
+      "step": 4729
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5038935802199896e-05,
+      "loss": 0.487,
+      "step": 4730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.502058838711337e-05,
+      "loss": 1.2226,
+      "step": 4731
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.500224164920125e-05,
+      "loss": 1.1039,
+      "step": 4732
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.498389559095859e-05,
+      "loss": 0.5101,
+      "step": 4733
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.496555021488035e-05,
+      "loss": 0.7501,
+      "step": 4734
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.494720552346144e-05,
+      "loss": 1.123,
+      "step": 4735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4928861519196607e-05,
+      "loss": 1.127,
+      "step": 4736
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4910518204580573e-05,
+      "loss": 0.4226,
+      "step": 4737
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4892175582107884e-05,
+      "loss": 1.0507,
+      "step": 4738
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.487383365427311e-05,
+      "loss": 0.744,
+      "step": 4739
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4855492423570625e-05,
+      "loss": 1.4508,
+      "step": 4740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.483715189249474e-05,
+      "loss": 1.3041,
+      "step": 4741
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.481881206353966e-05,
+      "loss": 0.9342,
+      "step": 4742
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.480047293919953e-05,
+      "loss": 0.893,
+      "step": 4743
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4782134521968386e-05,
+      "loss": 1.2379,
+      "step": 4744
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4763796814340134e-05,
+      "loss": 0.5318,
+      "step": 4745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4745459818808605e-05,
+      "loss": 1.0782,
+      "step": 4746
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.472712353786756e-05,
+      "loss": 1.0814,
+      "step": 4747
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.470878797401063e-05,
+      "loss": 1.0558,
+      "step": 4748
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.469045312973135e-05,
+      "loss": 1.218,
+      "step": 4749
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.467211900752316e-05,
+      "loss": 0.6179,
+      "step": 4750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.465378560987944e-05,
+      "loss": 1.0796,
+      "step": 4751
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.46354529392934e-05,
+      "loss": 1.0017,
+      "step": 4752
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.461712099825819e-05,
+      "loss": 1.3876,
+      "step": 4753
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.459878978926688e-05,
+      "loss": 1.4349,
+      "step": 4754
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.458045931481243e-05,
+      "loss": 1.3846,
+      "step": 4755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.456212957738766e-05,
+      "loss": 1.0567,
+      "step": 4756
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.454380057948533e-05,
+      "loss": 1.1759,
+      "step": 4757
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4525472323598105e-05,
+      "loss": 0.4834,
+      "step": 4758
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.450714481221851e-05,
+      "loss": 0.6392,
+      "step": 4759
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.448881804783901e-05,
+      "loss": 1.0646,
+      "step": 4760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4470492032951925e-05,
+      "loss": 1.0838,
+      "step": 4761
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.445216677004955e-05,
+      "loss": 0.8275,
+      "step": 4762
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4433842261623984e-05,
+      "loss": 1.2446,
+      "step": 4763
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.441551851016728e-05,
+      "loss": 1.1085,
+      "step": 4764
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.439719551817136e-05,
+      "loss": 0.9248,
+      "step": 4765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.437887328812807e-05,
+      "loss": 1.0478,
+      "step": 4766
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.436055182252914e-05,
+      "loss": 1.3134,
+      "step": 4767
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4342231123866204e-05,
+      "loss": 0.746,
+      "step": 4768
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.432391119463075e-05,
+      "loss": 0.5544,
+      "step": 4769
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4305592037314223e-05,
+      "loss": 1.1559,
+      "step": 4770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.428727365440794e-05,
+      "loss": 1.1751,
+      "step": 4771
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.426895604840307e-05,
+      "loss": 1.3366,
+      "step": 4772
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4250639221790766e-05,
+      "loss": 1.1111,
+      "step": 4773
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4232323177061975e-05,
+      "loss": 1.0636,
+      "step": 4774
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4214007916707614e-05,
+      "loss": 1.3127,
+      "step": 4775
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.419569344321844e-05,
+      "loss": 1.1897,
+      "step": 4776
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.417737975908515e-05,
+      "loss": 1.2747,
+      "step": 4777
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.415906686679831e-05,
+      "loss": 0.6207,
+      "step": 4778
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.414075476884836e-05,
+      "loss": 0.5507,
+      "step": 4779
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.412244346772566e-05,
+      "loss": 1.2072,
+      "step": 4780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4104132965920456e-05,
+      "loss": 1.4924,
+      "step": 4781
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.40858232659229e-05,
+      "loss": 1.2161,
+      "step": 4782
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4067514370222986e-05,
+      "loss": 1.1702,
+      "step": 4783
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.404920628131063e-05,
+      "loss": 1.559,
+      "step": 4784
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.403089900167567e-05,
+      "loss": 1.6535,
+      "step": 4785
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.401259253380778e-05,
+      "loss": 1.5258,
+      "step": 4786
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.399428688019656e-05,
+      "loss": 1.5759,
+      "step": 4787
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3975982043331425e-05,
+      "loss": 1.3705,
+      "step": 4788
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.395767802570183e-05,
+      "loss": 1.271,
+      "step": 4789
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3939374829796976e-05,
+      "loss": 0.5882,
+      "step": 4790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3921072458106e-05,
+      "loss": 1.5334,
+      "step": 4791
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3902770913117964e-05,
+      "loss": 1.0776,
+      "step": 4792
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.388447019732175e-05,
+      "loss": 1.2479,
+      "step": 4793
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.386617031320618e-05,
+      "loss": 0.5271,
+      "step": 4794
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.384787126325991e-05,
+      "loss": 1.0391,
+      "step": 4795
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.382957304997159e-05,
+      "loss": 1.4511,
+      "step": 4796
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.381127567582961e-05,
+      "loss": 0.9642,
+      "step": 4797
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.379297914332236e-05,
+      "loss": 1.4146,
+      "step": 4798
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.377468345493804e-05,
+      "loss": 1.2346,
+      "step": 4799
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3756388613164806e-05,
+      "loss": 0.6716,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3738094620490656e-05,
+      "loss": 1.021,
+      "step": 4801
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.371980147940346e-05,
+      "loss": 0.9936,
+      "step": 4802
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.370150919239099e-05,
+      "loss": 0.8616,
+      "step": 4803
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.368321776194093e-05,
+      "loss": 1.2776,
+      "step": 4804
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.366492719054081e-05,
+      "loss": 1.6603,
+      "step": 4805
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3646637480678036e-05,
+      "loss": 0.6339,
+      "step": 4806
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.362834863483992e-05,
+      "loss": 0.9159,
+      "step": 4807
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3610060655513684e-05,
+      "loss": 1.2254,
+      "step": 4808
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3591773545186364e-05,
+      "loss": 0.6633,
+      "step": 4809
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.357348730634492e-05,
+      "loss": 1.521,
+      "step": 4810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3555201941476174e-05,
+      "loss": 0.7561,
+      "step": 4811
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.353691745306689e-05,
+      "loss": 1.3051,
+      "step": 4812
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.351863384360361e-05,
+      "loss": 1.3217,
+      "step": 4813
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.350035111557283e-05,
+      "loss": 1.4247,
+      "step": 4814
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.348206927146091e-05,
+      "loss": 0.9462,
+      "step": 4815
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.346378831375411e-05,
+      "loss": 1.1117,
+      "step": 4816
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.344550824493851e-05,
+      "loss": 1.5756,
+      "step": 4817
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3427229067500106e-05,
+      "loss": 1.3177,
+      "step": 4818
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.34089507839248e-05,
+      "loss": 1.3351,
+      "step": 4819
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3390673396698324e-05,
+      "loss": 1.1506,
+      "step": 4820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.337239690830632e-05,
+      "loss": 1.5438,
+      "step": 4821
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.335412132123427e-05,
+      "loss": 1.4308,
+      "step": 4822
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.33358466379676e-05,
+      "loss": 1.4954,
+      "step": 4823
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.331757286099155e-05,
+      "loss": 0.8534,
+      "step": 4824
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3299299992791276e-05,
+      "loss": 0.6226,
+      "step": 4825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.328102803585176e-05,
+      "loss": 1.2558,
+      "step": 4826
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.326275699265792e-05,
+      "loss": 0.7951,
+      "step": 4827
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.324448686569453e-05,
+      "loss": 1.5696,
+      "step": 4828
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.322621765744622e-05,
+      "loss": 1.3084,
+      "step": 4829
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.32079493703975e-05,
+      "loss": 1.2254,
+      "step": 4830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.318968200703278e-05,
+      "loss": 1.2118,
+      "step": 4831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3171415569836346e-05,
+      "loss": 1.4335,
+      "step": 4832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.315315006129228e-05,
+      "loss": 0.5936,
+      "step": 4833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.313488548388466e-05,
+      "loss": 0.9831,
+      "step": 4834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3116621840097344e-05,
+      "loss": 1.4209,
+      "step": 4835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3098359132414094e-05,
+      "loss": 1.0044,
+      "step": 4836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.308009736331854e-05,
+      "loss": 1.5043,
+      "step": 4837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.30618365352942e-05,
+      "loss": 1.0372,
+      "step": 4838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.304357665082447e-05,
+      "loss": 1.1469,
+      "step": 4839
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3025317712392564e-05,
+      "loss": 1.6322,
+      "step": 4840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3007059722481616e-05,
+      "loss": 1.4066,
+      "step": 4841
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.298880268357463e-05,
+      "loss": 0.4583,
+      "step": 4842
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.297054659815447e-05,
+      "loss": 0.8277,
+      "step": 4843
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.295229146870385e-05,
+      "loss": 0.4798,
+      "step": 4844
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.293403729770539e-05,
+      "loss": 0.5537,
+      "step": 4845
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.291578408764157e-05,
+      "loss": 0.6656,
+      "step": 4846
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.289753184099471e-05,
+      "loss": 1.3829,
+      "step": 4847
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2879280560247046e-05,
+      "loss": 1.1435,
+      "step": 4848
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.286103024788063e-05,
+      "loss": 1.3124,
+      "step": 4849
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.284278090637744e-05,
+      "loss": 1.4012,
+      "step": 4850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2824532538219274e-05,
+      "loss": 1.3091,
+      "step": 4851
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.280628514588784e-05,
+      "loss": 1.1944,
+      "step": 4852
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.278803873186465e-05,
+      "loss": 1.6073,
+      "step": 4853
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.276979329863115e-05,
+      "loss": 1.2063,
+      "step": 4854
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.275154884866864e-05,
+      "loss": 1.2497,
+      "step": 4855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.273330538445822e-05,
+      "loss": 1.2186,
+      "step": 4856
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.271506290848095e-05,
+      "loss": 0.5721,
+      "step": 4857
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2696821423217706e-05,
+      "loss": 1.109,
+      "step": 4858
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.267858093114925e-05,
+      "loss": 1.7038,
+      "step": 4859
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.266034143475615e-05,
+      "loss": 1.085,
+      "step": 4860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.264210293651892e-05,
+      "loss": 1.4616,
+      "step": 4861
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2623865438917906e-05,
+      "loss": 1.0552,
+      "step": 4862
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.260562894443329e-05,
+      "loss": 1.1931,
+      "step": 4863
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.258739345554514e-05,
+      "loss": 1.5394,
+      "step": 4864
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2569158974733414e-05,
+      "loss": 1.549,
+      "step": 4865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.25509255044779e-05,
+      "loss": 1.2716,
+      "step": 4866
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.253269304725825e-05,
+      "loss": 1.2376,
+      "step": 4867
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.251446160555398e-05,
+      "loss": 1.4034,
+      "step": 4868
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2496231181844495e-05,
+      "loss": 1.07,
+      "step": 4869
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.247800177860901e-05,
+      "loss": 0.7162,
+      "step": 4870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.245977339832665e-05,
+      "loss": 0.8157,
+      "step": 4871
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.244154604347634e-05,
+      "loss": 1.7536,
+      "step": 4872
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.242331971653699e-05,
+      "loss": 1.2564,
+      "step": 4873
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.240509441998721e-05,
+      "loss": 1.0978,
+      "step": 4874
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.238687015630557e-05,
+      "loss": 1.0815,
+      "step": 4875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.23686469279705e-05,
+      "loss": 1.3765,
+      "step": 4876
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.235042473746023e-05,
+      "loss": 1.3294,
+      "step": 4877
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.233220358725292e-05,
+      "loss": 1.2652,
+      "step": 4878
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2313983479826494e-05,
+      "loss": 1.3851,
+      "step": 4879
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2295764417658865e-05,
+      "loss": 1.1572,
+      "step": 4880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.22775464032277e-05,
+      "loss": 1.3734,
+      "step": 4881
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.225932943901055e-05,
+      "loss": 1.4388,
+      "step": 4882
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2241113527484815e-05,
+      "loss": 1.1533,
+      "step": 4883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.22228986711278e-05,
+      "loss": 0.9878,
+      "step": 4884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2204684872416616e-05,
+      "loss": 1.4409,
+      "step": 4885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.218647213382824e-05,
+      "loss": 0.5453,
+      "step": 4886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.21682604578395e-05,
+      "loss": 1.2478,
+      "step": 4887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.215004984692713e-05,
+      "loss": 1.4781,
+      "step": 4888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.213184030356766e-05,
+      "loss": 1.2795,
+      "step": 4889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.211363183023748e-05,
+      "loss": 1.1832,
+      "step": 4890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.209542442941284e-05,
+      "loss": 0.9732,
+      "step": 4891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2077218103569903e-05,
+      "loss": 0.96,
+      "step": 4892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.205901285518459e-05,
+      "loss": 0.5798,
+      "step": 4893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2040808686732735e-05,
+      "loss": 1.4781,
+      "step": 4894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.202260560069001e-05,
+      "loss": 1.2897,
+      "step": 4895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2004403599531955e-05,
+      "loss": 0.7163,
+      "step": 4896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1986202685733934e-05,
+      "loss": 1.9143,
+      "step": 4897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.196800286177116e-05,
+      "loss": 1.5545,
+      "step": 4898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.194980413011877e-05,
+      "loss": 0.8521,
+      "step": 4899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.193160649325165e-05,
+      "loss": 0.7964,
+      "step": 4900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.19134099536446e-05,
+      "loss": 1.1906,
+      "step": 4901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.189521451377224e-05,
+      "loss": 1.128,
+      "step": 4902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.18770201761091e-05,
+      "loss": 1.3912,
+      "step": 4903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1858826943129476e-05,
+      "loss": 0.8014,
+      "step": 4904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.184063481730758e-05,
+      "loss": 1.3521,
+      "step": 4905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1822443801117404e-05,
+      "loss": 0.9973,
+      "step": 4906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.180425389703288e-05,
+      "loss": 1.2585,
+      "step": 4907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.178606510752773e-05,
+      "loss": 1.1156,
+      "step": 4908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.176787743507553e-05,
+      "loss": 0.9387,
+      "step": 4909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.17496908821497e-05,
+      "loss": 1.3581,
+      "step": 4910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.173150545122354e-05,
+      "loss": 1.333,
+      "step": 4911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1713321144770165e-05,
+      "loss": 0.5737,
+      "step": 4912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1695137965262546e-05,
+      "loss": 0.6538,
+      "step": 4913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.167695591517348e-05,
+      "loss": 1.331,
+      "step": 4914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.165877499697567e-05,
+      "loss": 0.4757,
+      "step": 4915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.164059521314162e-05,
+      "loss": 1.2089,
+      "step": 4916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1622416566143676e-05,
+      "loss": 1.1074,
+      "step": 4917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1604239058454025e-05,
+      "loss": 1.5356,
+      "step": 4918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.158606269254476e-05,
+      "loss": 0.8268,
+      "step": 4919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.156788747088773e-05,
+      "loss": 0.717,
+      "step": 4920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.154971339595468e-05,
+      "loss": 1.352,
+      "step": 4921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.153154047021721e-05,
+      "loss": 0.4551,
+      "step": 4922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.151336869614675e-05,
+      "loss": 1.3475,
+      "step": 4923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.149519807621454e-05,
+      "loss": 1.7364,
+      "step": 4924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.147702861289168e-05,
+      "loss": 1.0994,
+      "step": 4925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1458860308649185e-05,
+      "loss": 0.8961,
+      "step": 4926
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.144069316595779e-05,
+      "loss": 0.6832,
+      "step": 4927
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.142252718728816e-05,
+      "loss": 1.4249,
+      "step": 4928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.140436237511076e-05,
+      "loss": 0.9905,
+      "step": 4929
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.138619873189594e-05,
+      "loss": 0.6844,
+      "step": 4930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1368036260113836e-05,
+      "loss": 1.2351,
+      "step": 4931
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.134987496223447e-05,
+      "loss": 1.3542,
+      "step": 4932
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.133171484072765e-05,
+      "loss": 1.2022,
+      "step": 4933
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.131355589806309e-05,
+      "loss": 1.2951,
+      "step": 4934
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.129539813671031e-05,
+      "loss": 0.9676,
+      "step": 4935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.127724155913868e-05,
+      "loss": 1.3075,
+      "step": 4936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.125908616781737e-05,
+      "loss": 1.4452,
+      "step": 4937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.124093196521545e-05,
+      "loss": 0.6352,
+      "step": 4938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.122277895380179e-05,
+      "loss": 0.8885,
+      "step": 4939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.120462713604509e-05,
+      "loss": 1.2384,
+      "step": 4940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.118647651441392e-05,
+      "loss": 1.4193,
+      "step": 4941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.116832709137668e-05,
+      "loss": 0.8706,
+      "step": 4942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1150178869401594e-05,
+      "loss": 1.3566,
+      "step": 4943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.11320318509567e-05,
+      "loss": 1.1168,
+      "step": 4944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.111388603850993e-05,
+      "loss": 1.2528,
+      "step": 4945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1095741434529026e-05,
+      "loss": 1.2226,
+      "step": 4946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.107759804148152e-05,
+      "loss": 0.5146,
+      "step": 4947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.105945586183485e-05,
+      "loss": 1.4323,
+      "step": 4948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.104131489805627e-05,
+      "loss": 1.29,
+      "step": 4949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.102317515261284e-05,
+      "loss": 1.3915,
+      "step": 4950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1005036627971474e-05,
+      "loss": 1.4116,
+      "step": 4951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.098689932659891e-05,
+      "loss": 1.1798,
+      "step": 4952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.096876325096175e-05,
+      "loss": 1.6919,
+      "step": 4953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.095062840352639e-05,
+      "loss": 1.4859,
+      "step": 4954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.09324947867591e-05,
+      "loss": 1.382,
+      "step": 4955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0914362403125895e-05,
+      "loss": 1.3698,
+      "step": 4956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.089623125509276e-05,
+      "loss": 0.3523,
+      "step": 4957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.087810134512541e-05,
+      "loss": 0.8164,
+      "step": 4958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0859972675689426e-05,
+      "loss": 1.0639,
+      "step": 4959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.084184524925018e-05,
+      "loss": 1.2723,
+      "step": 4960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.082371906827295e-05,
+      "loss": 1.3501,
+      "step": 4961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.08055941352228e-05,
+      "loss": 1.524,
+      "step": 4962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.078747045256458e-05,
+      "loss": 1.3139,
+      "step": 4963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.076934802276309e-05,
+      "loss": 1.0828,
+      "step": 4964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.075122684828283e-05,
+      "loss": 0.5631,
+      "step": 4965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.073310693158823e-05,
+      "loss": 0.6787,
+      "step": 4966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.071498827514346e-05,
+      "loss": 1.1845,
+      "step": 4967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0696870881412596e-05,
+      "loss": 0.8934,
+      "step": 4968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0678754752859514e-05,
+      "loss": 1.2468,
+      "step": 4969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0660639891947894e-05,
+      "loss": 0.899,
+      "step": 4970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.064252630114126e-05,
+      "loss": 1.0267,
+      "step": 4971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0624413982902994e-05,
+      "loss": 1.4274,
+      "step": 4972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0606302939696275e-05,
+      "loss": 0.8529,
+      "step": 4973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.058819317398408e-05,
+      "loss": 0.5372,
+      "step": 4974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057008468822927e-05,
+      "loss": 1.1927,
+      "step": 4975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055197748489451e-05,
+      "loss": 1.3838,
+      "step": 4976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0533871566442274e-05,
+      "loss": 1.3511,
+      "step": 4977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0515766935334875e-05,
+      "loss": 0.772,
+      "step": 4978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.049766359403444e-05,
+      "loss": 0.7044,
+      "step": 4979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.047956154500297e-05,
+      "loss": 1.1746,
+      "step": 4980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.046146079070221e-05,
+      "loss": 0.7317,
+      "step": 4981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.044336133359377e-05,
+      "loss": 1.3826,
+      "step": 4982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0425263176139125e-05,
+      "loss": 1.1254,
+      "step": 4983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.040716632079948e-05,
+      "loss": 1.1712,
+      "step": 4984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.038907077003595e-05,
+      "loss": 1.1238,
+      "step": 4985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0370976526309415e-05,
+      "loss": 1.2892,
+      "step": 4986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.035288359208062e-05,
+      "loss": 1.1803,
+      "step": 4987
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0334791969810096e-05,
+      "loss": 0.5157,
+      "step": 4988
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0316701661958216e-05,
+      "loss": 1.0555,
+      "step": 4989
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.029861267098515e-05,
+      "loss": 1.3888,
+      "step": 4990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.028052499935094e-05,
+      "loss": 0.814,
+      "step": 4991
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.02624386495154e-05,
+      "loss": 1.4899,
+      "step": 4992
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.024435362393819e-05,
+      "loss": 0.6686,
+      "step": 4993
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.022626992507875e-05,
+      "loss": 0.7039,
+      "step": 4994
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.020818755539641e-05,
+      "loss": 1.2283,
+      "step": 4995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0190106517350265e-05,
+      "loss": 1.5009,
+      "step": 4996
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.017202681339924e-05,
+      "loss": 1.1983,
+      "step": 4997
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.015394844600207e-05,
+      "loss": 1.206,
+      "step": 4998
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0135871417617333e-05,
+      "loss": 0.7918,
+      "step": 4999
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.011779573070343e-05,
+      "loss": 0.9825,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.009972138771853e-05,
+      "loss": 0.7903,
+      "step": 5001
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.008164839112066e-05,
+      "loss": 1.3486,
+      "step": 5002
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.006357674336767e-05,
+      "loss": 1.7001,
+      "step": 5003
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.004550644691719e-05,
+      "loss": 0.5668,
+      "step": 5004
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0027437504226696e-05,
+      "loss": 0.5828,
+      "step": 5005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0009369917753475e-05,
+      "loss": 0.9508,
+      "step": 5006
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.999130368995464e-05,
+      "loss": 0.576,
+      "step": 5007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.997323882328707e-05,
+      "loss": 0.9176,
+      "step": 5008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.99551753202075e-05,
+      "loss": 0.6021,
+      "step": 5009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9937113183172525e-05,
+      "loss": 0.939,
+      "step": 5010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.991905241463844e-05,
+      "loss": 1.2073,
+      "step": 5011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.990099301706146e-05,
+      "loss": 1.4248,
+      "step": 5012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.988293499289751e-05,
+      "loss": 1.201,
+      "step": 5013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.986487834460247e-05,
+      "loss": 1.4382,
+      "step": 5014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.98468230746319e-05,
+      "loss": 1.493,
+      "step": 5015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.982876918544124e-05,
+      "loss": 0.7555,
+      "step": 5016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.98107166794857e-05,
+      "loss": 1.35,
+      "step": 5017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9792665559220375e-05,
+      "loss": 1.3794,
+      "step": 5018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.977461582710009e-05,
+      "loss": 1.2761,
+      "step": 5019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.975656748557953e-05,
+      "loss": 1.1122,
+      "step": 5020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.975656748557953e-05,
+      "loss": 1.0455,
+      "step": 5021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9738520537113154e-05,
+      "loss": 1.3056,
+      "step": 5022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.972047498415528e-05,
+      "loss": 0.4674,
+      "step": 5023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.970243082916002e-05,
+      "loss": 1.0636,
+      "step": 5024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.968438807458124e-05,
+      "loss": 1.0886,
+      "step": 5025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.966634672287271e-05,
+      "loss": 1.1188,
+      "step": 5026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.964830677648794e-05,
+      "loss": 1.2351,
+      "step": 5027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.963026823788028e-05,
+      "loss": 0.5136,
+      "step": 5028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.961223110950285e-05,
+      "loss": 0.6237,
+      "step": 5029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.959419539380864e-05,
+      "loss": 1.068,
+      "step": 5030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.957616109325041e-05,
+      "loss": 1.297,
+      "step": 5031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9558128210280706e-05,
+      "loss": 1.3044,
+      "step": 5032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.954009674735193e-05,
+      "loss": 0.813,
+      "step": 5033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9522066706916275e-05,
+      "loss": 1.2238,
+      "step": 5034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.950403809142572e-05,
+      "loss": 1.066,
+      "step": 5035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9486010903332065e-05,
+      "loss": 0.8515,
+      "step": 5036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9467985145086906e-05,
+      "loss": 1.1829,
+      "step": 5037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9449960819141684e-05,
+      "loss": 1.1666,
+      "step": 5038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.943193792794759e-05,
+      "loss": 1.1158,
+      "step": 5039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.941391647395566e-05,
+      "loss": 0.6626,
+      "step": 5040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.939589645961667e-05,
+      "loss": 1.1247,
+      "step": 5041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9377877887381326e-05,
+      "loss": 1.5809,
+      "step": 5042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9359860759700015e-05,
+      "loss": 0.6134,
+      "step": 5043
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.934184507902299e-05,
+      "loss": 1.1373,
+      "step": 5044
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9323830847800264e-05,
+      "loss": 1.3219,
+      "step": 5045
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.930581806848171e-05,
+      "loss": 1.5141,
+      "step": 5046
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.928780674351697e-05,
+      "loss": 0.8268,
+      "step": 5047
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9269796875355456e-05,
+      "loss": 0.8816,
+      "step": 5048
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.925178846644648e-05,
+      "loss": 1.4013,
+      "step": 5049
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.923378151923904e-05,
+      "loss": 0.7437,
+      "step": 5050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.921577603618203e-05,
+      "loss": 1.4545,
+      "step": 5051
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.919777201972405e-05,
+      "loss": 1.1006,
+      "step": 5052
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9179769472313604e-05,
+      "loss": 1.0308,
+      "step": 5053
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.916176839639893e-05,
+      "loss": 1.2627,
+      "step": 5054
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9143768794428076e-05,
+      "loss": 1.251,
+      "step": 5055
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.912577066884888e-05,
+      "loss": 1.3188,
+      "step": 5056
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.910777402210903e-05,
+      "loss": 1.0584,
+      "step": 5057
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.908977885665596e-05,
+      "loss": 0.6597,
+      "step": 5058
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.907178517493692e-05,
+      "loss": 1.2819,
+      "step": 5059
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.905379297939894e-05,
+      "loss": 1.732,
+      "step": 5060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.903580227248891e-05,
+      "loss": 0.5082,
+      "step": 5061
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.901781305665344e-05,
+      "loss": 1.0149,
+      "step": 5062
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8999825334338975e-05,
+      "loss": 1.0775,
+      "step": 5063
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.898183910799175e-05,
+      "loss": 1.2557,
+      "step": 5064
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.896385438005782e-05,
+      "loss": 1.2387,
+      "step": 5065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8945871152983e-05,
+      "loss": 1.3525,
+      "step": 5066
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8927889429212906e-05,
+      "loss": 1.1238,
+      "step": 5067
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8909909211192984e-05,
+      "loss": 1.0139,
+      "step": 5068
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.889193050136844e-05,
+      "loss": 0.8451,
+      "step": 5069
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.887395330218429e-05,
+      "loss": 0.9434,
+      "step": 5070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8855977616085314e-05,
+      "loss": 0.8296,
+      "step": 5071
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.883800344551616e-05,
+      "loss": 0.647,
+      "step": 5072
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.882003079292119e-05,
+      "loss": 0.7851,
+      "step": 5073
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.880205966074462e-05,
+      "loss": 1.3962,
+      "step": 5074
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8784090051430376e-05,
+      "loss": 1.3925,
+      "step": 5075
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8766121967422295e-05,
+      "loss": 0.5219,
+      "step": 5076
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8748155411163914e-05,
+      "loss": 1.3881,
+      "step": 5077
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.87301903850986e-05,
+      "loss": 0.3855,
+      "step": 5078
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8712226891669486e-05,
+      "loss": 1.6521,
+      "step": 5079
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.869426493331954e-05,
+      "loss": 1.0296,
+      "step": 5080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.867630451249149e-05,
+      "loss": 0.8411,
+      "step": 5081
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.865834563162785e-05,
+      "loss": 1.4324,
+      "step": 5082
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8640388293170935e-05,
+      "loss": 1.1896,
+      "step": 5083
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.862243249956287e-05,
+      "loss": 1.4476,
+      "step": 5084
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8604478253245544e-05,
+      "loss": 1.1032,
+      "step": 5085
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.858652555666063e-05,
+      "loss": 1.3667,
+      "step": 5086
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.85685744122496e-05,
+      "loss": 0.6216,
+      "step": 5087
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.855062482245375e-05,
+      "loss": 0.7626,
+      "step": 5088
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.853267678971409e-05,
+      "loss": 0.7115,
+      "step": 5089
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.851473031647147e-05,
+      "loss": 1.1091,
+      "step": 5090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.849678540516655e-05,
+      "loss": 0.7552,
+      "step": 5091
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8478842058239725e-05,
+      "loss": 1.0883,
+      "step": 5092
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8460900278131186e-05,
+      "loss": 1.2422,
+      "step": 5093
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.844296006728092e-05,
+      "loss": 1.2352,
+      "step": 5094
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.842502142812874e-05,
+      "loss": 1.2302,
+      "step": 5095
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.840708436311418e-05,
+      "loss": 1.0319,
+      "step": 5096
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.83891488746766e-05,
+      "loss": 1.1718,
+      "step": 5097
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.83712149652551e-05,
+      "loss": 1.3405,
+      "step": 5098
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.835328263728866e-05,
+      "loss": 0.6433,
+      "step": 5099
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.833535189321593e-05,
+      "loss": 1.1428,
+      "step": 5100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.831742273547544e-05,
+      "loss": 1.2216,
+      "step": 5101
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.829949516650543e-05,
+      "loss": 0.6861,
+      "step": 5102
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8281569188743984e-05,
+      "loss": 1.7443,
+      "step": 5103
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.826364480462894e-05,
+      "loss": 0.5702,
+      "step": 5104
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.824572201659789e-05,
+      "loss": 1.2911,
+      "step": 5105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.822780082708827e-05,
+      "loss": 1.1184,
+      "step": 5106
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.820988123853726e-05,
+      "loss": 1.2824,
+      "step": 5107
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.819196325338186e-05,
+      "loss": 0.8171,
+      "step": 5108
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.817404687405878e-05,
+      "loss": 0.6563,
+      "step": 5109
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8156132103004566e-05,
+      "loss": 1.1605,
+      "step": 5110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.813821894265557e-05,
+      "loss": 1.1823,
+      "step": 5111
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.812030739544785e-05,
+      "loss": 1.0812,
+      "step": 5112
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.810239746381729e-05,
+      "loss": 1.1114,
+      "step": 5113
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.808448915019957e-05,
+      "loss": 1.6201,
+      "step": 5114
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.806658245703013e-05,
+      "loss": 1.304,
+      "step": 5115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8048677386744155e-05,
+      "loss": 1.2936,
+      "step": 5116
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.803077394177665e-05,
+      "loss": 1.2859,
+      "step": 5117
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.801287212456243e-05,
+      "loss": 1.4887,
+      "step": 5118
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.799497193753602e-05,
+      "loss": 1.5586,
+      "step": 5119
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7977073383131744e-05,
+      "loss": 1.3339,
+      "step": 5120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7959176463783724e-05,
+      "loss": 1.1214,
+      "step": 5121
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.794128118192587e-05,
+      "loss": 0.6608,
+      "step": 5122
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.792338753999182e-05,
+      "loss": 1.1975,
+      "step": 5123
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7905495540415034e-05,
+      "loss": 0.7641,
+      "step": 5124
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.78876051856287e-05,
+      "loss": 1.2044,
+      "step": 5125
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.786971647806587e-05,
+      "loss": 1.1338,
+      "step": 5126
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.785182942015927e-05,
+      "loss": 0.7458,
+      "step": 5127
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7833944014341485e-05,
+      "loss": 1.484,
+      "step": 5128
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.781606026304479e-05,
+      "loss": 1.1428,
+      "step": 5129
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.779817816870132e-05,
+      "loss": 1.4123,
+      "step": 5130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7780297733742955e-05,
+      "loss": 1.1795,
+      "step": 5131
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.776241896060129e-05,
+      "loss": 0.4529,
+      "step": 5132
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7744541851707806e-05,
+      "loss": 1.0978,
+      "step": 5133
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.772666640949368e-05,
+      "loss": 0.9377,
+      "step": 5134
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.770879263638988e-05,
+      "loss": 1.2901,
+      "step": 5135
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.769092053482711e-05,
+      "loss": 1.2302,
+      "step": 5136
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.767305010723594e-05,
+      "loss": 0.7384,
+      "step": 5137
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.765518135604664e-05,
+      "loss": 0.8874,
+      "step": 5138
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.763731428368925e-05,
+      "loss": 1.5805,
+      "step": 5139
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.761944889259359e-05,
+      "loss": 0.9485,
+      "step": 5140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.76015851851893e-05,
+      "loss": 0.8414,
+      "step": 5141
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.758372316390575e-05,
+      "loss": 0.5672,
+      "step": 5142
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.756586283117205e-05,
+      "loss": 1.4972,
+      "step": 5143
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.754800418941713e-05,
+      "loss": 0.341,
+      "step": 5144
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.753014724106969e-05,
+      "loss": 1.1646,
+      "step": 5145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.751229198855816e-05,
+      "loss": 0.9748,
+      "step": 5146
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7494438434310765e-05,
+      "loss": 1.6186,
+      "step": 5147
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.747658658075549e-05,
+      "loss": 0.948,
+      "step": 5148
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.745873643032013e-05,
+      "loss": 1.0635,
+      "step": 5149
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.744088798543219e-05,
+      "loss": 0.9483,
+      "step": 5150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7423041248518975e-05,
+      "loss": 1.2252,
+      "step": 5151
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.740519622200751e-05,
+      "loss": 1.3208,
+      "step": 5152
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.738735290832469e-05,
+      "loss": 1.1122,
+      "step": 5153
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.736951130989707e-05,
+      "loss": 1.1684,
+      "step": 5154
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.735167142915102e-05,
+      "loss": 1.3968,
+      "step": 5155
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7333833268512706e-05,
+      "loss": 1.2034,
+      "step": 5156
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7315996830408e-05,
+      "loss": 1.0609,
+      "step": 5157
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.729816211726256e-05,
+      "loss": 0.6451,
+      "step": 5158
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.728032913150181e-05,
+      "loss": 0.5841,
+      "step": 5159
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.726249787555097e-05,
+      "loss": 0.9117,
+      "step": 5160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.724466835183499e-05,
+      "loss": 1.3857,
+      "step": 5161
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.72268405627786e-05,
+      "loss": 1.1123,
+      "step": 5162
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.720901451080625e-05,
+      "loss": 1.5152,
+      "step": 5163
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.719119019834224e-05,
+      "loss": 1.4907,
+      "step": 5164
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.717336762781056e-05,
+      "loss": 1.0656,
+      "step": 5165
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7155546801634996e-05,
+      "loss": 0.6751,
+      "step": 5166
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.713772772223908e-05,
+      "loss": 1.4469,
+      "step": 5167
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7119910392046114e-05,
+      "loss": 1.0929,
+      "step": 5168
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7102094813479196e-05,
+      "loss": 1.2086,
+      "step": 5169
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.708428098896111e-05,
+      "loss": 1.294,
+      "step": 5170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.706646892091444e-05,
+      "loss": 0.6866,
+      "step": 5171
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.704865861176159e-05,
+      "loss": 1.3055,
+      "step": 5172
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.703085006392463e-05,
+      "loss": 1.3111,
+      "step": 5173
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.701304327982543e-05,
+      "loss": 1.2357,
+      "step": 5174
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.699523826188563e-05,
+      "loss": 1.4561,
+      "step": 5175
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.697743501252664e-05,
+      "loss": 0.89,
+      "step": 5176
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.695963353416958e-05,
+      "loss": 0.7684,
+      "step": 5177
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6941833829235365e-05,
+      "loss": 1.2431,
+      "step": 5178
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.692403590014469e-05,
+      "loss": 1.4809,
+      "step": 5179
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.690623974931794e-05,
+      "loss": 0.9684,
+      "step": 5180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.688844537917534e-05,
+      "loss": 1.2319,
+      "step": 5181
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.687065279213679e-05,
+      "loss": 0.6574,
+      "step": 5182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.685286199062205e-05,
+      "loss": 0.5864,
+      "step": 5183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.683507297705051e-05,
+      "loss": 0.8034,
+      "step": 5184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6817285753841444e-05,
+      "loss": 1.1024,
+      "step": 5185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.679950032341377e-05,
+      "loss": 0.7123,
+      "step": 5186
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.678171668818626e-05,
+      "loss": 1.3218,
+      "step": 5187
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.676393485057739e-05,
+      "loss": 1.1191,
+      "step": 5188
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.674615481300536e-05,
+      "loss": 1.4931,
+      "step": 5189
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.672837657788819e-05,
+      "loss": 1.4916,
+      "step": 5190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.671060014764365e-05,
+      "loss": 1.099,
+      "step": 5191
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.669282552468922e-05,
+      "loss": 0.4726,
+      "step": 5192
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.667505271144215e-05,
+      "loss": 0.7421,
+      "step": 5193
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.665728171031946e-05,
+      "loss": 0.9417,
+      "step": 5194
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.663951252373793e-05,
+      "loss": 1.1308,
+      "step": 5195
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.662174515411405e-05,
+      "loss": 0.4321,
+      "step": 5196
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.66039796038641e-05,
+      "loss": 0.5089,
+      "step": 5197
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6586215875404126e-05,
+      "loss": 1.1064,
+      "step": 5198
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.656845397114989e-05,
+      "loss": 1.5457,
+      "step": 5199
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6550693893516906e-05,
+      "loss": 1.2486,
+      "step": 5200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.653293564492046e-05,
+      "loss": 0.6568,
+      "step": 5201
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.65151792277756e-05,
+      "loss": 0.9795,
+      "step": 5202
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.649742464449708e-05,
+      "loss": 1.342,
+      "step": 5203
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.647967189749946e-05,
+      "loss": 1.5239,
+      "step": 5204
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6461920989196984e-05,
+      "loss": 0.5704,
+      "step": 5205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.644417192200374e-05,
+      "loss": 0.5042,
+      "step": 5206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.642642469833346e-05,
+      "loss": 1.1679,
+      "step": 5207
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.640867932059971e-05,
+      "loss": 1.389,
+      "step": 5208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.639093579121573e-05,
+      "loss": 1.5456,
+      "step": 5209
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.637319411259459e-05,
+      "loss": 1.1145,
+      "step": 5210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.635545428714905e-05,
+      "loss": 0.7613,
+      "step": 5211
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.633771631729164e-05,
+      "loss": 1.3319,
+      "step": 5212
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.631998020543461e-05,
+      "loss": 0.4685,
+      "step": 5213
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.630224595399e-05,
+      "loss": 0.7462,
+      "step": 5214
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.62845135653696e-05,
+      "loss": 1.2926,
+      "step": 5215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6266783041984865e-05,
+      "loss": 1.4646,
+      "step": 5216
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.624905438624712e-05,
+      "loss": 1.5472,
+      "step": 5217
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.623132760056732e-05,
+      "loss": 1.4408,
+      "step": 5218
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.621360268735625e-05,
+      "loss": 1.1062,
+      "step": 5219
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.619587964902438e-05,
+      "loss": 1.6085,
+      "step": 5220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.617815848798198e-05,
+      "loss": 0.6062,
+      "step": 5221
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.616043920663903e-05,
+      "loss": 1.7851,
+      "step": 5222
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6142721807405235e-05,
+      "loss": 1.113,
+      "step": 5223
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.612500629269009e-05,
+      "loss": 1.0877,
+      "step": 5224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.610729266490282e-05,
+      "loss": 1.8972,
+      "step": 5225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6089580926452396e-05,
+      "loss": 1.2619,
+      "step": 5226
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.607187107974749e-05,
+      "loss": 1.4421,
+      "step": 5227
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.605416312719657e-05,
+      "loss": 0.8618,
+      "step": 5228
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.603645707120784e-05,
+      "loss": 0.4997,
+      "step": 5229
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.60187529141892e-05,
+      "loss": 1.3449,
+      "step": 5230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.600105065854835e-05,
+      "loss": 1.4008,
+      "step": 5231
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5983350306692695e-05,
+      "loss": 1.1034,
+      "step": 5232
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.596565186102941e-05,
+      "loss": 1.4264,
+      "step": 5233
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.594795532396537e-05,
+      "loss": 1.2976,
+      "step": 5234
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.593026069790723e-05,
+      "loss": 1.3983,
+      "step": 5235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5912567985261333e-05,
+      "loss": 0.6845,
+      "step": 5236
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5894877188433834e-05,
+      "loss": 1.4169,
+      "step": 5237
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.587718830983058e-05,
+      "loss": 1.1751,
+      "step": 5238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.585950135185716e-05,
+      "loss": 1.4781,
+      "step": 5239
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.584181631691894e-05,
+      "loss": 1.422,
+      "step": 5240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5824133207420966e-05,
+      "loss": 1.0637,
+      "step": 5241
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5806452025768056e-05,
+      "loss": 1.1561,
+      "step": 5242
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.578877277436474e-05,
+      "loss": 0.7072,
+      "step": 5243
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5771095455615336e-05,
+      "loss": 1.4335,
+      "step": 5244
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.575342007192385e-05,
+      "loss": 0.8795,
+      "step": 5245
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5735746625694066e-05,
+      "loss": 0.6875,
+      "step": 5246
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5718075119329444e-05,
+      "loss": 0.6308,
+      "step": 5247
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.570040555523325e-05,
+      "loss": 0.9887,
+      "step": 5248
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5682737935808455e-05,
+      "loss": 1.306,
+      "step": 5249
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.566507226345774e-05,
+      "loss": 0.7039,
+      "step": 5250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.564740854058356e-05,
+      "loss": 1.0454,
+      "step": 5251
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.562974676958809e-05,
+      "loss": 1.3325,
+      "step": 5252
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.561208695287325e-05,
+      "loss": 0.7111,
+      "step": 5253
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.559442909284067e-05,
+      "loss": 1.5107,
+      "step": 5254
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5576773191891724e-05,
+      "loss": 0.8386,
+      "step": 5255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.555911925242754e-05,
+      "loss": 1.248,
+      "step": 5256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.554146727684895e-05,
+      "loss": 1.0415,
+      "step": 5257
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.552381726755656e-05,
+      "loss": 1.4232,
+      "step": 5258
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.550616922695062e-05,
+      "loss": 1.3381,
+      "step": 5259
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.548852315743123e-05,
+      "loss": 1.2915,
+      "step": 5260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.547087906139814e-05,
+      "loss": 0.4711,
+      "step": 5261
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.545323694125086e-05,
+      "loss": 1.212,
+      "step": 5262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.543559679938865e-05,
+      "loss": 1.1139,
+      "step": 5263
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.541795863821045e-05,
+      "loss": 1.0714,
+      "step": 5264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5400322460114966e-05,
+      "loss": 1.5633,
+      "step": 5265
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.538268826750061e-05,
+      "loss": 1.3487,
+      "step": 5266
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5365056062765586e-05,
+      "loss": 0.8415,
+      "step": 5267
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.534742584830775e-05,
+      "loss": 1.2622,
+      "step": 5268
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.532979762652474e-05,
+      "loss": 0.8841,
+      "step": 5269
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.531217139981387e-05,
+      "loss": 1.0046,
+      "step": 5270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.529454717057224e-05,
+      "loss": 1.4221,
+      "step": 5271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.527692494119668e-05,
+      "loss": 0.9751,
+      "step": 5272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.525930471408367e-05,
+      "loss": 1.5726,
+      "step": 5273
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5241686491629497e-05,
+      "loss": 1.0814,
+      "step": 5274
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.522407027623016e-05,
+      "loss": 0.7719,
+      "step": 5275
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.520645607028137e-05,
+      "loss": 1.6212,
+      "step": 5276
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5188843876178547e-05,
+      "loss": 0.8371,
+      "step": 5277
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.517123369631686e-05,
+      "loss": 1.1187,
+      "step": 5278
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.515362553309125e-05,
+      "loss": 1.1923,
+      "step": 5279
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.513601938889628e-05,
+      "loss": 1.2437,
+      "step": 5280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.511841526612631e-05,
+      "loss": 1.4707,
+      "step": 5281
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.510081316717543e-05,
+      "loss": 1.5176,
+      "step": 5282
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.508321309443743e-05,
+      "loss": 1.0605,
+      "step": 5283
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.506561505030581e-05,
+      "loss": 1.9168,
+      "step": 5284
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5048019037173815e-05,
+      "loss": 1.076,
+      "step": 5285
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.503042505743445e-05,
+      "loss": 0.7859,
+      "step": 5286
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.501283311348036e-05,
+      "loss": 1.4167,
+      "step": 5287
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.499524320770399e-05,
+      "loss": 1.401,
+      "step": 5288
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.497765534249744e-05,
+      "loss": 1.376,
+      "step": 5289
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.496006952025262e-05,
+      "loss": 1.0884,
+      "step": 5290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.494248574336109e-05,
+      "loss": 0.9126,
+      "step": 5291
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.492490401421414e-05,
+      "loss": 1.0206,
+      "step": 5292
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.490732433520279e-05,
+      "loss": 1.1135,
+      "step": 5293
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4889746708717805e-05,
+      "loss": 1.426,
+      "step": 5294
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.487217113714966e-05,
+      "loss": 0.9764,
+      "step": 5295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.485459762288853e-05,
+      "loss": 0.7809,
+      "step": 5296
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.483702616832431e-05,
+      "loss": 0.4468,
+      "step": 5297
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.481945677584664e-05,
+      "loss": 1.2544,
+      "step": 5298
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.480188944784488e-05,
+      "loss": 1.1271,
+      "step": 5299
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.478432418670807e-05,
+      "loss": 1.0694,
+      "step": 5300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4766760994824995e-05,
+      "loss": 1.3778,
+      "step": 5301
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.474919987458419e-05,
+      "loss": 1.2629,
+      "step": 5302
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.473164082837387e-05,
+      "loss": 1.1701,
+      "step": 5303
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.471408385858194e-05,
+      "loss": 1.0361,
+      "step": 5304
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.469652896759609e-05,
+      "loss": 0.6362,
+      "step": 5305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.46789761578037e-05,
+      "loss": 1.4875,
+      "step": 5306
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4661425431591843e-05,
+      "loss": 1.2265,
+      "step": 5307
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.464387679134732e-05,
+      "loss": 1.3961,
+      "step": 5308
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.462633023945668e-05,
+      "loss": 1.2028,
+      "step": 5309
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.460878577830616e-05,
+      "loss": 1.3679,
+      "step": 5310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4591243410281705e-05,
+      "loss": 1.2605,
+      "step": 5311
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.457370313776897e-05,
+      "loss": 1.3287,
+      "step": 5312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.455616496315341e-05,
+      "loss": 1.3017,
+      "step": 5313
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.453862888882005e-05,
+      "loss": 1.4544,
+      "step": 5314
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.452109491715375e-05,
+      "loss": 1.4987,
+      "step": 5315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4503563050539e-05,
+      "loss": 0.7196,
+      "step": 5316
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4486033291360094e-05,
+      "loss": 0.994,
+      "step": 5317
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4468505642000956e-05,
+      "loss": 1.1883,
+      "step": 5318
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.445098010484528e-05,
+      "loss": 1.1028,
+      "step": 5319
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4433456682276414e-05,
+      "loss": 1.0311,
+      "step": 5320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.441593537667749e-05,
+      "loss": 1.487,
+      "step": 5321
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4398416190431305e-05,
+      "loss": 1.0062,
+      "step": 5322
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4380899125920344e-05,
+      "loss": 0.6221,
+      "step": 5323
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4363384185526894e-05,
+      "loss": 0.7513,
+      "step": 5324
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4345871371632866e-05,
+      "loss": 1.0233,
+      "step": 5325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.432836068661992e-05,
+      "loss": 1.3806,
+      "step": 5326
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4310852132869395e-05,
+      "loss": 1.0396,
+      "step": 5327
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4293345712762405e-05,
+      "loss": 0.989,
+      "step": 5328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.427584142867971e-05,
+      "loss": 0.9884,
+      "step": 5329
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4258339283001816e-05,
+      "loss": 1.3562,
+      "step": 5330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4240839278108895e-05,
+      "loss": 1.277,
+      "step": 5331
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4223341416380885e-05,
+      "loss": 1.2569,
+      "step": 5332
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.42058457001974e-05,
+      "loss": 1.0085,
+      "step": 5333
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.418835213193777e-05,
+      "loss": 1.0829,
+      "step": 5334
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.417086071398101e-05,
+      "loss": 1.1636,
+      "step": 5335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4153371448705897e-05,
+      "loss": 0.6217,
+      "step": 5336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.413588433849086e-05,
+      "loss": 1.3618,
+      "step": 5337
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.411839938571406e-05,
+      "loss": 1.0111,
+      "step": 5338
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.410091659275335e-05,
+      "loss": 1.1776,
+      "step": 5339
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4083435961986325e-05,
+      "loss": 0.6942,
+      "step": 5340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.406595749579024e-05,
+      "loss": 1.2058,
+      "step": 5341
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4048481196542095e-05,
+      "loss": 0.9764,
+      "step": 5342
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4031007066618524e-05,
+      "loss": 0.6856,
+      "step": 5343
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4013535108396e-05,
+      "loss": 0.6174,
+      "step": 5344
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.399606532425057e-05,
+      "loss": 1.0824,
+      "step": 5345
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.397859771655803e-05,
+      "loss": 1.2679,
+      "step": 5346
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.396113228769392e-05,
+      "loss": 0.9217,
+      "step": 5347
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.394366904003342e-05,
+      "loss": 1.1127,
+      "step": 5348
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.392620797595146e-05,
+      "loss": 1.2499,
+      "step": 5349
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.390874909782261e-05,
+      "loss": 1.302,
+      "step": 5350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3891292408021256e-05,
+      "loss": 1.1465,
+      "step": 5351
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.387383790892137e-05,
+      "loss": 1.7305,
+      "step": 5352
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.385638560289671e-05,
+      "loss": 0.9932,
+      "step": 5353
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.383893549232064e-05,
+      "loss": 0.7767,
+      "step": 5354
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3821487579566355e-05,
+      "loss": 1.2466,
+      "step": 5355
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.380404186700665e-05,
+      "loss": 1.151,
+      "step": 5356
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.378659835701404e-05,
+      "loss": 1.4941,
+      "step": 5357
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3769157051960766e-05,
+      "loss": 0.7597,
+      "step": 5358
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.375171795421876e-05,
+      "loss": 0.863,
+      "step": 5359
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.373428106615966e-05,
+      "loss": 1.5371,
+      "step": 5360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.371684639015476e-05,
+      "loss": 0.6074,
+      "step": 5361
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3699413928575095e-05,
+      "loss": 1.0393,
+      "step": 5362
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.368198368379143e-05,
+      "loss": 0.797,
+      "step": 5363
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.366455565817414e-05,
+      "loss": 0.6849,
+      "step": 5364
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3647129854093364e-05,
+      "loss": 1.3591,
+      "step": 5365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3629706273918926e-05,
+      "loss": 0.6433,
+      "step": 5366
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.361228492002035e-05,
+      "loss": 0.7896,
+      "step": 5367
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.359486579476684e-05,
+      "loss": 1.4756,
+      "step": 5368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.357744890052729e-05,
+      "loss": 0.6565,
+      "step": 5369
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3560034239670355e-05,
+      "loss": 1.1616,
+      "step": 5370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.35426218145643e-05,
+      "loss": 0.8649,
+      "step": 5371
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.352521162757714e-05,
+      "loss": 0.8586,
+      "step": 5372
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.350780368107655e-05,
+      "loss": 0.5477,
+      "step": 5373
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.349039797742997e-05,
+      "loss": 0.6483,
+      "step": 5374
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3472994519004436e-05,
+      "loss": 1.7288,
+      "step": 5375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.345559330816677e-05,
+      "loss": 1.3999,
+      "step": 5376
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.343819434728339e-05,
+      "loss": 0.8982,
+      "step": 5377
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3420797638720524e-05,
+      "loss": 1.0181,
+      "step": 5378
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.340340318484401e-05,
+      "loss": 1.3471,
+      "step": 5379
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.338601098801941e-05,
+      "loss": 0.9718,
+      "step": 5380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.336862105061196e-05,
+      "loss": 1.2109,
+      "step": 5381
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.335123337498662e-05,
+      "loss": 1.3453,
+      "step": 5382
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3333847963508025e-05,
+      "loss": 1.0367,
+      "step": 5383
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.331646481854048e-05,
+      "loss": 1.2882,
+      "step": 5384
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.329908394244801e-05,
+      "loss": 1.2023,
+      "step": 5385
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.328170533759434e-05,
+      "loss": 0.7428,
+      "step": 5386
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.326432900634287e-05,
+      "loss": 1.0599,
+      "step": 5387
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.324695495105667e-05,
+      "loss": 0.8274,
+      "step": 5388
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3229583174098554e-05,
+      "loss": 1.3409,
+      "step": 5389
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.321221367783098e-05,
+      "loss": 0.9892,
+      "step": 5390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.31948464646161e-05,
+      "loss": 1.1825,
+      "step": 5391
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3177481536815757e-05,
+      "loss": 0.8694,
+      "step": 5392
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.316011889679153e-05,
+      "loss": 1.1494,
+      "step": 5393
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3142758546904626e-05,
+      "loss": 0.6446,
+      "step": 5394
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.312540048951596e-05,
+      "loss": 0.7889,
+      "step": 5395
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.310804472698612e-05,
+      "loss": 0.9033,
+      "step": 5396
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3090691261675454e-05,
+      "loss": 1.4247,
+      "step": 5397
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3073340095943906e-05,
+      "loss": 0.7189,
+      "step": 5398
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3055991232151164e-05,
+      "loss": 0.5973,
+      "step": 5399
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3038644672656544e-05,
+      "loss": 1.2545,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.302130041981914e-05,
+      "loss": 0.9414,
+      "step": 5401
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.300395847599766e-05,
+      "loss": 1.3225,
+      "step": 5402
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.298661884355053e-05,
+      "loss": 1.3131,
+      "step": 5403
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2969281524835815e-05,
+      "loss": 1.5232,
+      "step": 5404
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.295194652221133e-05,
+      "loss": 1.1969,
+      "step": 5405
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.293461383803457e-05,
+      "loss": 1.4524,
+      "step": 5406
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.291728347466262e-05,
+      "loss": 1.2955,
+      "step": 5407
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.28999554344524e-05,
+      "loss": 1.2038,
+      "step": 5408
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.288262971976038e-05,
+      "loss": 0.8031,
+      "step": 5409
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.28653063329428e-05,
+      "loss": 0.9345,
+      "step": 5410
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2847985276355524e-05,
+      "loss": 1.4189,
+      "step": 5411
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.283066655235415e-05,
+      "loss": 0.95,
+      "step": 5412
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2813350163293935e-05,
+      "loss": 1.33,
+      "step": 5413
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2796036111529805e-05,
+      "loss": 0.5313,
+      "step": 5414
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.277872439941637e-05,
+      "loss": 1.1194,
+      "step": 5415
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2761415029307966e-05,
+      "loss": 1.4006,
+      "step": 5416
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2744108003558574e-05,
+      "loss": 0.9735,
+      "step": 5417
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.272680332452184e-05,
+      "loss": 1.1802,
+      "step": 5418
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.27095009945511e-05,
+      "loss": 1.1173,
+      "step": 5419
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2692201015999444e-05,
+      "loss": 1.3108,
+      "step": 5420
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2674903391219523e-05,
+      "loss": 1.4019,
+      "step": 5421
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.265760812256374e-05,
+      "loss": 0.5962,
+      "step": 5422
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2640315212384154e-05,
+      "loss": 0.5752,
+      "step": 5423
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.262302466303255e-05,
+      "loss": 0.9774,
+      "step": 5424
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.260573647686031e-05,
+      "loss": 1.1899,
+      "step": 5425
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2588450656218574e-05,
+      "loss": 0.6823,
+      "step": 5426
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.257116720345807e-05,
+      "loss": 1.1914,
+      "step": 5427
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2553886120929324e-05,
+      "loss": 1.5121,
+      "step": 5428
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2536607410982434e-05,
+      "loss": 1.5441,
+      "step": 5429
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.251933107596722e-05,
+      "loss": 1.104,
+      "step": 5430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.25020571182332e-05,
+      "loss": 1.2886,
+      "step": 5431
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2484785540129515e-05,
+      "loss": 1.6932,
+      "step": 5432
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2467516344005034e-05,
+      "loss": 1.2352,
+      "step": 5433
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.245024953220824e-05,
+      "loss": 1.174,
+      "step": 5434
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2432985107087385e-05,
+      "loss": 0.9308,
+      "step": 5435
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.24157230709903e-05,
+      "loss": 0.5264,
+      "step": 5436
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2398463426264564e-05,
+      "loss": 1.426,
+      "step": 5437
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.238120617525736e-05,
+      "loss": 0.5348,
+      "step": 5438
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2363951320315616e-05,
+      "loss": 1.1007,
+      "step": 5439
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.234669886378591e-05,
+      "loss": 1.3365,
+      "step": 5440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.232944880801446e-05,
+      "loss": 0.9202,
+      "step": 5441
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.23122011553472e-05,
+      "loss": 1.571,
+      "step": 5442
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2294955908129724e-05,
+      "loss": 1.2194,
+      "step": 5443
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.227771306870731e-05,
+      "loss": 0.9144,
+      "step": 5444
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.226047263942486e-05,
+      "loss": 1.623,
+      "step": 5445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2243234622626994e-05,
+      "loss": 0.9523,
+      "step": 5446
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.222599902065803e-05,
+      "loss": 0.9669,
+      "step": 5447
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.220876583586189e-05,
+      "loss": 1.375,
+      "step": 5448
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.219153507058219e-05,
+      "loss": 1.2988,
+      "step": 5449
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.217430672716223e-05,
+      "loss": 1.1006,
+      "step": 5450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2157080807945016e-05,
+      "loss": 0.9231,
+      "step": 5451
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.213985731527313e-05,
+      "loss": 1.4358,
+      "step": 5452
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.212263625148891e-05,
+      "loss": 0.5002,
+      "step": 5453
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.210541761893433e-05,
+      "loss": 0.7999,
+      "step": 5454
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2088201419951026e-05,
+      "loss": 1.1455,
+      "step": 5455
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2070987656880315e-05,
+      "loss": 1.0786,
+      "step": 5456
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2053776332063176e-05,
+      "loss": 1.0961,
+      "step": 5457
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.203656744784028e-05,
+      "loss": 1.3964,
+      "step": 5458
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.201936100655193e-05,
+      "loss": 1.309,
+      "step": 5459
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.200215701053812e-05,
+      "loss": 0.8992,
+      "step": 5460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.198495546213849e-05,
+      "loss": 0.9882,
+      "step": 5461
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1967756363692375e-05,
+      "loss": 0.5628,
+      "step": 5462
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.195055971753877e-05,
+      "loss": 0.9943,
+      "step": 5463
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1933365526016336e-05,
+      "loss": 0.8729,
+      "step": 5464
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1916173791463355e-05,
+      "loss": 1.5931,
+      "step": 5465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.189898451621786e-05,
+      "loss": 1.0483,
+      "step": 5466
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.188179770261749e-05,
+      "loss": 0.6045,
+      "step": 5467
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.186461335299955e-05,
+      "loss": 1.544,
+      "step": 5468
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1847431469701034e-05,
+      "loss": 1.3382,
+      "step": 5469
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.183025205505859e-05,
+      "loss": 1.7227,
+      "step": 5470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.181307511140854e-05,
+      "loss": 0.7231,
+      "step": 5471
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.179590064108683e-05,
+      "loss": 1.6034,
+      "step": 5472
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1778728646429136e-05,
+      "loss": 1.1591,
+      "step": 5473
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1761559129770745e-05,
+      "loss": 1.1746,
+      "step": 5474
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.174439209344661e-05,
+      "loss": 0.6811,
+      "step": 5475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1727227539791374e-05,
+      "loss": 1.4744,
+      "step": 5476
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.171006547113932e-05,
+      "loss": 1.1804,
+      "step": 5477
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.169290588982443e-05,
+      "loss": 1.4409,
+      "step": 5478
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1675748798180285e-05,
+      "loss": 1.1263,
+      "step": 5479
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.165859419854015e-05,
+      "loss": 1.0374,
+      "step": 5480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1641442093237016e-05,
+      "loss": 1.3148,
+      "step": 5481
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.162429248460344e-05,
+      "loss": 0.5869,
+      "step": 5482
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.16071453749717e-05,
+      "loss": 1.0142,
+      "step": 5483
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.159000076667367e-05,
+      "loss": 0.573,
+      "step": 5484
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.157285866204101e-05,
+      "loss": 0.8828,
+      "step": 5485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.155571906340488e-05,
+      "loss": 1.082,
+      "step": 5486
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1538581973096235e-05,
+      "loss": 1.274,
+      "step": 5487
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.152144739344558e-05,
+      "loss": 0.4941,
+      "step": 5488
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.150431532678317e-05,
+      "loss": 1.3123,
+      "step": 5489
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.148718577543887e-05,
+      "loss": 1.4548,
+      "step": 5490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.147005874174219e-05,
+      "loss": 0.9351,
+      "step": 5491
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.145293422802234e-05,
+      "loss": 1.4708,
+      "step": 5492
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.143581223660815e-05,
+      "loss": 0.8689,
+      "step": 5493
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.141869276982815e-05,
+      "loss": 1.041,
+      "step": 5494
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1401575830010464e-05,
+      "loss": 0.5435,
+      "step": 5495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1384461419482934e-05,
+      "loss": 1.115,
+      "step": 5496
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.136734954057305e-05,
+      "loss": 1.363,
+      "step": 5497
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.135024019560788e-05,
+      "loss": 1.4697,
+      "step": 5498
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.133313338691424e-05,
+      "loss": 0.7791,
+      "step": 5499
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1316029116818565e-05,
+      "loss": 1.664,
+      "step": 5500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.129892738764698e-05,
+      "loss": 0.9353,
+      "step": 5501
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1281828201725185e-05,
+      "loss": 0.8885,
+      "step": 5502
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.126473156137859e-05,
+      "loss": 1.2577,
+      "step": 5503
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1247637468932276e-05,
+      "loss": 1.4235,
+      "step": 5504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.123054592671094e-05,
+      "loss": 0.6889,
+      "step": 5505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1213456937038934e-05,
+      "loss": 1.3363,
+      "step": 5506
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.119637050224026e-05,
+      "loss": 1.5935,
+      "step": 5507
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.117928662463863e-05,
+      "loss": 1.6548,
+      "step": 5508
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.116220530655733e-05,
+      "loss": 1.0355,
+      "step": 5509
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.114512655031935e-05,
+      "loss": 1.0481,
+      "step": 5510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.112805035824727e-05,
+      "loss": 1.2238,
+      "step": 5511
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.111097673266342e-05,
+      "loss": 1.2988,
+      "step": 5512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.10939056758897e-05,
+      "loss": 0.8562,
+      "step": 5513
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1076837190247674e-05,
+      "loss": 1.1715,
+      "step": 5514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.10597712780586e-05,
+      "loss": 1.4374,
+      "step": 5515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.104270794164334e-05,
+      "loss": 0.5708,
+      "step": 5516
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.102564718332242e-05,
+      "loss": 1.2643,
+      "step": 5517
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1008589005415986e-05,
+      "loss": 1.3907,
+      "step": 5518
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.099153341024391e-05,
+      "loss": 0.7377,
+      "step": 5519
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.097448040012564e-05,
+      "loss": 1.4403,
+      "step": 5520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.09574299773803e-05,
+      "loss": 1.4304,
+      "step": 5521
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0940382144326666e-05,
+      "loss": 1.3017,
+      "step": 5522
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.092333690328314e-05,
+      "loss": 1.4749,
+      "step": 5523
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.090629425656783e-05,
+      "loss": 1.7257,
+      "step": 5524
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.088925420649839e-05,
+      "loss": 1.2527,
+      "step": 5525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.08722167553922e-05,
+      "loss": 0.7892,
+      "step": 5526
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.085518190556628e-05,
+      "loss": 1.0376,
+      "step": 5527
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.083814965933729e-05,
+      "loss": 1.3759,
+      "step": 5528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0821120019021504e-05,
+      "loss": 0.6993,
+      "step": 5529
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.080409298693486e-05,
+      "loss": 0.9279,
+      "step": 5530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.078706856539298e-05,
+      "loss": 1.2841,
+      "step": 5531
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.077004675671108e-05,
+      "loss": 0.9938,
+      "step": 5532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0753027563204034e-05,
+      "loss": 0.9169,
+      "step": 5533
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0736010987186345e-05,
+      "loss": 1.4833,
+      "step": 5534
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.071899703097223e-05,
+      "loss": 0.5514,
+      "step": 5535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.070198569687547e-05,
+      "loss": 1.2566,
+      "step": 5536
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.06849769872095e-05,
+      "loss": 0.5944,
+      "step": 5537
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.066797090428747e-05,
+      "loss": 1.1425,
+      "step": 5538
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0650967450422074e-05,
+      "loss": 0.6503,
+      "step": 5539
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0633966627925716e-05,
+      "loss": 0.9355,
+      "step": 5540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.061696843911039e-05,
+      "loss": 1.1896,
+      "step": 5541
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.059997288628782e-05,
+      "loss": 1.0779,
+      "step": 5542
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0582979971769266e-05,
+      "loss": 1.1333,
+      "step": 5543
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0565989697865704e-05,
+      "loss": 0.9135,
+      "step": 5544
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.054900206688769e-05,
+      "loss": 1.0598,
+      "step": 5545
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0532017081145484e-05,
+      "loss": 0.874,
+      "step": 5546
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0515034742948944e-05,
+      "loss": 1.4822,
+      "step": 5547
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.04980550546076e-05,
+      "loss": 1.0759,
+      "step": 5548
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0481078018430565e-05,
+      "loss": 1.3767,
+      "step": 5549
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0464103636726664e-05,
+      "loss": 1.1486,
+      "step": 5550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0447131911804327e-05,
+      "loss": 1.6254,
+      "step": 5551
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.043016284597159e-05,
+      "loss": 1.497,
+      "step": 5552
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0413196441536172e-05,
+      "loss": 1.1183,
+      "step": 5553
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0396232700805428e-05,
+      "loss": 1.1286,
+      "step": 5554
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0379271626086347e-05,
+      "loss": 0.8727,
+      "step": 5555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.036231321968551e-05,
+      "loss": 0.9855,
+      "step": 5556
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.034535748390921e-05,
+      "loss": 0.6093,
+      "step": 5557
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0328404421063334e-05,
+      "loss": 1.1331,
+      "step": 5558
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0311454033453397e-05,
+      "loss": 1.0303,
+      "step": 5559
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0294506323384558e-05,
+      "loss": 0.8784,
+      "step": 5560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.027756129316165e-05,
+      "loss": 1.4072,
+      "step": 5561
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0260618945089092e-05,
+      "loss": 1.1362,
+      "step": 5562
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0243679281470955e-05,
+      "loss": 0.924,
+      "step": 5563
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0226742304610932e-05,
+      "loss": 0.525,
+      "step": 5564
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0209808016812413e-05,
+      "loss": 1.3781,
+      "step": 5565
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0192876420378323e-05,
+      "loss": 1.3046,
+      "step": 5566
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0175947517611303e-05,
+      "loss": 1.0819,
+      "step": 5567
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.015902131081355e-05,
+      "loss": 0.5967,
+      "step": 5568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0142097802287018e-05,
+      "loss": 0.9193,
+      "step": 5569
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0125176994333154e-05,
+      "loss": 1.3726,
+      "step": 5570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0108258889253138e-05,
+      "loss": 1.7687,
+      "step": 5571
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.009134348934771e-05,
+      "loss": 1.4725,
+      "step": 5572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0074430796917308e-05,
+      "loss": 0.5871,
+      "step": 5573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0057520814261964e-05,
+      "loss": 1.1482,
+      "step": 5574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0040613543681345e-05,
+      "loss": 1.4439,
+      "step": 5575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.002370898747474e-05,
+      "loss": 0.9941,
+      "step": 5576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0006807147941102e-05,
+      "loss": 1.0643,
+      "step": 5577
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9989908027379e-05,
+      "loss": 0.4155,
+      "step": 5578
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9973011628086595e-05,
+      "loss": 0.7164,
+      "step": 5579
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.995611795236174e-05,
+      "loss": 1.6633,
+      "step": 5580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.993922700250189e-05,
+      "loss": 0.6739,
+      "step": 5581
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9922338780804106e-05,
+      "loss": 1.143,
+      "step": 5582
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.99054532895651e-05,
+      "loss": 1.2324,
+      "step": 5583
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9888570531081227e-05,
+      "loss": 1.2028,
+      "step": 5584
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9871690507648465e-05,
+      "loss": 1.4442,
+      "step": 5585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.985481322156239e-05,
+      "loss": 1.2354,
+      "step": 5586
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9837938675118217e-05,
+      "loss": 0.7176,
+      "step": 5587
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9821066870610826e-05,
+      "loss": 0.8414,
+      "step": 5588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.980419781033468e-05,
+      "loss": 1.0706,
+      "step": 5589
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9787331496583887e-05,
+      "loss": 1.0234,
+      "step": 5590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9770467931652168e-05,
+      "loss": 1.8975,
+      "step": 5591
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9753607117832903e-05,
+      "loss": 1.5368,
+      "step": 5592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9736749057419066e-05,
+      "loss": 1.2077,
+      "step": 5593
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.971989375270327e-05,
+      "loss": 0.9926,
+      "step": 5594
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9703041205977734e-05,
+      "loss": 0.9623,
+      "step": 5595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9686191419534327e-05,
+      "loss": 1.353,
+      "step": 5596
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9669344395664544e-05,
+      "loss": 1.4806,
+      "step": 5597
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.965250013665949e-05,
+      "loss": 1.0653,
+      "step": 5598
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9635658644809872e-05,
+      "loss": 0.6205,
+      "step": 5599
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9618819922406083e-05,
+      "loss": 1.418,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9601983971738097e-05,
+      "loss": 1.4005,
+      "step": 5601
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9585150795095485e-05,
+      "loss": 1.1923,
+      "step": 5602
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9568320394767502e-05,
+      "loss": 1.7904,
+      "step": 5603
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9551492773042994e-05,
+      "loss": 1.264,
+      "step": 5604
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9534667932210435e-05,
+      "loss": 1.2692,
+      "step": 5605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9517845874557883e-05,
+      "loss": 0.6004,
+      "step": 5606
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9501026602373083e-05,
+      "loss": 1.2849,
+      "step": 5607
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9484210117943377e-05,
+      "loss": 1.4533,
+      "step": 5608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9467396423555706e-05,
+      "loss": 1.0426,
+      "step": 5609
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.945058552149662e-05,
+      "loss": 1.1344,
+      "step": 5610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.943377741405236e-05,
+      "loss": 1.5066,
+      "step": 5611
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9416972103508734e-05,
+      "loss": 1.3083,
+      "step": 5612
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9400169592151154e-05,
+      "loss": 1.1491,
+      "step": 5613
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.938336988226468e-05,
+      "loss": 1.1353,
+      "step": 5614
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.936657297613402e-05,
+      "loss": 1.0565,
+      "step": 5615
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.934977887604342e-05,
+      "loss": 1.7099,
+      "step": 5616
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.933298758427684e-05,
+      "loss": 0.7508,
+      "step": 5617
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9316199103117736e-05,
+      "loss": 0.9811,
+      "step": 5618
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9299413434849342e-05,
+      "loss": 1.4183,
+      "step": 5619
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9282630581754373e-05,
+      "loss": 1.3028,
+      "step": 5620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.926585054611521e-05,
+      "loss": 1.2951,
+      "step": 5621
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9249073330213884e-05,
+      "loss": 0.9923,
+      "step": 5622
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.923229893633197e-05,
+      "loss": 1.449,
+      "step": 5623
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9215527366750733e-05,
+      "loss": 1.1587,
+      "step": 5624
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9198758623750977e-05,
+      "loss": 1.2244,
+      "step": 5625
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9181992709613216e-05,
+      "loss": 1.4059,
+      "step": 5626
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.91652296266175e-05,
+      "loss": 1.468,
+      "step": 5627
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.914846937704353e-05,
+      "loss": 1.5072,
+      "step": 5628
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9131711963170605e-05,
+      "loss": 1.3129,
+      "step": 5629
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9114957387277657e-05,
+      "loss": 1.6299,
+      "step": 5630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9098205651643218e-05,
+      "loss": 1.327,
+      "step": 5631
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.908145675854544e-05,
+      "loss": 1.3874,
+      "step": 5632
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9064710710262084e-05,
+      "loss": 1.4659,
+      "step": 5633
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.904796750907053e-05,
+      "loss": 1.1959,
+      "step": 5634
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9031227157247758e-05,
+      "loss": 1.0838,
+      "step": 5635
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9014489657070398e-05,
+      "loss": 1.1014,
+      "step": 5636
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8997755010814607e-05,
+      "loss": 0.9724,
+      "step": 5637
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8981023220756263e-05,
+      "loss": 0.5773,
+      "step": 5638
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8964294289170797e-05,
+      "loss": 1.0392,
+      "step": 5639
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.894756821833326e-05,
+      "loss": 0.8571,
+      "step": 5640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.893084501051826e-05,
+      "loss": 1.566,
+      "step": 5641
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8914124668000127e-05,
+      "loss": 0.5384,
+      "step": 5642
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.889740719305274e-05,
+      "loss": 1.1772,
+      "step": 5643
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.888069258794953e-05,
+      "loss": 1.1408,
+      "step": 5644
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8863980854963668e-05,
+      "loss": 1.2261,
+      "step": 5645
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8847271996367832e-05,
+      "loss": 1.1335,
+      "step": 5646
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8830566014434358e-05,
+      "loss": 1.0861,
+      "step": 5647
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8813862911435124e-05,
+      "loss": 0.7216,
+      "step": 5648
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8797162689641722e-05,
+      "loss": 1.2651,
+      "step": 5649
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8780465351325293e-05,
+      "loss": 1.0242,
+      "step": 5650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8763770898756552e-05,
+      "loss": 1.7281,
+      "step": 5651
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8747079334205862e-05,
+      "loss": 1.242,
+      "step": 5652
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8730390659943225e-05,
+      "loss": 1.1192,
+      "step": 5653
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8713704878238217e-05,
+      "loss": 1.2077,
+      "step": 5654
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.869702199135998e-05,
+      "loss": 1.5879,
+      "step": 5655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.86803420015773e-05,
+      "loss": 1.2148,
+      "step": 5656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8663664911158626e-05,
+      "loss": 1.1651,
+      "step": 5657
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.86469907223719e-05,
+      "loss": 1.3617,
+      "step": 5658
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8630319437484753e-05,
+      "loss": 0.7669,
+      "step": 5659
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8613651058764357e-05,
+      "loss": 0.9598,
+      "step": 5660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8596985588477604e-05,
+      "loss": 1.3173,
+      "step": 5661
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.858032302889084e-05,
+      "loss": 0.463,
+      "step": 5662
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8563663382270094e-05,
+      "loss": 1.1633,
+      "step": 5663
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8547006650881055e-05,
+      "loss": 1.0977,
+      "step": 5664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8530352836988884e-05,
+      "loss": 1.175,
+      "step": 5665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8513701942858436e-05,
+      "loss": 1.2483,
+      "step": 5666
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8497053970754127e-05,
+      "loss": 0.424,
+      "step": 5667
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8480408922940067e-05,
+      "loss": 1.1894,
+      "step": 5668
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8463766801679826e-05,
+      "loss": 1.4452,
+      "step": 5669
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.844712760923666e-05,
+      "loss": 0.9956,
+      "step": 5670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8430491347873433e-05,
+      "loss": 0.6259,
+      "step": 5671
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.841385801985258e-05,
+      "loss": 1.2579,
+      "step": 5672
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.839722762743614e-05,
+      "loss": 1.3494,
+      "step": 5673
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8380600172885778e-05,
+      "loss": 0.9961,
+      "step": 5674
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.836397565846274e-05,
+      "loss": 1.284,
+      "step": 5675
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8347354086427857e-05,
+      "loss": 1.077,
+      "step": 5676
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.833073545904159e-05,
+      "loss": 1.3062,
+      "step": 5677
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8314119778563997e-05,
+      "loss": 0.5519,
+      "step": 5678
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8297507047254708e-05,
+      "loss": 1.4072,
+      "step": 5679
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8280897267372974e-05,
+      "loss": 0.6391,
+      "step": 5680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.826429044117764e-05,
+      "loss": 1.2304,
+      "step": 5681
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8247686570927158e-05,
+      "loss": 1.4342,
+      "step": 5682
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8231085658879554e-05,
+      "loss": 0.9252,
+      "step": 5683
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.821448770729248e-05,
+      "loss": 1.2429,
+      "step": 5684
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.819789271842316e-05,
+      "loss": 0.7946,
+      "step": 5685
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.818130069452844e-05,
+      "loss": 1.3783,
+      "step": 5686
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.816471163786474e-05,
+      "loss": 1.0864,
+      "step": 5687
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8148125550688094e-05,
+      "loss": 0.5965,
+      "step": 5688
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8131542435254116e-05,
+      "loss": 1.5672,
+      "step": 5689
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.811496229381803e-05,
+      "loss": 1.3843,
+      "step": 5690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8098385128634653e-05,
+      "loss": 1.7688,
+      "step": 5691
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8081810941958385e-05,
+      "loss": 1.139,
+      "step": 5692
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.806523973604326e-05,
+      "loss": 1.4235,
+      "step": 5693
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.804867151314281e-05,
+      "loss": 0.8368,
+      "step": 5694
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8032106275510296e-05,
+      "loss": 1.3068,
+      "step": 5695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8015544025398476e-05,
+      "loss": 1.7092,
+      "step": 5696
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7998984765059755e-05,
+      "loss": 1.072,
+      "step": 5697
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7982428496746048e-05,
+      "loss": 0.5089,
+      "step": 5698
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7965875222708988e-05,
+      "loss": 0.7222,
+      "step": 5699
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.794932494519972e-05,
+      "loss": 0.7861,
+      "step": 5700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7932777666468974e-05,
+      "loss": 1.2021,
+      "step": 5701
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.791623338876708e-05,
+      "loss": 1.3613,
+      "step": 5702
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7899692114344028e-05,
+      "loss": 1.3156,
+      "step": 5703
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.788315384544934e-05,
+      "loss": 1.3018,
+      "step": 5704
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7866618584332073e-05,
+      "loss": 1.2723,
+      "step": 5705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7850086333240994e-05,
+      "loss": 1.6976,
+      "step": 5706
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7833557094424417e-05,
+      "loss": 1.4201,
+      "step": 5707
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7817030870130184e-05,
+      "loss": 0.8023,
+      "step": 5708
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.780050766260578e-05,
+      "loss": 1.0154,
+      "step": 5709
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7783987474098318e-05,
+      "loss": 0.7533,
+      "step": 5710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.776747030685446e-05,
+      "loss": 0.9781,
+      "step": 5711
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.77509561631204e-05,
+      "loss": 1.3706,
+      "step": 5712
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7734445045142e-05,
+      "loss": 0.9739,
+      "step": 5713
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7717936955164736e-05,
+      "loss": 0.9716,
+      "step": 5714
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7701431895433562e-05,
+      "loss": 1.2987,
+      "step": 5715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.76849298681931e-05,
+      "loss": 1.1172,
+      "step": 5716
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7668430875687533e-05,
+      "loss": 0.501,
+      "step": 5717
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.765193492016068e-05,
+      "loss": 0.8422,
+      "step": 5718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.763544200385586e-05,
+      "loss": 1.8706,
+      "step": 5719
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7618952129016047e-05,
+      "loss": 1.0115,
+      "step": 5720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7602465297883774e-05,
+      "loss": 1.2281,
+      "step": 5721
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7585981512701153e-05,
+      "loss": 0.6187,
+      "step": 5722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.756950077570992e-05,
+      "loss": 1.0645,
+      "step": 5723
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7553023089151354e-05,
+      "loss": 1.0466,
+      "step": 5724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7536548455266332e-05,
+      "loss": 0.7105,
+      "step": 5725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7520076876295332e-05,
+      "loss": 1.1302,
+      "step": 5726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7503608354478393e-05,
+      "loss": 1.6319,
+      "step": 5727
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7487142892055155e-05,
+      "loss": 1.3434,
+      "step": 5728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7470680491264845e-05,
+      "loss": 1.0643,
+      "step": 5729
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.745422115434625e-05,
+      "loss": 0.5909,
+      "step": 5730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7437764883537755e-05,
+      "loss": 1.2656,
+      "step": 5731
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7421311681077344e-05,
+      "loss": 1.2835,
+      "step": 5732
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7404861549202555e-05,
+      "loss": 1.4365,
+      "step": 5733
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.738841449015052e-05,
+      "loss": 1.414,
+      "step": 5734
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.737197050615796e-05,
+      "loss": 1.3293,
+      "step": 5735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7355529599461184e-05,
+      "loss": 1.2081,
+      "step": 5736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7339091772296045e-05,
+      "loss": 1.2848,
+      "step": 5737
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7322657026898023e-05,
+      "loss": 1.2121,
+      "step": 5738
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.730622536550215e-05,
+      "loss": 1.1307,
+      "step": 5739
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7289796790343058e-05,
+      "loss": 1.1593,
+      "step": 5740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.727337130365493e-05,
+      "loss": 1.752,
+      "step": 5741
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7256948907671566e-05,
+      "loss": 1.059,
+      "step": 5742
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7240529604626335e-05,
+      "loss": 0.8974,
+      "step": 5743
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7224113396752128e-05,
+      "loss": 1.4129,
+      "step": 5744
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7207700286281517e-05,
+      "loss": 0.5413,
+      "step": 5745
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.719129027544659e-05,
+      "loss": 1.2671,
+      "step": 5746
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.717488336647901e-05,
+      "loss": 1.4916,
+      "step": 5747
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.715847956161004e-05,
+      "loss": 1.154,
+      "step": 5748
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.714207886307052e-05,
+      "loss": 1.0084,
+      "step": 5749
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7125681273090874e-05,
+      "loss": 1.2401,
+      "step": 5750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7109286793901023e-05,
+      "loss": 0.9558,
+      "step": 5751
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7092895427730604e-05,
+      "loss": 1.6817,
+      "step": 5752
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7076507176808736e-05,
+      "loss": 0.8895,
+      "step": 5753
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7060122043364155e-05,
+      "loss": 1.1038,
+      "step": 5754
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7043740029625085e-05,
+      "loss": 1.3201,
+      "step": 5755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7027361137819472e-05,
+      "loss": 1.2474,
+      "step": 5756
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.701098537017476e-05,
+      "loss": 1.2878,
+      "step": 5757
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6994612728917923e-05,
+      "loss": 1.5814,
+      "step": 5758
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.697824321627555e-05,
+      "loss": 1.3647,
+      "step": 5759
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6961876834473866e-05,
+      "loss": 1.3642,
+      "step": 5760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6945513585738598e-05,
+      "loss": 0.8648,
+      "step": 5761
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6929153472295032e-05,
+      "loss": 0.7727,
+      "step": 5762
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6912796496368063e-05,
+      "loss": 0.7979,
+      "step": 5763
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6896442660182213e-05,
+      "loss": 0.9147,
+      "step": 5764
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.688009196596146e-05,
+      "loss": 0.6039,
+      "step": 5765
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6863744415929436e-05,
+      "loss": 0.6445,
+      "step": 5766
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.68474000123093e-05,
+      "loss": 0.9383,
+      "step": 5767
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6831058757323857e-05,
+      "loss": 1.5163,
+      "step": 5768
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6814720653195397e-05,
+      "loss": 1.6014,
+      "step": 5769
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.67983857021458e-05,
+      "loss": 0.9986,
+      "step": 5770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6782053906396608e-05,
+      "loss": 1.212,
+      "step": 5771
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6765725268168797e-05,
+      "loss": 0.8156,
+      "step": 5772
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.674939978968298e-05,
+      "loss": 1.1082,
+      "step": 5773
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.673307747315935e-05,
+      "loss": 1.28,
+      "step": 5774
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6716758320817696e-05,
+      "loss": 1.2383,
+      "step": 5775
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6700442334877285e-05,
+      "loss": 0.9192,
+      "step": 5776
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6684129517557033e-05,
+      "loss": 1.2776,
+      "step": 5777
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6667819871075365e-05,
+      "loss": 1.3529,
+      "step": 5778
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6651513397650384e-05,
+      "loss": 0.3843,
+      "step": 5779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6635210099499618e-05,
+      "loss": 1.2879,
+      "step": 5780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6618909978840257e-05,
+      "loss": 1.0627,
+      "step": 5781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6602613037889028e-05,
+      "loss": 0.6919,
+      "step": 5782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.658631927886224e-05,
+      "loss": 1.1045,
+      "step": 5783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6570028703975748e-05,
+      "loss": 0.6031,
+      "step": 5784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6553741315444995e-05,
+      "loss": 1.635,
+      "step": 5785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.653745711548498e-05,
+      "loss": 1.0895,
+      "step": 5786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6521176106310276e-05,
+      "loss": 0.8683,
+      "step": 5787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6504898290135004e-05,
+      "loss": 0.8294,
+      "step": 5788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6488623669172873e-05,
+      "loss": 0.9578,
+      "step": 5789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6472352245637155e-05,
+      "loss": 1.2535,
+      "step": 5790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.645608402174066e-05,
+      "loss": 1.1921,
+      "step": 5791
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.64398189996958e-05,
+      "loss": 0.6533,
+      "step": 5792
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6423557181714525e-05,
+      "loss": 1.5028,
+      "step": 5793
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6407298570008364e-05,
+      "loss": 1.4011,
+      "step": 5794
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.63910431667884e-05,
+      "loss": 1.2026,
+      "step": 5795
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.637479097426529e-05,
+      "loss": 1.3394,
+      "step": 5796
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6358541994649245e-05,
+      "loss": 0.6961,
+      "step": 5797
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6342296230150042e-05,
+      "loss": 1.2413,
+      "step": 5798
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6326053682977024e-05,
+      "loss": 1.3062,
+      "step": 5799
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.630981435533909e-05,
+      "loss": 1.1217,
+      "step": 5800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6293578249444713e-05,
+      "loss": 1.2985,
+      "step": 5801
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.627734536750191e-05,
+      "loss": 0.6485,
+      "step": 5802
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6261115711718275e-05,
+      "loss": 0.8331,
+      "step": 5803
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6244889284300978e-05,
+      "loss": 1.4614,
+      "step": 5804
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6228666087456676e-05,
+      "loss": 1.2317,
+      "step": 5805
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6212446123391686e-05,
+      "loss": 1.3338,
+      "step": 5806
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6196229394311832e-05,
+      "loss": 1.7677,
+      "step": 5807
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6180015902422516e-05,
+      "loss": 1.4004,
+      "step": 5808
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.616380564992864e-05,
+      "loss": 1.2295,
+      "step": 5809
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6147598639034766e-05,
+      "loss": 1.5226,
+      "step": 5810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.613139487194497e-05,
+      "loss": 1.1428,
+      "step": 5811
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.611519435086282e-05,
+      "loss": 1.2954,
+      "step": 5812
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6098997077991562e-05,
+      "loss": 0.5898,
+      "step": 5813
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.608280305553392e-05,
+      "loss": 1.5473,
+      "step": 5814
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6066612285692228e-05,
+      "loss": 1.3822,
+      "step": 5815
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6050424770668268e-05,
+      "loss": 0.8194,
+      "step": 5816
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6034240512663542e-05,
+      "loss": 1.0301,
+      "step": 5817
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.601805951387901e-05,
+      "loss": 1.2717,
+      "step": 5818
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6001881776515175e-05,
+      "loss": 1.6204,
+      "step": 5819
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5985707302772117e-05,
+      "loss": 1.7643,
+      "step": 5820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5969536094849524e-05,
+      "loss": 1.1387,
+      "step": 5821
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.59533681549466e-05,
+      "loss": 1.3245,
+      "step": 5822
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5937203485262063e-05,
+      "loss": 0.689,
+      "step": 5823
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5921042087994217e-05,
+      "loss": 1.1806,
+      "step": 5824
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5904883965340997e-05,
+      "loss": 1.4531,
+      "step": 5825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5888729119499765e-05,
+      "loss": 1.1194,
+      "step": 5826
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.587257755266751e-05,
+      "loss": 1.3118,
+      "step": 5827
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5856429267040748e-05,
+      "loss": 1.3791,
+      "step": 5828
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.584028426481562e-05,
+      "loss": 1.1587,
+      "step": 5829
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5824142548187703e-05,
+      "loss": 1.3941,
+      "step": 5830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.580800411935221e-05,
+      "loss": 0.5565,
+      "step": 5831
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5791868980503885e-05,
+      "loss": 0.577,
+      "step": 5832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.577573713383702e-05,
+      "loss": 1.7226,
+      "step": 5833
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5759608581545468e-05,
+      "loss": 0.3823,
+      "step": 5834
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5743483325822602e-05,
+      "loss": 1.2069,
+      "step": 5835
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5727361368861446e-05,
+      "loss": 0.3865,
+      "step": 5836
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5711242712854435e-05,
+      "loss": 0.6856,
+      "step": 5837
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5695127359993635e-05,
+      "loss": 1.4296,
+      "step": 5838
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5679015312470668e-05,
+      "loss": 1.0661,
+      "step": 5839
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5662906572476685e-05,
+      "loss": 1.2402,
+      "step": 5840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5646801142202387e-05,
+      "loss": 0.9471,
+      "step": 5841
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5630699023838034e-05,
+      "loss": 1.3926,
+      "step": 5842
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5614600219573427e-05,
+      "loss": 1.3638,
+      "step": 5843
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.559850473159792e-05,
+      "loss": 0.8659,
+      "step": 5844
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5582412562100422e-05,
+      "loss": 1.198,
+      "step": 5845
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.556632371326939e-05,
+      "loss": 1.3235,
+      "step": 5846
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5550238187292807e-05,
+      "loss": 0.5191,
+      "step": 5847
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5534155986358243e-05,
+      "loss": 1.1762,
+      "step": 5848
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5518077112652783e-05,
+      "loss": 1.7847,
+      "step": 5849
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5502001568363065e-05,
+      "loss": 0.829,
+      "step": 5850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5485929355675296e-05,
+      "loss": 1.4866,
+      "step": 5851
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5469860476775204e-05,
+      "loss": 1.6443,
+      "step": 5852
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5453794933848075e-05,
+      "loss": 1.5951,
+      "step": 5853
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5437732729078744e-05,
+      "loss": 1.2418,
+      "step": 5854
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5421673864651592e-05,
+      "loss": 1.3602,
+      "step": 5855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5405618342750526e-05,
+      "loss": 1.1861,
+      "step": 5856
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5389566165559032e-05,
+      "loss": 0.5072,
+      "step": 5857
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.537351733526011e-05,
+      "loss": 1.0562,
+      "step": 5858
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5357471854036334e-05,
+      "loss": 1.5204,
+      "step": 5859
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.534142972406979e-05,
+      "loss": 0.6237,
+      "step": 5860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.532539094754216e-05,
+      "loss": 1.4466,
+      "step": 5861
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5309355526634572e-05,
+      "loss": 1.2164,
+      "step": 5862
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5293323463527806e-05,
+      "loss": 1.2537,
+      "step": 5863
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5277294760402137e-05,
+      "loss": 0.6078,
+      "step": 5864
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.52612694194374e-05,
+      "loss": 0.7968,
+      "step": 5865
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5245247442812903e-05,
+      "loss": 0.8671,
+      "step": 5866
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5229228832707606e-05,
+      "loss": 1.3916,
+      "step": 5867
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5213213591299955e-05,
+      "loss": 1.224,
+      "step": 5868
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5197201720767917e-05,
+      "loss": 0.657,
+      "step": 5869
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.518119322328901e-05,
+      "loss": 0.893,
+      "step": 5870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5165188101040356e-05,
+      "loss": 1.3347,
+      "step": 5871
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5149186356198562e-05,
+      "loss": 1.0868,
+      "step": 5872
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5133187990939745e-05,
+      "loss": 1.4902,
+      "step": 5873
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.51171930074396e-05,
+      "loss": 1.4721,
+      "step": 5874
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.510120140787342e-05,
+      "loss": 1.4484,
+      "step": 5875
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5085213194415936e-05,
+      "loss": 1.0177,
+      "step": 5876
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.506922836924144e-05,
+      "loss": 1.4512,
+      "step": 5877
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5053246934523843e-05,
+      "loss": 0.8963,
+      "step": 5878
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5037268892436526e-05,
+      "loss": 1.6234,
+      "step": 5879
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5021294245152393e-05,
+      "loss": 1.2084,
+      "step": 5880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.500532299484391e-05,
+      "loss": 1.3505,
+      "step": 5881
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.498935514368314e-05,
+      "loss": 0.8773,
+      "step": 5882
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.497339069384157e-05,
+      "loss": 1.3118,
+      "step": 5883
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4957429647490303e-05,
+      "loss": 1.4305,
+      "step": 5884
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.494147200679995e-05,
+      "loss": 1.4413,
+      "step": 5885
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.492551777394071e-05,
+      "loss": 1.5421,
+      "step": 5886
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.490956695108223e-05,
+      "loss": 1.3128,
+      "step": 5887
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.489361954039376e-05,
+      "loss": 1.4354,
+      "step": 5888
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4877675544044053e-05,
+      "loss": 1.2994,
+      "step": 5889
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.486173496420143e-05,
+      "loss": 1.2591,
+      "step": 5890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.484579780303371e-05,
+      "loss": 0.6154,
+      "step": 5891
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4829864062708274e-05,
+      "loss": 1.2028,
+      "step": 5892
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4813933745392026e-05,
+      "loss": 0.5509,
+      "step": 5893
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.479800685325141e-05,
+      "loss": 0.8113,
+      "step": 5894
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4782083388452404e-05,
+      "loss": 1.3813,
+      "step": 5895
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.476616335316051e-05,
+      "loss": 1.0744,
+      "step": 5896
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4750246749540774e-05,
+      "loss": 1.1868,
+      "step": 5897
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4734333579757774e-05,
+      "loss": 1.2021,
+      "step": 5898
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.471842384597562e-05,
+      "loss": 1.4808,
+      "step": 5899
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4702517550357946e-05,
+      "loss": 0.8129,
+      "step": 5900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.468661469506794e-05,
+      "loss": 0.6208,
+      "step": 5901
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4670715282268297e-05,
+      "loss": 0.9558,
+      "step": 5902
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4654819314121262e-05,
+      "loss": 1.3024,
+      "step": 5903
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.46389267927886e-05,
+      "loss": 1.1804,
+      "step": 5904
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4623037720431624e-05,
+      "loss": 1.6882,
+      "step": 5905
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.460715209921115e-05,
+      "loss": 0.9805,
+      "step": 5906
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4591269931287546e-05,
+      "loss": 1.2274,
+      "step": 5907
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4575391218820713e-05,
+      "loss": 1.2125,
+      "step": 5908
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.455951596397007e-05,
+      "loss": 1.2898,
+      "step": 5909
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4543644168894565e-05,
+      "loss": 1.3279,
+      "step": 5910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.452777583575271e-05,
+      "loss": 0.9005,
+      "step": 5911
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4511910966702445e-05,
+      "loss": 1.4562,
+      "step": 5912
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.449604956390138e-05,
+      "loss": 1.4371,
+      "step": 5913
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.448019162950656e-05,
+      "loss": 1.1007,
+      "step": 5914
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4464337165674606e-05,
+      "loss": 0.6402,
+      "step": 5915
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.444848617456158e-05,
+      "loss": 1.5123,
+      "step": 5916
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.44326386583232e-05,
+      "loss": 1.1997,
+      "step": 5917
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.441679461911464e-05,
+      "loss": 0.7598,
+      "step": 5918
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.440095405909056e-05,
+      "loss": 0.6496,
+      "step": 5919
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4385116980405236e-05,
+      "loss": 1.024,
+      "step": 5920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4369283385212428e-05,
+      "loss": 0.6275,
+      "step": 5921
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4353453275665427e-05,
+      "loss": 1.4374,
+      "step": 5922
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4337626653917e-05,
+      "loss": 0.7352,
+      "step": 5923
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.432180352211954e-05,
+      "loss": 1.3515,
+      "step": 5924
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4305983882424915e-05,
+      "loss": 1.2428,
+      "step": 5925
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.429016773698447e-05,
+      "loss": 1.3295,
+      "step": 5926
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.427435508794913e-05,
+      "loss": 0.944,
+      "step": 5927
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4258545937469355e-05,
+      "loss": 1.0284,
+      "step": 5928
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4242740287695127e-05,
+      "loss": 1.4574,
+      "step": 5929
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4226938140775884e-05,
+      "loss": 1.5192,
+      "step": 5930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.421113949886064e-05,
+      "loss": 1.3912,
+      "step": 5931
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4195344364097982e-05,
+      "loss": 1.2588,
+      "step": 5932
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.417955273863592e-05,
+      "loss": 1.5157,
+      "step": 5933
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4163764624622043e-05,
+      "loss": 1.3185,
+      "step": 5934
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4147980024203436e-05,
+      "loss": 1.249,
+      "step": 5935
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.413219893952679e-05,
+      "loss": 1.5332,
+      "step": 5936
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4116421372738186e-05,
+      "loss": 1.5184,
+      "step": 5937
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.410064732598329e-05,
+      "loss": 1.2233,
+      "step": 5938
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4084876801407346e-05,
+      "loss": 1.3862,
+      "step": 5939
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.406910980115502e-05,
+      "loss": 0.7746,
+      "step": 5940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4053346327370558e-05,
+      "loss": 1.126,
+      "step": 5941
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4037586382197685e-05,
+      "loss": 0.6087,
+      "step": 5942
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4021829967779734e-05,
+      "loss": 0.5451,
+      "step": 5943
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4006077086259437e-05,
+      "loss": 1.1303,
+      "step": 5944
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3990327739779135e-05,
+      "loss": 0.5495,
+      "step": 5945
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3974581930480638e-05,
+      "loss": 1.0164,
+      "step": 5946
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3958839660505316e-05,
+      "loss": 1.1527,
+      "step": 5947
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3943100931994022e-05,
+      "loss": 1.0804,
+      "step": 5948
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3927365747087154e-05,
+      "loss": 1.033,
+      "step": 5949
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.39116341079246e-05,
+      "loss": 1.2568,
+      "step": 5950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3895906016645802e-05,
+      "loss": 0.6165,
+      "step": 5951
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.388018147538969e-05,
+      "loss": 1.0809,
+      "step": 5952
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3864460486294716e-05,
+      "loss": 1.3714,
+      "step": 5953
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3848743051498863e-05,
+      "loss": 1.7327,
+      "step": 5954
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.383302917313961e-05,
+      "loss": 0.901,
+      "step": 5955
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3817318853353977e-05,
+      "loss": 0.8345,
+      "step": 5956
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3801612094278476e-05,
+      "loss": 0.9824,
+      "step": 5957
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3785908898049153e-05,
+      "loss": 1.0431,
+      "step": 5958
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3770209266801558e-05,
+      "loss": 1.217,
+      "step": 5959
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3754513202670762e-05,
+      "loss": 0.974,
+      "step": 5960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3738820707791348e-05,
+      "loss": 0.325,
+      "step": 5961
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3723131784297413e-05,
+      "loss": 1.0336,
+      "step": 5962
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3707446434322573e-05,
+      "loss": 1.2854,
+      "step": 5963
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3691764659999955e-05,
+      "loss": 1.1917,
+      "step": 5964
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.36760864634622e-05,
+      "loss": 1.052,
+      "step": 5965
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3660411846841458e-05,
+      "loss": 0.9771,
+      "step": 5966
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.36447408122694e-05,
+      "loss": 1.1347,
+      "step": 5967
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.362907336187723e-05,
+      "loss": 0.589,
+      "step": 5968
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.361340949779557e-05,
+      "loss": 1.0676,
+      "step": 5969
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3597749222154696e-05,
+      "loss": 1.0225,
+      "step": 5970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.35820925370843e-05,
+      "loss": 0.9622,
+      "step": 5971
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3566439444713624e-05,
+      "loss": 1.4533,
+      "step": 5972
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3550789947171366e-05,
+      "loss": 0.8445,
+      "step": 5973
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3535144046585817e-05,
+      "loss": 1.2625,
+      "step": 5974
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3519501745084753e-05,
+      "loss": 1.2051,
+      "step": 5975
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3503863044795404e-05,
+      "loss": 1.2487,
+      "step": 5976
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.348822794784455e-05,
+      "loss": 1.4448,
+      "step": 5977
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.347259645635852e-05,
+      "loss": 1.2631,
+      "step": 5978
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.345696857246313e-05,
+      "loss": 1.2382,
+      "step": 5979
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3441344298283634e-05,
+      "loss": 1.6077,
+      "step": 5980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3425723635944874e-05,
+      "loss": 1.3514,
+      "step": 5981
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.341010658757122e-05,
+      "loss": 1.1278,
+      "step": 5982
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.339449315528646e-05,
+      "loss": 0.5546,
+      "step": 5983
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3378883341213942e-05,
+      "loss": 0.7335,
+      "step": 5984
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.336327714747656e-05,
+      "loss": 0.8566,
+      "step": 5985
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3347674576196665e-05,
+      "loss": 1.2279,
+      "step": 5986
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3332075629496104e-05,
+      "loss": 0.9961,
+      "step": 5987
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3316480309496242e-05,
+      "loss": 1.2399,
+      "step": 5988
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3300888618318005e-05,
+      "loss": 1.2932,
+      "step": 5989
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3285300558081785e-05,
+      "loss": 0.7299,
+      "step": 5990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3269716130907437e-05,
+      "loss": 0.9578,
+      "step": 5991
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3254135338914363e-05,
+      "loss": 0.8301,
+      "step": 5992
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3238558184221528e-05,
+      "loss": 0.6828,
+      "step": 5993
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3222984668947283e-05,
+      "loss": 1.3444,
+      "step": 5994
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.320741479520958e-05,
+      "loss": 1.1536,
+      "step": 5995
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3191848565125805e-05,
+      "loss": 1.1435,
+      "step": 5996
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3176285980812958e-05,
+      "loss": 1.2995,
+      "step": 5997
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3160727044387404e-05,
+      "loss": 1.4334,
+      "step": 5998
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.31451717579651e-05,
+      "loss": 0.5996,
+      "step": 5999
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3129620123661482e-05,
+      "loss": 1.1032,
+      "step": 6000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3114072143591498e-05,
+      "loss": 1.014,
+      "step": 6001
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3098527819869586e-05,
+      "loss": 1.4524,
+      "step": 6002
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.308298715460968e-05,
+      "loss": 0.4455,
+      "step": 6003
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.306745014992529e-05,
+      "loss": 1.0866,
+      "step": 6004
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3051916807929315e-05,
+      "loss": 1.1659,
+      "step": 6005
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.303638713073421e-05,
+      "loss": 0.5987,
+      "step": 6006
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3020861120451953e-05,
+      "loss": 1.1932,
+      "step": 6007
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3005338779193993e-05,
+      "loss": 0.9316,
+      "step": 6008
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.298982010907128e-05,
+      "loss": 1.3169,
+      "step": 6009
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.297430511219429e-05,
+      "loss": 1.1808,
+      "step": 6010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2958793790672973e-05,
+      "loss": 1.3101,
+      "step": 6011
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2943286146616788e-05,
+      "loss": 0.9857,
+      "step": 6012
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2927782182134698e-05,
+      "loss": 1.4908,
+      "step": 6013
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2912281899335164e-05,
+      "loss": 1.1262,
+      "step": 6014
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2896785300326146e-05,
+      "loss": 1.0153,
+      "step": 6015
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2881292387215097e-05,
+      "loss": 1.4558,
+      "step": 6016
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2865803162108972e-05,
+      "loss": 1.4007,
+      "step": 6017
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2850317627114233e-05,
+      "loss": 1.4299,
+      "step": 6018
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2834835784336828e-05,
+      "loss": 0.4954,
+      "step": 6019
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.281935763588221e-05,
+      "loss": 1.4253,
+      "step": 6020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.280388318385533e-05,
+      "loss": 1.2349,
+      "step": 6021
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2788412430360645e-05,
+      "loss": 1.1388,
+      "step": 6022
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2772945377502052e-05,
+      "loss": 0.4341,
+      "step": 6023
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.275748202738304e-05,
+      "loss": 0.8462,
+      "step": 6024
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.274202238210653e-05,
+      "loss": 1.1615,
+      "step": 6025
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2726566443774948e-05,
+      "loss": 1.292,
+      "step": 6026
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.271111421449023e-05,
+      "loss": 1.4905,
+      "step": 6027
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.26956656963538e-05,
+      "loss": 1.262,
+      "step": 6028
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2680220891466593e-05,
+      "loss": 0.7652,
+      "step": 6029
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.266477980192896e-05,
+      "loss": 1.2724,
+      "step": 6030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2649342429840874e-05,
+      "loss": 1.1323,
+      "step": 6031
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.263390877730172e-05,
+      "loss": 1.3213,
+      "step": 6032
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.261847884641042e-05,
+      "loss": 1.0407,
+      "step": 6033
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.260305263926529e-05,
+      "loss": 1.4299,
+      "step": 6034
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2587630157964285e-05,
+      "loss": 1.0585,
+      "step": 6035
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2572211404604787e-05,
+      "loss": 1.3363,
+      "step": 6036
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2556796381283617e-05,
+      "loss": 1.3895,
+      "step": 6037
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.254138509009715e-05,
+      "loss": 0.7892,
+      "step": 6038
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2525977533141275e-05,
+      "loss": 1.5672,
+      "step": 6039
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2510573712511336e-05,
+      "loss": 0.8482,
+      "step": 6040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2495173630302146e-05,
+      "loss": 0.9255,
+      "step": 6041
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2479777288608027e-05,
+      "loss": 0.5245,
+      "step": 6042
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2464384689522872e-05,
+      "loss": 0.7727,
+      "step": 6043
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2448995835139925e-05,
+      "loss": 0.8521,
+      "step": 6044
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2433610727552e-05,
+      "loss": 0.7903,
+      "step": 6045
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2418229368851428e-05,
+      "loss": 0.4013,
+      "step": 6046
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.240285176112999e-05,
+      "loss": 1.7272,
+      "step": 6047
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.238747790647893e-05,
+      "loss": 0.957,
+      "step": 6048
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2372107806989012e-05,
+      "loss": 1.2332,
+      "step": 6049
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2356741464750547e-05,
+      "loss": 1.209,
+      "step": 6050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.234137888185322e-05,
+      "loss": 0.9172,
+      "step": 6051
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2326020060386287e-05,
+      "loss": 1.3017,
+      "step": 6052
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2310665002438446e-05,
+      "loss": 0.7297,
+      "step": 6053
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.229531371009796e-05,
+      "loss": 1.4391,
+      "step": 6054
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2279966185452474e-05,
+      "loss": 1.5336,
+      "step": 6055
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.226462243058919e-05,
+      "loss": 1.9435,
+      "step": 6056
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2249282447594782e-05,
+      "loss": 1.0717,
+      "step": 6057
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2233946238555405e-05,
+      "loss": 1.6145,
+      "step": 6058
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2218613805556705e-05,
+      "loss": 1.5373,
+      "step": 6059
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2203285150683818e-05,
+      "loss": 0.5371,
+      "step": 6060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2187960276021358e-05,
+      "loss": 0.892,
+      "step": 6061
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2172639183653428e-05,
+      "loss": 1.5708,
+      "step": 6062
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2157321875663623e-05,
+      "loss": 0.8054,
+      "step": 6063
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.214200835413502e-05,
+      "loss": 1.2717,
+      "step": 6064
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2126698621150172e-05,
+      "loss": 1.2456,
+      "step": 6065
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2111392678791128e-05,
+      "loss": 1.2643,
+      "step": 6066
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2096090529139418e-05,
+      "loss": 1.1309,
+      "step": 6067
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2080792174276048e-05,
+      "loss": 1.1982,
+      "step": 6068
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2065497616281534e-05,
+      "loss": 1.4099,
+      "step": 6069
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.205020685723584e-05,
+      "loss": 1.1532,
+      "step": 6070
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.203491989921843e-05,
+      "loss": 0.6618,
+      "step": 6071
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2019636744308264e-05,
+      "loss": 1.6946,
+      "step": 6072
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2004357394583767e-05,
+      "loss": 0.7207,
+      "step": 6073
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.198908185212285e-05,
+      "loss": 1.3866,
+      "step": 6074
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.197381011900291e-05,
+      "loss": 1.0443,
+      "step": 6075
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.195854219730083e-05,
+      "loss": 1.1724,
+      "step": 6076
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1943278089092955e-05,
+      "loss": 1.1016,
+      "step": 6077
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.192801779645514e-05,
+      "loss": 1.0311,
+      "step": 6078
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1912761321462717e-05,
+      "loss": 1.5355,
+      "step": 6079
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.189750866619043e-05,
+      "loss": 1.0689,
+      "step": 6080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1882259832712625e-05,
+      "loss": 1.3094,
+      "step": 6081
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1867014823103044e-05,
+      "loss": 1.4097,
+      "step": 6082
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1851773639434942e-05,
+      "loss": 1.3439,
+      "step": 6083
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1836536283780996e-05,
+      "loss": 1.2865,
+      "step": 6084
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1821302758213458e-05,
+      "loss": 0.7057,
+      "step": 6085
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.180607306480401e-05,
+      "loss": 0.7034,
+      "step": 6086
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.179084720562376e-05,
+      "loss": 1.1727,
+      "step": 6087
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1775625182743398e-05,
+      "loss": 0.9233,
+      "step": 6088
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.176040699823302e-05,
+      "loss": 1.6773,
+      "step": 6089
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1745192654162245e-05,
+      "loss": 0.7559,
+      "step": 6090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1729982152600088e-05,
+      "loss": 1.0595,
+      "step": 6091
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.171477549561515e-05,
+      "loss": 0.731,
+      "step": 6092
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.169957268527547e-05,
+      "loss": 0.487,
+      "step": 6093
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.168437372364851e-05,
+      "loss": 0.6846,
+      "step": 6094
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1669178612801245e-05,
+      "loss": 1.4082,
+      "step": 6095
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.165398735480017e-05,
+      "loss": 1.447,
+      "step": 6096
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1638799951711225e-05,
+      "loss": 1.2425,
+      "step": 6097
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1623616405599776e-05,
+      "loss": 1.6343,
+      "step": 6098
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1608436718530707e-05,
+      "loss": 1.1347,
+      "step": 6099
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.159326089256844e-05,
+      "loss": 0.8874,
+      "step": 6100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.157808892977674e-05,
+      "loss": 0.4098,
+      "step": 6101
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1562920832218952e-05,
+      "loss": 0.6475,
+      "step": 6102
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1547756601957832e-05,
+      "loss": 0.9516,
+      "step": 6103
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1532596241055687e-05,
+      "loss": 1.3148,
+      "step": 6104
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.15174397515742e-05,
+      "loss": 1.0123,
+      "step": 6105
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1502287135574596e-05,
+      "loss": 0.7625,
+      "step": 6106
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1487138395117546e-05,
+      "loss": 0.8305,
+      "step": 6107
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1471993532263206e-05,
+      "loss": 1.0394,
+      "step": 6108
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1456852549071198e-05,
+      "loss": 1.4341,
+      "step": 6109
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.14417154476006e-05,
+      "loss": 1.4844,
+      "step": 6110
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1426582229910038e-05,
+      "loss": 1.2806,
+      "step": 6111
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.141145289805749e-05,
+      "loss": 1.2112,
+      "step": 6112
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.139632745410049e-05,
+      "loss": 1.4038,
+      "step": 6113
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1381205900096026e-05,
+      "loss": 0.8493,
+      "step": 6114
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.136608823810054e-05,
+      "loss": 1.0838,
+      "step": 6115
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.135097447016996e-05,
+      "loss": 1.1539,
+      "step": 6116
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1335864598359688e-05,
+      "loss": 0.878,
+      "step": 6117
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.132075862472458e-05,
+      "loss": 1.3969,
+      "step": 6118
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.130565655131898e-05,
+      "loss": 0.9256,
+      "step": 6119
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1290558380196685e-05,
+      "loss": 1.2978,
+      "step": 6120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.127546411341097e-05,
+      "loss": 0.5948,
+      "step": 6121
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1260373753014577e-05,
+      "loss": 1.0647,
+      "step": 6122
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.124528730105972e-05,
+      "loss": 0.866,
+      "step": 6123
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1230204759598083e-05,
+      "loss": 0.5258,
+      "step": 6124
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1215126130680803e-05,
+      "loss": 0.6958,
+      "step": 6125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1200051416358503e-05,
+      "loss": 1.205,
+      "step": 6126
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.118498061868127e-05,
+      "loss": 1.1146,
+      "step": 6127
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1169913739698648e-05,
+      "loss": 1.0045,
+      "step": 6128
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1154850781459678e-05,
+      "loss": 1.3129,
+      "step": 6129
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.113979174601279e-05,
+      "loss": 1.0074,
+      "step": 6130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1124736635405983e-05,
+      "loss": 1.3688,
+      "step": 6131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1109685451686666e-05,
+      "loss": 1.4671,
+      "step": 6132
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1094638196901723e-05,
+      "loss": 1.085,
+      "step": 6133
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1079594873097497e-05,
+      "loss": 0.7612,
+      "step": 6134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1064555482319803e-05,
+      "loss": 1.2294,
+      "step": 6135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1049520026613938e-05,
+      "loss": 0.6394,
+      "step": 6136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1034488508024602e-05,
+      "loss": 1.0648,
+      "step": 6137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1019460928596052e-05,
+      "loss": 0.646,
+      "step": 6138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1004437290371944e-05,
+      "loss": 1.6707,
+      "step": 6139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0989417595395426e-05,
+      "loss": 1.1345,
+      "step": 6140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.097440184570906e-05,
+      "loss": 0.5022,
+      "step": 6141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0959390043354955e-05,
+      "loss": 1.2098,
+      "step": 6142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0944382190374646e-05,
+      "loss": 1.0769,
+      "step": 6143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0929378288809076e-05,
+      "loss": 1.5582,
+      "step": 6144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.091437834069871e-05,
+      "loss": 1.4092,
+      "step": 6145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0899382348083496e-05,
+      "loss": 1.1883,
+      "step": 6146
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.088439031300281e-05,
+      "loss": 0.9117,
+      "step": 6147
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.086940223749546e-05,
+      "loss": 1.4051,
+      "step": 6148
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.085441812359974e-05,
+      "loss": 1.3937,
+      "step": 6149
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0839437973353475e-05,
+      "loss": 1.429,
+      "step": 6150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0824461788793832e-05,
+      "loss": 1.3789,
+      "step": 6151
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.080948957195749e-05,
+      "loss": 0.7,
+      "step": 6152
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.079452132488064e-05,
+      "loss": 1.0396,
+      "step": 6153
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0779557049598874e-05,
+      "loss": 0.6653,
+      "step": 6154
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.076459674814723e-05,
+      "loss": 1.5089,
+      "step": 6155
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.074964042256023e-05,
+      "loss": 1.4835,
+      "step": 6156
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0734688074871912e-05,
+      "loss": 1.1123,
+      "step": 6157
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0719739707115658e-05,
+      "loss": 1.2027,
+      "step": 6158
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0704795321324393e-05,
+      "loss": 0.854,
+      "step": 6159
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.068985491953046e-05,
+      "loss": 1.146,
+      "step": 6160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0674918503765727e-05,
+      "loss": 1.0259,
+      "step": 6161
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.065998607606142e-05,
+      "loss": 1.3205,
+      "step": 6162
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0645057638448284e-05,
+      "loss": 1.2042,
+      "step": 6163
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.063013319295652e-05,
+      "loss": 1.6075,
+      "step": 6164
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0615212741615765e-05,
+      "loss": 1.1509,
+      "step": 6165
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0600296286455127e-05,
+      "loss": 1.4431,
+      "step": 6166
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0585383829503175e-05,
+      "loss": 1.3712,
+      "step": 6167
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.057047537278792e-05,
+      "loss": 0.8205,
+      "step": 6168
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.055557091833683e-05,
+      "loss": 0.721,
+      "step": 6169
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0540670468176844e-05,
+      "loss": 1.1167,
+      "step": 6170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0525774024334336e-05,
+      "loss": 1.1771,
+      "step": 6171
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0510881588835162e-05,
+      "loss": 1.2991,
+      "step": 6172
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0495993163704603e-05,
+      "loss": 1.2101,
+      "step": 6173
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0481108750967415e-05,
+      "loss": 1.3616,
+      "step": 6174
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.04662283526478e-05,
+      "loss": 0.8201,
+      "step": 6175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0451351970769412e-05,
+      "loss": 1.4601,
+      "step": 6176
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0436479607355362e-05,
+      "loss": 1.4244,
+      "step": 6177
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0421611264428224e-05,
+      "loss": 1.0929,
+      "step": 6178
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0406746944010012e-05,
+      "loss": 0.7002,
+      "step": 6179
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0391886648122193e-05,
+      "loss": 1.2789,
+      "step": 6180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0377030378785694e-05,
+      "loss": 0.5727,
+      "step": 6181
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0362178138020894e-05,
+      "loss": 1.6675,
+      "step": 6182
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.034732992784762e-05,
+      "loss": 0.6256,
+      "step": 6183
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0332485750285147e-05,
+      "loss": 1.3023,
+      "step": 6184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0317645607352215e-05,
+      "loss": 1.0595,
+      "step": 6185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0302809501067015e-05,
+      "loss": 1.0863,
+      "step": 6186
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0287977433447135e-05,
+      "loss": 1.049,
+      "step": 6187
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0273149406509717e-05,
+      "loss": 1.4234,
+      "step": 6188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0258325422271275e-05,
+      "loss": 0.6525,
+      "step": 6189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0243505482747804e-05,
+      "loss": 1.3783,
+      "step": 6190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0228689589954695e-05,
+      "loss": 0.8579,
+      "step": 6191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0213877745906897e-05,
+      "loss": 1.2001,
+      "step": 6192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0199069952618704e-05,
+      "loss": 1.4428,
+      "step": 6193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0184266212103914e-05,
+      "loss": 1.2884,
+      "step": 6194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.016946652637575e-05,
+      "loss": 1.3779,
+      "step": 6195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0154670897446908e-05,
+      "loss": 1.3431,
+      "step": 6196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.013987932732953e-05,
+      "loss": 1.4266,
+      "step": 6197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.012509181803513e-05,
+      "loss": 0.7033,
+      "step": 6198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.01103083715748e-05,
+      "loss": 1.1472,
+      "step": 6199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0095528989958984e-05,
+      "loss": 0.5677,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0080753675197627e-05,
+      "loss": 0.8953,
+      "step": 6201
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0065982429300047e-05,
+      "loss": 0.4801,
+      "step": 6202
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0051215254275096e-05,
+      "loss": 0.4852,
+      "step": 6203
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0036452152131047e-05,
+      "loss": 1.2903,
+      "step": 6204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0021693124875563e-05,
+      "loss": 1.4779,
+      "step": 6205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0006938174515798e-05,
+      "loss": 1.6674,
+      "step": 6206
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9992187303058386e-05,
+      "loss": 1.27,
+      "step": 6207
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9977440512509373e-05,
+      "loss": 1.0252,
+      "step": 6208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9962697804874203e-05,
+      "loss": 0.5946,
+      "step": 6209
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9947959182157817e-05,
+      "loss": 0.8673,
+      "step": 6210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.993322464636464e-05,
+      "loss": 0.6456,
+      "step": 6211
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9918494199498444e-05,
+      "loss": 1.4953,
+      "step": 6212
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9903767843562505e-05,
+      "loss": 1.2757,
+      "step": 6213
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9889045580559518e-05,
+      "loss": 1.5765,
+      "step": 6214
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9874327412491694e-05,
+      "loss": 1.3474,
+      "step": 6215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9859613341360567e-05,
+      "loss": 0.9112,
+      "step": 6216
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9844903369167172e-05,
+      "loss": 0.9638,
+      "step": 6217
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9830197497912055e-05,
+      "loss": 1.3623,
+      "step": 6218
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9815495729595078e-05,
+      "loss": 1.1295,
+      "step": 6219
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9800798066215625e-05,
+      "loss": 1.1828,
+      "step": 6220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9786104509772486e-05,
+      "loss": 0.8984,
+      "step": 6221
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9771415062263965e-05,
+      "loss": 1.4379,
+      "step": 6222
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9756729725687688e-05,
+      "loss": 1.3293,
+      "step": 6223
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9742048502040812e-05,
+      "loss": 0.5067,
+      "step": 6224
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9727371393319906e-05,
+      "loss": 1.3052,
+      "step": 6225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.971269840152098e-05,
+      "loss": 1.0857,
+      "step": 6226
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.969802952863949e-05,
+      "loss": 1.3543,
+      "step": 6227
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.968336477667032e-05,
+      "loss": 0.9264,
+      "step": 6228
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.96687041476078e-05,
+      "loss": 1.2532,
+      "step": 6229
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9654047643445716e-05,
+      "loss": 1.3094,
+      "step": 6230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9639395266177264e-05,
+      "loss": 1.0195,
+      "step": 6231
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9624747017795092e-05,
+      "loss": 0.8283,
+      "step": 6232
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9610102900291288e-05,
+      "loss": 1.5043,
+      "step": 6233
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.959546291565738e-05,
+      "loss": 0.7751,
+      "step": 6234
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9580827065884328e-05,
+      "loss": 0.4197,
+      "step": 6235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.956619535296253e-05,
+      "loss": 1.2769,
+      "step": 6236
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.955156777888183e-05,
+      "loss": 1.4246,
+      "step": 6237
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9536944345631496e-05,
+      "loss": 0.9291,
+      "step": 6238
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9522325055200246e-05,
+      "loss": 1.3402,
+      "step": 6239
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9507709909576228e-05,
+      "loss": 0.7239,
+      "step": 6240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9493098910747022e-05,
+      "loss": 1.4543,
+      "step": 6241
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.947849206069965e-05,
+      "loss": 1.4356,
+      "step": 6242
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9463889361420575e-05,
+      "loss": 1.2338,
+      "step": 6243
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9449290814895677e-05,
+      "loss": 1.2054,
+      "step": 6244
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9434696423110292e-05,
+      "loss": 0.9159,
+      "step": 6245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9420106188049185e-05,
+      "loss": 0.5709,
+      "step": 6246
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.940552011169656e-05,
+      "loss": 1.4496,
+      "step": 6247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9390938196036006e-05,
+      "loss": 1.4541,
+      "step": 6248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.937636044305064e-05,
+      "loss": 0.7923,
+      "step": 6249
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9361786854722936e-05,
+      "loss": 1.1094,
+      "step": 6250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9347217433034854e-05,
+      "loss": 1.5395,
+      "step": 6251
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.93326521799677e-05,
+      "loss": 1.2283,
+      "step": 6252
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9318091097502328e-05,
+      "loss": 0.9136,
+      "step": 6253
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9303534187618976e-05,
+      "loss": 0.734,
+      "step": 6254
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9288981452297273e-05,
+      "loss": 0.5771,
+      "step": 6255
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.927443289351631e-05,
+      "loss": 1.2091,
+      "step": 6256
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9259888513254653e-05,
+      "loss": 0.5685,
+      "step": 6257
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.924534831349027e-05,
+      "loss": 0.5133,
+      "step": 6258
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.923081229620049e-05,
+      "loss": 1.5678,
+      "step": 6259
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.92162804633622e-05,
+      "loss": 1.3085,
+      "step": 6260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9201752816951645e-05,
+      "loss": 1.3992,
+      "step": 6261
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9187229358944486e-05,
+      "loss": 1.0349,
+      "step": 6262
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9172710091315828e-05,
+      "loss": 1.1732,
+      "step": 6263
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9158195016040258e-05,
+      "loss": 1.1604,
+      "step": 6264
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.914368413509175e-05,
+      "loss": 1.296,
+      "step": 6265
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9129177450443675e-05,
+      "loss": 1.0185,
+      "step": 6266
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9114674964068862e-05,
+      "loss": 1.2877,
+      "step": 6267
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.910017667793964e-05,
+      "loss": 1.3171,
+      "step": 6268
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9085682594027638e-05,
+      "loss": 0.9909,
+      "step": 6269
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9071192714303992e-05,
+      "loss": 1.1631,
+      "step": 6270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9056707040739248e-05,
+      "loss": 0.5927,
+      "step": 6271
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.904222557530342e-05,
+      "loss": 1.6044,
+      "step": 6272
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.902774831996587e-05,
+      "loss": 1.1501,
+      "step": 6273
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9013275276695446e-05,
+      "loss": 1.2823,
+      "step": 6274
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8998806447460406e-05,
+      "loss": 1.1742,
+      "step": 6275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8984341834228438e-05,
+      "loss": 0.842,
+      "step": 6276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8969881438966654e-05,
+      "loss": 1.0492,
+      "step": 6277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.895542526364157e-05,
+      "loss": 0.454,
+      "step": 6278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8940973310219217e-05,
+      "loss": 1.039,
+      "step": 6279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8926525580664922e-05,
+      "loss": 0.6989,
+      "step": 6280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8912082076943533e-05,
+      "loss": 1.1942,
+      "step": 6281
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8897642801019277e-05,
+      "loss": 0.9192,
+      "step": 6282
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8883207754855832e-05,
+      "loss": 1.0067,
+      "step": 6283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8868776940416282e-05,
+      "loss": 1.2653,
+      "step": 6284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8854350359663153e-05,
+      "loss": 0.84,
+      "step": 6285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8839928014558378e-05,
+      "loss": 1.2833,
+      "step": 6286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8825509907063327e-05,
+      "loss": 1.3362,
+      "step": 6287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8811096039138786e-05,
+      "loss": 0.8546,
+      "step": 6288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8796686412744967e-05,
+      "loss": 1.4789,
+      "step": 6289
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8782281029841502e-05,
+      "loss": 1.315,
+      "step": 6290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8767879892387458e-05,
+      "loss": 1.2962,
+      "step": 6291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.875348300234131e-05,
+      "loss": 0.3619,
+      "step": 6292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8739090361660962e-05,
+      "loss": 0.5622,
+      "step": 6293
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8724701972303733e-05,
+      "loss": 1.5671,
+      "step": 6294
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.871031783622638e-05,
+      "loss": 1.3491,
+      "step": 6295
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.869593795538507e-05,
+      "loss": 1.4295,
+      "step": 6296
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8681562331735402e-05,
+      "loss": 0.5319,
+      "step": 6297
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.866719096723234e-05,
+      "loss": 1.1049,
+      "step": 6298
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.865282386383037e-05,
+      "loss": 0.9711,
+      "step": 6299
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.863846102348333e-05,
+      "loss": 0.6856,
+      "step": 6300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.862410244814448e-05,
+      "loss": 1.0082,
+      "step": 6301
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8609748139766525e-05,
+      "loss": 0.9302,
+      "step": 6302
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8595398100301575e-05,
+      "loss": 1.178,
+      "step": 6303
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.858105233170117e-05,
+      "loss": 1.4237,
+      "step": 6304
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.856671083591622e-05,
+      "loss": 1.238,
+      "step": 6305
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.855237361489714e-05,
+      "loss": 1.3987,
+      "step": 6306
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8538040670593705e-05,
+      "loss": 1.164,
+      "step": 6307
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.852371200495513e-05,
+      "loss": 0.4016,
+      "step": 6308
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.850938761993e-05,
+      "loss": 1.191,
+      "step": 6309
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.849506751746641e-05,
+      "loss": 1.6381,
+      "step": 6310
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8480751699511806e-05,
+      "loss": 1.2475,
+      "step": 6311
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8466440168013043e-05,
+      "loss": 0.5451,
+      "step": 6312
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8452132924916414e-05,
+      "loss": 1.1477,
+      "step": 6313
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.843782997216767e-05,
+      "loss": 1.3689,
+      "step": 6314
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8423531311711928e-05,
+      "loss": 1.3987,
+      "step": 6315
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8409236945493702e-05,
+      "loss": 1.0832,
+      "step": 6316
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8394946875456952e-05,
+      "loss": 0.9577,
+      "step": 6317
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.838066110354511e-05,
+      "loss": 0.6136,
+      "step": 6318
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8366379631700913e-05,
+      "loss": 1.253,
+      "step": 6319
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8352102461866587e-05,
+      "loss": 1.3947,
+      "step": 6320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8337829595983734e-05,
+      "loss": 1.1848,
+      "step": 6321
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8323561035993447e-05,
+      "loss": 0.5488,
+      "step": 6322
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8309296783836117e-05,
+      "loss": 1.4789,
+      "step": 6323
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8295036841451612e-05,
+      "loss": 1.1692,
+      "step": 6324
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8280781210779264e-05,
+      "loss": 0.7422,
+      "step": 6325
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8266529893757712e-05,
+      "loss": 0.8622,
+      "step": 6326
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8252282892325085e-05,
+      "loss": 0.9673,
+      "step": 6327
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8238040208418877e-05,
+      "loss": 1.2002,
+      "step": 6328
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.822380184397608e-05,
+      "loss": 1.3981,
+      "step": 6329
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8209567800932975e-05,
+      "loss": 0.9973,
+      "step": 6330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.819533808122535e-05,
+      "loss": 1.4277,
+      "step": 6331
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8181112686788358e-05,
+      "loss": 1.1767,
+      "step": 6332
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8166891619556597e-05,
+      "loss": 1.1914,
+      "step": 6333
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8152674881464042e-05,
+      "loss": 1.1947,
+      "step": 6334
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.813846247444411e-05,
+      "loss": 1.1181,
+      "step": 6335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8124254400429608e-05,
+      "loss": 0.6765,
+      "step": 6336
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.811005066135277e-05,
+      "loss": 1.4727,
+      "step": 6337
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8095851259145214e-05,
+      "loss": 1.0747,
+      "step": 6338
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8081656195738e-05,
+      "loss": 1.2646,
+      "step": 6339
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.806746547306158e-05,
+      "loss": 1.1671,
+      "step": 6340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8053279093045822e-05,
+      "loss": 1.243,
+      "step": 6341
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8039097057619992e-05,
+      "loss": 1.4937,
+      "step": 6342
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8024919368712784e-05,
+      "loss": 1.272,
+      "step": 6343
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.801074602825229e-05,
+      "loss": 0.9441,
+      "step": 6344
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7996577038166e-05,
+      "loss": 1.2831,
+      "step": 6345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7982412400380845e-05,
+      "loss": 1.5198,
+      "step": 6346
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.796825211682312e-05,
+      "loss": 1.1696,
+      "step": 6347
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7954096189418563e-05,
+      "loss": 1.2399,
+      "step": 6348
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7939944620092307e-05,
+      "loss": 1.4954,
+      "step": 6349
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.792579741076889e-05,
+      "loss": 1.3149,
+      "step": 6350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7911654563372266e-05,
+      "loss": 1.227,
+      "step": 6351
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7897516079825784e-05,
+      "loss": 1.5503,
+      "step": 6352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.78833819620522e-05,
+      "loss": 0.5935,
+      "step": 6353
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.786925221197371e-05,
+      "loss": 0.6272,
+      "step": 6354
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7855126831511826e-05,
+      "loss": 0.4385,
+      "step": 6355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.784100582258758e-05,
+      "loss": 1.2124,
+      "step": 6356
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.782688918712135e-05,
+      "loss": 0.7734,
+      "step": 6357
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.781277692703293e-05,
+      "loss": 1.6118,
+      "step": 6358
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7798669044241474e-05,
+      "loss": 0.6098,
+      "step": 6359
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7784565540665626e-05,
+      "loss": 1.3182,
+      "step": 6360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.777046641822339e-05,
+      "loss": 0.9383,
+      "step": 6361
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7756371678832145e-05,
+      "loss": 0.7447,
+      "step": 6362
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.77422813244087e-05,
+      "loss": 0.5291,
+      "step": 6363
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7728195356869303e-05,
+      "loss": 0.8923,
+      "step": 6364
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771411377812958e-05,
+      "loss": 1.3977,
+      "step": 6365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7700036590104495e-05,
+      "loss": 0.7743,
+      "step": 6366
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7685963794708526e-05,
+      "loss": 0.6614,
+      "step": 6367
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7671895393855503e-05,
+      "loss": 1.1327,
+      "step": 6368
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7657831389458618e-05,
+      "loss": 1.338,
+      "step": 6369
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7643771783430506e-05,
+      "loss": 1.1684,
+      "step": 6370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.762971657768324e-05,
+      "loss": 1.6291,
+      "step": 6371
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7615665774128255e-05,
+      "loss": 1.3176,
+      "step": 6372
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7601619374676343e-05,
+      "loss": 0.9235,
+      "step": 6373
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7587577381237754e-05,
+      "loss": 1.6062,
+      "step": 6374
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7573539795722172e-05,
+      "loss": 1.4592,
+      "step": 6375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7559506620038583e-05,
+      "loss": 1.3714,
+      "step": 6376
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7545477856095455e-05,
+      "loss": 1.071,
+      "step": 6377
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.75314535058006e-05,
+      "loss": 1.3868,
+      "step": 6378
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7517433571061315e-05,
+      "loss": 0.4651,
+      "step": 6379
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.750341805378418e-05,
+      "loss": 1.125,
+      "step": 6380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7489406955875263e-05,
+      "loss": 1.5771,
+      "step": 6381
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.747540027923999e-05,
+      "loss": 1.4618,
+      "step": 6382
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7461398025783205e-05,
+      "loss": 1.1742,
+      "step": 6383
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7447400197409134e-05,
+      "loss": 0.9112,
+      "step": 6384
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7433406796021402e-05,
+      "loss": 1.5489,
+      "step": 6385
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.741941782352308e-05,
+      "loss": 0.5641,
+      "step": 6386
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7405433281816557e-05,
+      "loss": 0.9501,
+      "step": 6387
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7391453172803672e-05,
+      "loss": 0.6244,
+      "step": 6388
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.737747749838563e-05,
+      "loss": 1.2014,
+      "step": 6389
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7363506260463096e-05,
+      "loss": 1.2082,
+      "step": 6390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7349539460936042e-05,
+      "loss": 0.9433,
+      "step": 6391
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7335577101703904e-05,
+      "loss": 1.531,
+      "step": 6392
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7321619184665477e-05,
+      "loss": 0.9982,
+      "step": 6393
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7307665711718974e-05,
+      "loss": 0.5635,
+      "step": 6394
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7293716684761994e-05,
+      "loss": 1.5791,
+      "step": 6395
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7279772105691538e-05,
+      "loss": 1.2973,
+      "step": 6396
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.726583197640399e-05,
+      "loss": 1.5677,
+      "step": 6397
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.725189629879514e-05,
+      "loss": 0.6122,
+      "step": 6398
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.723796507476017e-05,
+      "loss": 1.2628,
+      "step": 6399
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7224038306193646e-05,
+      "loss": 0.6571,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.721011599498955e-05,
+      "loss": 1.3394,
+      "step": 6401
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7196198143041236e-05,
+      "loss": 1.2693,
+      "step": 6402
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7182284752241462e-05,
+      "loss": 1.221,
+      "step": 6403
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.716837582448238e-05,
+      "loss": 1.2023,
+      "step": 6404
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7154471361655527e-05,
+      "loss": 0.5134,
+      "step": 6405
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7140571365651842e-05,
+      "loss": 1.1134,
+      "step": 6406
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7126675838361662e-05,
+      "loss": 0.9088,
+      "step": 6407
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.711278478167469e-05,
+      "loss": 1.0349,
+      "step": 6408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.709889819748005e-05,
+      "loss": 1.0787,
+      "step": 6409
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7085016087666243e-05,
+      "loss": 1.1804,
+      "step": 6410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7071138454121167e-05,
+      "loss": 1.0307,
+      "step": 6411
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7057265298732106e-05,
+      "loss": 1.312,
+      "step": 6412
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7043396623385738e-05,
+      "loss": 0.8975,
+      "step": 6413
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.702953242996813e-05,
+      "loss": 1.0314,
+      "step": 6414
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7015672720364768e-05,
+      "loss": 0.9533,
+      "step": 6415
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7001817496460437e-05,
+      "loss": 1.2941,
+      "step": 6416
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6987966760139433e-05,
+      "loss": 0.5557,
+      "step": 6417
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.697412051328537e-05,
+      "loss": 1.3606,
+      "step": 6418
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.696027875778128e-05,
+      "loss": 0.566,
+      "step": 6419
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6946441495509517e-05,
+      "loss": 1.2791,
+      "step": 6420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.693260872835194e-05,
+      "loss": 1.2741,
+      "step": 6421
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6918780458189727e-05,
+      "loss": 1.2435,
+      "step": 6422
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6904956686903413e-05,
+      "loss": 1.5721,
+      "step": 6423
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6891137416372967e-05,
+      "loss": 0.9039,
+      "step": 6424
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6877322648477767e-05,
+      "loss": 1.393,
+      "step": 6425
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6863512385096558e-05,
+      "loss": 1.3282,
+      "step": 6426
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6849706628107408e-05,
+      "loss": 0.8716,
+      "step": 6427
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.683590537938788e-05,
+      "loss": 0.9256,
+      "step": 6428
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6822108640814876e-05,
+      "loss": 1.3365,
+      "step": 6429
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6808316414264648e-05,
+      "loss": 1.2048,
+      "step": 6430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6794528701612867e-05,
+      "loss": 0.8071,
+      "step": 6431
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6780745504734617e-05,
+      "loss": 0.6712,
+      "step": 6432
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6766966825504356e-05,
+      "loss": 1.3037,
+      "step": 6433
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6753192665795868e-05,
+      "loss": 1.2623,
+      "step": 6434
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.673942302748238e-05,
+      "loss": 1.3582,
+      "step": 6435
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6725657912436532e-05,
+      "loss": 1.4526,
+      "step": 6436
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.671189732253026e-05,
+      "loss": 1.2218,
+      "step": 6437
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6698141259634952e-05,
+      "loss": 0.8438,
+      "step": 6438
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6684389725621348e-05,
+      "loss": 1.0521,
+      "step": 6439
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6670642722359636e-05,
+      "loss": 1.5058,
+      "step": 6440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6656900251719282e-05,
+      "loss": 0.5683,
+      "step": 6441
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6643162315569212e-05,
+      "loss": 1.0352,
+      "step": 6442
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.662942891577771e-05,
+      "loss": 1.1605,
+      "step": 6443
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6615700054212457e-05,
+      "loss": 1.3792,
+      "step": 6444
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6601975732740504e-05,
+      "loss": 0.6668,
+      "step": 6445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6588255953228282e-05,
+      "loss": 1.1594,
+      "step": 6446
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.657454071754162e-05,
+      "loss": 1.1572,
+      "step": 6447
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.656083002754571e-05,
+      "loss": 0.6672,
+      "step": 6448
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6547123885105143e-05,
+      "loss": 1.5499,
+      "step": 6449
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6533422292083882e-05,
+      "loss": 0.9921,
+      "step": 6450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6519725250345276e-05,
+      "loss": 0.8114,
+      "step": 6451
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6506032761752042e-05,
+      "loss": 1.3099,
+      "step": 6452
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.64923448281663e-05,
+      "loss": 1.2711,
+      "step": 6453
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6478661451449535e-05,
+      "loss": 1.2575,
+      "step": 6454
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6464982633462612e-05,
+      "loss": 1.4552,
+      "step": 6455
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6451308376065777e-05,
+      "loss": 1.5926,
+      "step": 6456
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6437638681118672e-05,
+      "loss": 1.2927,
+      "step": 6457
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.64239735504803e-05,
+      "loss": 0.4258,
+      "step": 6458
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6410312986009042e-05,
+      "loss": 1.0088,
+      "step": 6459
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6396656989562664e-05,
+      "loss": 0.9936,
+      "step": 6460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.638300556299832e-05,
+      "loss": 0.9202,
+      "step": 6461
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6369358708172532e-05,
+      "loss": 1.1926,
+      "step": 6462
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6355716426941204e-05,
+      "loss": 1.1109,
+      "step": 6463
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6342078721159603e-05,
+      "loss": 1.1261,
+      "step": 6464
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.632844559268242e-05,
+      "loss": 0.6973,
+      "step": 6465
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6314817043363633e-05,
+      "loss": 1.3769,
+      "step": 6466
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6301193075056704e-05,
+      "loss": 1.0755,
+      "step": 6467
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6287573689614404e-05,
+      "loss": 1.0055,
+      "step": 6468
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6273958888888923e-05,
+      "loss": 0.842,
+      "step": 6469
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.626034867473175e-05,
+      "loss": 0.7658,
+      "step": 6470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6246743048993855e-05,
+      "loss": 1.2184,
+      "step": 6471
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.623314201352553e-05,
+      "loss": 1.4261,
+      "step": 6472
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6219545570176402e-05,
+      "loss": 1.093,
+      "step": 6473
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6205953720795563e-05,
+      "loss": 1.6048,
+      "step": 6474
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.619236646723142e-05,
+      "loss": 1.6051,
+      "step": 6475
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.617878381133178e-05,
+      "loss": 1.3126,
+      "step": 6476
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.616520575494378e-05,
+      "loss": 1.1278,
+      "step": 6477
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6151632299913998e-05,
+      "loss": 1.2999,
+      "step": 6478
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6138063448088363e-05,
+      "loss": 1.5818,
+      "step": 6479
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6124499201312137e-05,
+      "loss": 0.5157,
+      "step": 6480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6110939561429984e-05,
+      "loss": 1.0937,
+      "step": 6481
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6097384530285982e-05,
+      "loss": 1.0753,
+      "step": 6482
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6083834109723543e-05,
+      "loss": 0.7163,
+      "step": 6483
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.607028830158542e-05,
+      "loss": 1.9701,
+      "step": 6484
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6056747107713782e-05,
+      "loss": 1.1589,
+      "step": 6485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.60432105299502e-05,
+      "loss": 1.3588,
+      "step": 6486
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6029678570135535e-05,
+      "loss": 1.0566,
+      "step": 6487
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6016151230110083e-05,
+      "loss": 1.0409,
+      "step": 6488
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.600262851171347e-05,
+      "loss": 1.4748,
+      "step": 6489
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.598911041678477e-05,
+      "loss": 1.1754,
+      "step": 6490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5975596947162324e-05,
+      "loss": 1.1511,
+      "step": 6491
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.59620881046839e-05,
+      "loss": 1.45,
+      "step": 6492
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.594858389118667e-05,
+      "loss": 1.4391,
+      "step": 6493
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.593508430850709e-05,
+      "loss": 0.59,
+      "step": 6494
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.592158935848106e-05,
+      "loss": 1.3285,
+      "step": 6495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5908099042943796e-05,
+      "loss": 1.1049,
+      "step": 6496
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5894613363729972e-05,
+      "loss": 1.1486,
+      "step": 6497
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5881132322673515e-05,
+      "loss": 1.4608,
+      "step": 6498
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.586765592160779e-05,
+      "loss": 1.5759,
+      "step": 6499
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5854184162365527e-05,
+      "loss": 1.661,
+      "step": 6500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.584071704677881e-05,
+      "loss": 0.9856,
+      "step": 6501
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.58272545766791e-05,
+      "loss": 1.3827,
+      "step": 6502
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5813796753897226e-05,
+      "loss": 0.6533,
+      "step": 6503
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5800343580263376e-05,
+      "loss": 1.4659,
+      "step": 6504
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.578689505760712e-05,
+      "loss": 0.5717,
+      "step": 6505
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5773451187757382e-05,
+      "loss": 0.4925,
+      "step": 6506
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5760011972542466e-05,
+      "loss": 1.6899,
+      "step": 6507
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5746577413790025e-05,
+      "loss": 1.7534,
+      "step": 6508
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.57331475133271e-05,
+      "loss": 0.5769,
+      "step": 6509
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5719722272980092e-05,
+      "loss": 1.017,
+      "step": 6510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5706301694574754e-05,
+      "loss": 1.1068,
+      "step": 6511
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5692885779936217e-05,
+      "loss": 0.8425,
+      "step": 6512
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5679474530888982e-05,
+      "loss": 1.3053,
+      "step": 6513
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5666067949256907e-05,
+      "loss": 1.3305,
+      "step": 6514
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5652666036863216e-05,
+      "loss": 0.904,
+      "step": 6515
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.563926879553051e-05,
+      "loss": 0.8843,
+      "step": 6516
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.562587622708073e-05,
+      "loss": 1.2594,
+      "step": 6517
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.561248833333521e-05,
+      "loss": 1.2135,
+      "step": 6518
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5599105116114632e-05,
+      "loss": 0.8733,
+      "step": 6519
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5585726577239035e-05,
+      "loss": 0.9406,
+      "step": 6520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5572352718527843e-05,
+      "loss": 1.5893,
+      "step": 6521
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5558983541799842e-05,
+      "loss": 1.4089,
+      "step": 6522
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.554561904887312e-05,
+      "loss": 0.8966,
+      "step": 6523
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5532259241565243e-05,
+      "loss": 1.4258,
+      "step": 6524
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5518904121693046e-05,
+      "loss": 1.1905,
+      "step": 6525
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5505553691072773e-05,
+      "loss": 1.0302,
+      "step": 6526
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.549220795151997e-05,
+      "loss": 1.3776,
+      "step": 6527
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5478866904849636e-05,
+      "loss": 1.5619,
+      "step": 6528
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.546553055287609e-05,
+      "loss": 0.829,
+      "step": 6529
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.545219889741296e-05,
+      "loss": 1.1482,
+      "step": 6530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5438871940273304e-05,
+      "loss": 1.2753,
+      "step": 6531
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5425549683269536e-05,
+      "loss": 1.4112,
+      "step": 6532
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.541223212821342e-05,
+      "loss": 1.1689,
+      "step": 6533
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5398919276916024e-05,
+      "loss": 1.0477,
+      "step": 6534
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5385611131187884e-05,
+      "loss": 1.1367,
+      "step": 6535
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5372307692838827e-05,
+      "loss": 1.396,
+      "step": 6536
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.535900896367803e-05,
+      "loss": 1.421,
+      "step": 6537
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.534571494551405e-05,
+      "loss": 1.1473,
+      "step": 6538
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.533242564015483e-05,
+      "loss": 1.5052,
+      "step": 6539
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5319141049407658e-05,
+      "loss": 1.6551,
+      "step": 6540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5305861175079127e-05,
+      "loss": 1.6267,
+      "step": 6541
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5292586018975247e-05,
+      "loss": 1.2242,
+      "step": 6542
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.52793155829014e-05,
+      "loss": 1.1139,
+      "step": 6543
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5266049868662262e-05,
+      "loss": 1.4824,
+      "step": 6544
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5252788878061919e-05,
+      "loss": 0.5345,
+      "step": 6545
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5239532612903774e-05,
+      "loss": 1.2108,
+      "step": 6546
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5226281074990661e-05,
+      "loss": 1.1686,
+      "step": 6547
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5213034266124677e-05,
+      "loss": 1.4305,
+      "step": 6548
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5199792188107332e-05,
+      "loss": 1.3861,
+      "step": 6549
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5186554842739482e-05,
+      "loss": 0.5413,
+      "step": 6550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5173322231821342e-05,
+      "loss": 1.5857,
+      "step": 6551
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5160094357152472e-05,
+      "loss": 0.6385,
+      "step": 6552
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5146871220531794e-05,
+      "loss": 1.6307,
+      "step": 6553
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5133652823757594e-05,
+      "loss": 1.715,
+      "step": 6554
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5120439168627504e-05,
+      "loss": 0.71,
+      "step": 6555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5107230256938509e-05,
+      "loss": 1.0815,
+      "step": 6556
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5094026090486957e-05,
+      "loss": 1.3234,
+      "step": 6557
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5080826671068548e-05,
+      "loss": 1.4233,
+      "step": 6558
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5067632000478332e-05,
+      "loss": 1.058,
+      "step": 6559
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5054442080510721e-05,
+      "loss": 1.2916,
+      "step": 6560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5041256912959467e-05,
+      "loss": 0.8963,
+      "step": 6561
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5028076499617693e-05,
+      "loss": 1.0606,
+      "step": 6562
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5014900842277868e-05,
+      "loss": 0.8931,
+      "step": 6563
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5001729942731812e-05,
+      "loss": 1.4972,
+      "step": 6564
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4988563802770695e-05,
+      "loss": 1.2442,
+      "step": 6565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4975402424185053e-05,
+      "loss": 1.4136,
+      "step": 6566
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4962245808764758e-05,
+      "loss": 1.0734,
+      "step": 6567
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4949093958299042e-05,
+      "loss": 1.3799,
+      "step": 6568
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4935946874576495e-05,
+      "loss": 0.6637,
+      "step": 6569
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.492280455938505e-05,
+      "loss": 0.9627,
+      "step": 6570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4909667014511997e-05,
+      "loss": 1.4109,
+      "step": 6571
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4896534241743988e-05,
+      "loss": 1.0408,
+      "step": 6572
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4883406242866965e-05,
+      "loss": 0.6997,
+      "step": 6573
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4870283019666315e-05,
+      "loss": 1.1313,
+      "step": 6574
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4857164573926713e-05,
+      "loss": 0.8888,
+      "step": 6575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4844050907432206e-05,
+      "loss": 1.0733,
+      "step": 6576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4830942021966177e-05,
+      "loss": 1.1082,
+      "step": 6577
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4817837919311373e-05,
+      "loss": 1.166,
+      "step": 6578
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4804738601249896e-05,
+      "loss": 0.6191,
+      "step": 6579
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4791644069563138e-05,
+      "loss": 1.1804,
+      "step": 6580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4778554326031947e-05,
+      "loss": 1.4691,
+      "step": 6581
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4765469372436431e-05,
+      "loss": 1.3148,
+      "step": 6582
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4752389210556095e-05,
+      "loss": 1.3797,
+      "step": 6583
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4739313842169733e-05,
+      "loss": 1.3044,
+      "step": 6584
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4726243269055568e-05,
+      "loss": 1.3115,
+      "step": 6585
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4713177492991136e-05,
+      "loss": 1.2346,
+      "step": 6586
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4700116515753282e-05,
+      "loss": 0.896,
+      "step": 6587
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4687060339118235e-05,
+      "loss": 1.2956,
+      "step": 6588
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.467400896486159e-05,
+      "loss": 0.9024,
+      "step": 6589
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4660962394758282e-05,
+      "loss": 1.5598,
+      "step": 6590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.464792063058253e-05,
+      "loss": 0.6577,
+      "step": 6591
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4634883674107957e-05,
+      "loss": 1.2995,
+      "step": 6592
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.462185152710755e-05,
+      "loss": 1.1759,
+      "step": 6593
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4608824191353615e-05,
+      "loss": 1.1567,
+      "step": 6594
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.459580166861777e-05,
+      "loss": 1.1313,
+      "step": 6595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.458278396067101e-05,
+      "loss": 1.3301,
+      "step": 6596
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4569771069283727e-05,
+      "loss": 1.4224,
+      "step": 6597
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4556762996225565e-05,
+      "loss": 1.2589,
+      "step": 6598
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4543759743265545e-05,
+      "loss": 1.3336,
+      "step": 6599
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4530761312172075e-05,
+      "loss": 0.8347,
+      "step": 6600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.451776770471288e-05,
+      "loss": 0.5741,
+      "step": 6601
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4504778922654994e-05,
+      "loss": 0.4719,
+      "step": 6602
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4491794967764822e-05,
+      "loss": 0.7016,
+      "step": 6603
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4478815841808157e-05,
+      "loss": 1.1695,
+      "step": 6604
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4465841546550052e-05,
+      "loss": 1.0501,
+      "step": 6605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4452872083754959e-05,
+      "loss": 1.1881,
+      "step": 6606
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.443990745518664e-05,
+      "loss": 0.7253,
+      "step": 6607
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.442694766260827e-05,
+      "loss": 0.6751,
+      "step": 6608
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4413992707782265e-05,
+      "loss": 0.6678,
+      "step": 6609
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4401042592470449e-05,
+      "loss": 1.1414,
+      "step": 6610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4388097318433968e-05,
+      "loss": 0.758,
+      "step": 6611
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.437515688743331e-05,
+      "loss": 0.7073,
+      "step": 6612
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4362221301228318e-05,
+      "loss": 1.7385,
+      "step": 6613
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4349290561578161e-05,
+      "loss": 1.4104,
+      "step": 6614
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4336364670241347e-05,
+      "loss": 1.3124,
+      "step": 6615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4323443628975736e-05,
+      "loss": 0.5667,
+      "step": 6616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4310527439538524e-05,
+      "loss": 0.942,
+      "step": 6617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.429761610368624e-05,
+      "loss": 1.3298,
+      "step": 6618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4284709623174769e-05,
+      "loss": 0.6142,
+      "step": 6619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4271807999759319e-05,
+      "loss": 1.313,
+      "step": 6620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4258911235194444e-05,
+      "loss": 0.9014,
+      "step": 6621
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4246019331234046e-05,
+      "loss": 1.0666,
+      "step": 6622
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4233132289631346e-05,
+      "loss": 0.9225,
+      "step": 6623
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4220250112138927e-05,
+      "loss": 0.9515,
+      "step": 6624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.420737280050869e-05,
+      "loss": 1.0177,
+      "step": 6625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4194500356491886e-05,
+      "loss": 1.0242,
+      "step": 6626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4181632781839104e-05,
+      "loss": 1.1295,
+      "step": 6627
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4168770078300265e-05,
+      "loss": 1.1158,
+      "step": 6628
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4155912247624631e-05,
+      "loss": 1.0507,
+      "step": 6629
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4143059291560807e-05,
+      "loss": 1.4588,
+      "step": 6630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4130211211856714e-05,
+      "loss": 1.0816,
+      "step": 6631
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4117368010259641e-05,
+      "loss": 0.5709,
+      "step": 6632
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4104529688516204e-05,
+      "loss": 1.2646,
+      "step": 6633
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4091696248372305e-05,
+      "loss": 0.5603,
+      "step": 6634
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4078867691573272e-05,
+      "loss": 1.2001,
+      "step": 6635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4066044019863711e-05,
+      "loss": 0.5727,
+      "step": 6636
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4053225234987583e-05,
+      "loss": 1.1938,
+      "step": 6637
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4040411338688137e-05,
+      "loss": 0.7501,
+      "step": 6638
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.402760233270804e-05,
+      "loss": 1.3648,
+      "step": 6639
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4014798218789249e-05,
+      "loss": 1.3324,
+      "step": 6640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4001998998673021e-05,
+      "loss": 1.4136,
+      "step": 6641
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.398920467410002e-05,
+      "loss": 1.1224,
+      "step": 6642
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3976415246810198e-05,
+      "loss": 1.5465,
+      "step": 6643
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3963630718542869e-05,
+      "loss": 1.6849,
+      "step": 6644
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3950851091036609e-05,
+      "loss": 0.7535,
+      "step": 6645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3938076366029434e-05,
+      "loss": 0.5855,
+      "step": 6646
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3925306545258642e-05,
+      "loss": 1.2476,
+      "step": 6647
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3912541630460829e-05,
+      "loss": 1.0892,
+      "step": 6648
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.389978162337196e-05,
+      "loss": 1.4078,
+      "step": 6649
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3887026525727359e-05,
+      "loss": 0.6461,
+      "step": 6650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3874276339261655e-05,
+      "loss": 1.2162,
+      "step": 6651
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3861531065708777e-05,
+      "loss": 1.1471,
+      "step": 6652
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3848790706802018e-05,
+      "loss": 0.6908,
+      "step": 6653
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3836055264274046e-05,
+      "loss": 0.9862,
+      "step": 6654
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3823324739856763e-05,
+      "loss": 1.4267,
+      "step": 6655
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3810599135281487e-05,
+      "loss": 0.6098,
+      "step": 6656
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3797878452278806e-05,
+      "loss": 1.3449,
+      "step": 6657
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3785162692578713e-05,
+      "loss": 1.1452,
+      "step": 6658
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3772451857910446e-05,
+      "loss": 0.7159,
+      "step": 6659
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3759745950002629e-05,
+      "loss": 1.477,
+      "step": 6660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.37470449705832e-05,
+      "loss": 0.7725,
+      "step": 6661
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3734348921379419e-05,
+      "loss": 1.4175,
+      "step": 6662
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3721657804117893e-05,
+      "loss": 1.4581,
+      "step": 6663
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.370897162052453e-05,
+      "loss": 1.0355,
+      "step": 6664
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3696290372324632e-05,
+      "loss": 0.5072,
+      "step": 6665
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3683614061242734e-05,
+      "loss": 1.0917,
+      "step": 6666
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3670942689002769e-05,
+      "loss": 1.2551,
+      "step": 6667
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.365827625732798e-05,
+      "loss": 1.0435,
+      "step": 6668
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.364561476794093e-05,
+      "loss": 1.1241,
+      "step": 6669
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.363295822256352e-05,
+      "loss": 1.3364,
+      "step": 6670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3620306622916978e-05,
+      "loss": 0.9588,
+      "step": 6671
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3607659970721853e-05,
+      "loss": 1.3688,
+      "step": 6672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.359501826769803e-05,
+      "loss": 1.0081,
+      "step": 6673
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3582381515564706e-05,
+      "loss": 1.1646,
+      "step": 6674
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3569749716040419e-05,
+      "loss": 0.8255,
+      "step": 6675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3557122870843036e-05,
+      "loss": 1.1469,
+      "step": 6676
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3544500981689734e-05,
+      "loss": 0.8834,
+      "step": 6677
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3531884050297023e-05,
+      "loss": 1.1884,
+      "step": 6678
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.351927207838075e-05,
+      "loss": 0.9635,
+      "step": 6679
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3506665067656077e-05,
+      "loss": 1.3866,
+      "step": 6680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3494063019837488e-05,
+      "loss": 0.8388,
+      "step": 6681
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3481465936638793e-05,
+      "loss": 0.8375,
+      "step": 6682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3468873819773142e-05,
+      "loss": 1.117,
+      "step": 6683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3456286670952994e-05,
+      "loss": 1.077,
+      "step": 6684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.344370449189013e-05,
+      "loss": 0.8917,
+      "step": 6685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.343112728429567e-05,
+      "loss": 0.885,
+      "step": 6686
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3418555049880043e-05,
+      "loss": 1.2523,
+      "step": 6687
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3405987790353014e-05,
+      "loss": 1.2571,
+      "step": 6688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3393425507423662e-05,
+      "loss": 0.9396,
+      "step": 6689
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3380868202800411e-05,
+      "loss": 0.8306,
+      "step": 6690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3368315878190946e-05,
+      "loss": 1.2428,
+      "step": 6691
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3355768535302353e-05,
+      "loss": 1.2473,
+      "step": 6692
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3343226175841006e-05,
+      "loss": 1.279,
+      "step": 6693
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3330688801512609e-05,
+      "loss": 1.2958,
+      "step": 6694
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.331815641402213e-05,
+      "loss": 1.2168,
+      "step": 6695
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3305629015073967e-05,
+      "loss": 1.2523,
+      "step": 6696
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3293106606371774e-05,
+      "loss": 1.681,
+      "step": 6697
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3280589189618503e-05,
+      "loss": 0.5455,
+      "step": 6698
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3268076766516462e-05,
+      "loss": 1.3647,
+      "step": 6699
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.325556933876731e-05,
+      "loss": 1.4301,
+      "step": 6700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3243066908071984e-05,
+      "loss": 1.2867,
+      "step": 6701
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3230569476130722e-05,
+      "loss": 1.246,
+      "step": 6702
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3218077044643117e-05,
+      "loss": 1.2806,
+      "step": 6703
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3205589615308112e-05,
+      "loss": 1.3506,
+      "step": 6704
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.31931071898239e-05,
+      "loss": 1.1081,
+      "step": 6705
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3180629769888014e-05,
+      "loss": 1.1348,
+      "step": 6706
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3168157357197353e-05,
+      "loss": 1.2371,
+      "step": 6707
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3155689953448109e-05,
+      "loss": 1.1521,
+      "step": 6708
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.314322756033574e-05,
+      "loss": 1.415,
+      "step": 6709
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.313077017955508e-05,
+      "loss": 1.2946,
+      "step": 6710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3118317812800313e-05,
+      "loss": 0.5048,
+      "step": 6711
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3105870461764847e-05,
+      "loss": 1.0242,
+      "step": 6712
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3093428128141471e-05,
+      "loss": 1.0719,
+      "step": 6713
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3080990813622268e-05,
+      "loss": 1.386,
+      "step": 6714
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3068558519898693e-05,
+      "loss": 1.2903,
+      "step": 6715
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.305613124866143e-05,
+      "loss": 1.0705,
+      "step": 6716
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3043709001600535e-05,
+      "loss": 1.367,
+      "step": 6717
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3031291780405375e-05,
+      "loss": 0.5898,
+      "step": 6718
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3018879586764626e-05,
+      "loss": 1.4771,
+      "step": 6719
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3006472422366284e-05,
+      "loss": 0.8516,
+      "step": 6720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2994070288897659e-05,
+      "loss": 1.5064,
+      "step": 6721
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2981673188045374e-05,
+      "loss": 0.9709,
+      "step": 6722
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.296928112149537e-05,
+      "loss": 1.3397,
+      "step": 6723
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2956894090932914e-05,
+      "loss": 0.5933,
+      "step": 6724
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2944512098042567e-05,
+      "loss": 1.0749,
+      "step": 6725
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2932135144508229e-05,
+      "loss": 0.6108,
+      "step": 6726
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2919763232013088e-05,
+      "loss": 1.19,
+      "step": 6727
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2907396362239671e-05,
+      "loss": 1.4245,
+      "step": 6728
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2895034536869805e-05,
+      "loss": 0.6614,
+      "step": 6729
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2882677757584633e-05,
+      "loss": 0.6768,
+      "step": 6730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2870326026064622e-05,
+      "loss": 1.4795,
+      "step": 6731
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.285797934398953e-05,
+      "loss": 1.3269,
+      "step": 6732
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2845637713038455e-05,
+      "loss": 1.2722,
+      "step": 6733
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2833301134889786e-05,
+      "loss": 0.8309,
+      "step": 6734
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2820969611221245e-05,
+      "loss": 1.2667,
+      "step": 6735
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2808643143709842e-05,
+      "loss": 0.8372,
+      "step": 6736
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2796321734031929e-05,
+      "loss": 1.2277,
+      "step": 6737
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.278400538386314e-05,
+      "loss": 1.35,
+      "step": 6738
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2771694094878444e-05,
+      "loss": 1.5955,
+      "step": 6739
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2759387868752127e-05,
+      "loss": 1.4929,
+      "step": 6740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2747086707157723e-05,
+      "loss": 1.4622,
+      "step": 6741
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2734790611768177e-05,
+      "loss": 1.0031,
+      "step": 6742
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2722499584255676e-05,
+      "loss": 1.3611,
+      "step": 6743
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2710213626291751e-05,
+      "loss": 0.8325,
+      "step": 6744
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2697932739547192e-05,
+      "loss": 1.4446,
+      "step": 6745
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2685656925692175e-05,
+      "loss": 0.9268,
+      "step": 6746
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2673386186396147e-05,
+      "loss": 1.0225,
+      "step": 6747
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2661120523327824e-05,
+      "loss": 1.421,
+      "step": 6748
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2648859938155311e-05,
+      "loss": 0.9146,
+      "step": 6749
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2636604432545978e-05,
+      "loss": 1.5023,
+      "step": 6750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2624354008166527e-05,
+      "loss": 1.1563,
+      "step": 6751
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2612108666682898e-05,
+      "loss": 1.1138,
+      "step": 6752
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2599868409760452e-05,
+      "loss": 1.492,
+      "step": 6753
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2587633239063784e-05,
+      "loss": 1.2376,
+      "step": 6754
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2575403156256804e-05,
+      "loss": 0.8168,
+      "step": 6755
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2563178163002726e-05,
+      "loss": 1.2497,
+      "step": 6756
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2550958260964125e-05,
+      "loss": 0.6632,
+      "step": 6757
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2538743451802837e-05,
+      "loss": 1.3985,
+      "step": 6758
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2526533737179985e-05,
+      "loss": 0.5758,
+      "step": 6759
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2514329118756036e-05,
+      "loss": 1.3889,
+      "step": 6760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2502129598190793e-05,
+      "loss": 1.2328,
+      "step": 6761
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2489935177143281e-05,
+      "loss": 1.433,
+      "step": 6762
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2477745857271899e-05,
+      "loss": 1.4226,
+      "step": 6763
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2465561640234313e-05,
+      "loss": 1.2988,
+      "step": 6764
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2453382527687563e-05,
+      "loss": 0.8039,
+      "step": 6765
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.24412085212879e-05,
+      "loss": 1.3199,
+      "step": 6766
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2429039622690924e-05,
+      "loss": 1.2793,
+      "step": 6767
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2416875833551589e-05,
+      "loss": 0.7562,
+      "step": 6768
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.240471715552406e-05,
+      "loss": 0.9674,
+      "step": 6769
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2392563590261875e-05,
+      "loss": 1.0812,
+      "step": 6770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2380415139417839e-05,
+      "loss": 1.0649,
+      "step": 6771
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.236827180464412e-05,
+      "loss": 1.3382,
+      "step": 6772
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2356133587592111e-05,
+      "loss": 1.076,
+      "step": 6773
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2344000489912555e-05,
+      "loss": 1.0484,
+      "step": 6774
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.23318725132555e-05,
+      "loss": 0.555,
+      "step": 6775
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.231974965927028e-05,
+      "loss": 1.019,
+      "step": 6776
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2307631929605546e-05,
+      "loss": 1.0648,
+      "step": 6777
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2295519325909243e-05,
+      "loss": 1.1588,
+      "step": 6778
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.228341184982863e-05,
+      "loss": 1.2474,
+      "step": 6779
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2271309503010253e-05,
+      "loss": 0.8121,
+      "step": 6780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.225921228709997e-05,
+      "loss": 0.7528,
+      "step": 6781
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2247120203742945e-05,
+      "loss": 1.0009,
+      "step": 6782
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2235033254583639e-05,
+      "loss": 0.9521,
+      "step": 6783
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2222951441265813e-05,
+      "loss": 0.9937,
+      "step": 6784
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2210874765432528e-05,
+      "loss": 1.5332,
+      "step": 6785
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2198803228726157e-05,
+      "loss": 0.8626,
+      "step": 6786
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2186736832788365e-05,
+      "loss": 0.8362,
+      "step": 6787
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2174675579260114e-05,
+      "loss": 1.2466,
+      "step": 6788
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2162619469781682e-05,
+      "loss": 1.3406,
+      "step": 6789
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2150568505992627e-05,
+      "loss": 0.5762,
+      "step": 6790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2138522689531829e-05,
+      "loss": 1.0982,
+      "step": 6791
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.212648202203745e-05,
+      "loss": 1.3368,
+      "step": 6792
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.211444650514696e-05,
+      "loss": 0.8645,
+      "step": 6793
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.210241614049713e-05,
+      "loss": 1.1694,
+      "step": 6794
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2090390929724027e-05,
+      "loss": 0.797,
+      "step": 6795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2078370874463014e-05,
+      "loss": 0.6585,
+      "step": 6796
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2066355976348775e-05,
+      "loss": 1.1417,
+      "step": 6797
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2054346237015235e-05,
+      "loss": 0.7062,
+      "step": 6798
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2042341658095691e-05,
+      "loss": 1.0723,
+      "step": 6799
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2030342241222697e-05,
+      "loss": 1.1253,
+      "step": 6800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2018347988028122e-05,
+      "loss": 1.2151,
+      "step": 6801
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2006358900143083e-05,
+      "loss": 1.314,
+      "step": 6802
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1994374979198075e-05,
+      "loss": 1.3914,
+      "step": 6803
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.198239622682284e-05,
+      "loss": 1.0305,
+      "step": 6804
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1970422644646434e-05,
+      "loss": 1.4545,
+      "step": 6805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1958454234297162e-05,
+      "loss": 1.128,
+      "step": 6806
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.194649099740271e-05,
+      "loss": 1.1467,
+      "step": 6807
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.193453293559002e-05,
+      "loss": 0.8536,
+      "step": 6808
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1922580050485293e-05,
+      "loss": 1.4752,
+      "step": 6809
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1910632343714067e-05,
+      "loss": 0.9041,
+      "step": 6810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1898689816901187e-05,
+      "loss": 1.1697,
+      "step": 6811
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1886752471670786e-05,
+      "loss": 1.5302,
+      "step": 6812
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1874820309646228e-05,
+      "loss": 0.6867,
+      "step": 6813
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1862893332450281e-05,
+      "loss": 1.1912,
+      "step": 6814
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1850971541704942e-05,
+      "loss": 0.8946,
+      "step": 6815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.183905493903149e-05,
+      "loss": 0.401,
+      "step": 6816
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1827143526050522e-05,
+      "loss": 0.4846,
+      "step": 6817
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1815237304381955e-05,
+      "loss": 1.0674,
+      "step": 6818
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1803336275644978e-05,
+      "loss": 0.6442,
+      "step": 6819
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1791440441458036e-05,
+      "loss": 1.1888,
+      "step": 6820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.17795498034389e-05,
+      "loss": 1.4327,
+      "step": 6821
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1767664363204684e-05,
+      "loss": 1.4876,
+      "step": 6822
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1755784122371705e-05,
+      "loss": 0.8904,
+      "step": 6823
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1743909082555627e-05,
+      "loss": 0.9718,
+      "step": 6824
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1732039245371373e-05,
+      "loss": 1.002,
+      "step": 6825
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1720174612433233e-05,
+      "loss": 0.5961,
+      "step": 6826
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.170831518535469e-05,
+      "loss": 1.2868,
+      "step": 6827
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1696460965748578e-05,
+      "loss": 0.9868,
+      "step": 6828
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.168461195522701e-05,
+      "loss": 0.7016,
+      "step": 6829
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1672768155401393e-05,
+      "loss": 1.1344,
+      "step": 6830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1660929567882423e-05,
+      "loss": 1.0382,
+      "step": 6831
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1649096194280068e-05,
+      "loss": 1.3739,
+      "step": 6832
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1637268036203652e-05,
+      "loss": 1.3861,
+      "step": 6833
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.16254450952617e-05,
+      "loss": 1.3017,
+      "step": 6834
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1613627373062086e-05,
+      "loss": 1.1563,
+      "step": 6835
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1601814871211963e-05,
+      "loss": 1.4071,
+      "step": 6836
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1590007591317765e-05,
+      "loss": 1.0606,
+      "step": 6837
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1578205534985226e-05,
+      "loss": 1.0033,
+      "step": 6838
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.156640870381936e-05,
+      "loss": 0.9926,
+      "step": 6839
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1554617099424481e-05,
+      "loss": 0.9607,
+      "step": 6840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.154283072340418e-05,
+      "loss": 0.748,
+      "step": 6841
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1531049577361353e-05,
+      "loss": 0.6763,
+      "step": 6842
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1519273662898167e-05,
+      "loss": 1.246,
+      "step": 6843
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1507502981616086e-05,
+      "loss": 1.294,
+      "step": 6844
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1495737535115868e-05,
+      "loss": 1.2842,
+      "step": 6845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1483977324997552e-05,
+      "loss": 1.2281,
+      "step": 6846
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1472222352860462e-05,
+      "loss": 1.217,
+      "step": 6847
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1460472620303215e-05,
+      "loss": 1.2775,
+      "step": 6848
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.144872812892372e-05,
+      "loss": 1.2017,
+      "step": 6849
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1436988880319155e-05,
+      "loss": 0.6933,
+      "step": 6850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1425254876086028e-05,
+      "loss": 0.9169,
+      "step": 6851
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.141352611782005e-05,
+      "loss": 0.759,
+      "step": 6852
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1401802607116313e-05,
+      "loss": 0.804,
+      "step": 6853
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1390084345569147e-05,
+      "loss": 1.1252,
+      "step": 6854
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.137837133477217e-05,
+      "loss": 0.9515,
+      "step": 6855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1366663576318298e-05,
+      "loss": 1.3036,
+      "step": 6856
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1354961071799719e-05,
+      "loss": 0.819,
+      "step": 6857
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1343263822807931e-05,
+      "loss": 1.4453,
+      "step": 6858
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1331571830933657e-05,
+      "loss": 0.5333,
+      "step": 6859
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1319885097766986e-05,
+      "loss": 0.5282,
+      "step": 6860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1308203624897245e-05,
+      "loss": 1.4742,
+      "step": 6861
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1296527413913066e-05,
+      "loss": 0.9288,
+      "step": 6862
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1284856466402317e-05,
+      "loss": 1.0218,
+      "step": 6863
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.127319078395222e-05,
+      "loss": 0.9327,
+      "step": 6864
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1261530368149253e-05,
+      "loss": 1.594,
+      "step": 6865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1249875220579142e-05,
+      "loss": 1.2631,
+      "step": 6866
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1238225342826936e-05,
+      "loss": 1.554,
+      "step": 6867
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.122658073647697e-05,
+      "loss": 0.4898,
+      "step": 6868
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1214941403112867e-05,
+      "loss": 1.2489,
+      "step": 6869
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1203307344317476e-05,
+      "loss": 1.3985,
+      "step": 6870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.119167856167298e-05,
+      "loss": 1.2212,
+      "step": 6871
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1180055056760869e-05,
+      "loss": 1.2016,
+      "step": 6872
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1168436831161838e-05,
+      "loss": 1.2113,
+      "step": 6873
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1156823886455913e-05,
+      "loss": 0.5477,
+      "step": 6874
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.114521622422241e-05,
+      "loss": 1.302,
+      "step": 6875
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1133613846039925e-05,
+      "loss": 0.6503,
+      "step": 6876
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1122016753486286e-05,
+      "loss": 1.226,
+      "step": 6877
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1110424948138637e-05,
+      "loss": 0.9818,
+      "step": 6878
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.109883843157345e-05,
+      "loss": 1.0719,
+      "step": 6879
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1087257205366391e-05,
+      "loss": 0.9642,
+      "step": 6880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1075681271092459e-05,
+      "loss": 0.9816,
+      "step": 6881
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1064110630325902e-05,
+      "loss": 1.2529,
+      "step": 6882
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.105254528464032e-05,
+      "loss": 0.7777,
+      "step": 6883
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1040985235608491e-05,
+      "loss": 0.951,
+      "step": 6884
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1029430484802538e-05,
+      "loss": 0.4794,
+      "step": 6885
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1017881033793853e-05,
+      "loss": 1.2075,
+      "step": 6886
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1006336884153096e-05,
+      "loss": 0.5394,
+      "step": 6887
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.099479803745021e-05,
+      "loss": 1.1941,
+      "step": 6888
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0983264495254425e-05,
+      "loss": 0.7482,
+      "step": 6889
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.097173625913424e-05,
+      "loss": 1.252,
+      "step": 6890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0960213330657437e-05,
+      "loss": 1.0181,
+      "step": 6891
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0948695711391077e-05,
+      "loss": 0.8857,
+      "step": 6892
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0937183402901497e-05,
+      "loss": 0.9965,
+      "step": 6893
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0925676406754304e-05,
+      "loss": 1.2874,
+      "step": 6894
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0914174724514404e-05,
+      "loss": 1.1531,
+      "step": 6895
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0902678357745954e-05,
+      "loss": 1.149,
+      "step": 6896
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0891187308012407e-05,
+      "loss": 1.2993,
+      "step": 6897
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0879701576876483e-05,
+      "loss": 1.2853,
+      "step": 6898
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.086822116590019e-05,
+      "loss": 1.0884,
+      "step": 6899
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.085674607664479e-05,
+      "loss": 1.7586,
+      "step": 6900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0845276310670855e-05,
+      "loss": 1.285,
+      "step": 6901
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0833811869538202e-05,
+      "loss": 1.1931,
+      "step": 6902
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0822352754805942e-05,
+      "loss": 1.1989,
+      "step": 6903
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.081089896803245e-05,
+      "loss": 1.2114,
+      "step": 6904
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0799450510775389e-05,
+      "loss": 1.3587,
+      "step": 6905
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0788007384591686e-05,
+      "loss": 1.3355,
+      "step": 6906
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0776569591037544e-05,
+      "loss": 0.9469,
+      "step": 6907
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0765137131668467e-05,
+      "loss": 0.7816,
+      "step": 6908
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0753710008039163e-05,
+      "loss": 0.7144,
+      "step": 6909
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0742288221703706e-05,
+      "loss": 0.7773,
+      "step": 6910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0730871774215378e-05,
+      "loss": 0.342,
+      "step": 6911
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0719460667126785e-05,
+      "loss": 0.6255,
+      "step": 6912
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0708054901989733e-05,
+      "loss": 1.6846,
+      "step": 6913
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0696654480355384e-05,
+      "loss": 1.0081,
+      "step": 6914
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.068525940377414e-05,
+      "loss": 1.4992,
+      "step": 6915
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0673869673795627e-05,
+      "loss": 0.8383,
+      "step": 6916
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0662485291968837e-05,
+      "loss": 0.8204,
+      "step": 6917
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0651106259841975e-05,
+      "loss": 1.2134,
+      "step": 6918
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.063973257896254e-05,
+      "loss": 1.1385,
+      "step": 6919
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0628364250877254e-05,
+      "loss": 1.272,
+      "step": 6920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0617001277132193e-05,
+      "loss": 1.3903,
+      "step": 6921
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0605643659272668e-05,
+      "loss": 0.6756,
+      "step": 6922
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.059429139884322e-05,
+      "loss": 1.4361,
+      "step": 6923
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0582944497387698e-05,
+      "loss": 0.6477,
+      "step": 6924
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0571602956449261e-05,
+      "loss": 1.3299,
+      "step": 6925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0560266777570287e-05,
+      "loss": 1.2524,
+      "step": 6926
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0548935962292423e-05,
+      "loss": 0.7882,
+      "step": 6927
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0537610512156592e-05,
+      "loss": 1.3361,
+      "step": 6928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0526290428703045e-05,
+      "loss": 1.2668,
+      "step": 6929
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0514975713471208e-05,
+      "loss": 1.5651,
+      "step": 6930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0503666367999843e-05,
+      "loss": 0.9871,
+      "step": 6931
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0492362393826943e-05,
+      "loss": 0.8521,
+      "step": 6932
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0481063792489842e-05,
+      "loss": 0.7047,
+      "step": 6933
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.046977056552504e-05,
+      "loss": 1.182,
+      "step": 6934
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0458482714468377e-05,
+      "loss": 0.9539,
+      "step": 6935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0447200240854937e-05,
+      "loss": 1.1483,
+      "step": 6936
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.043592314621909e-05,
+      "loss": 1.1821,
+      "step": 6937
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.042465143209445e-05,
+      "loss": 1.0994,
+      "step": 6938
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0413385100013906e-05,
+      "loss": 1.1035,
+      "step": 6939
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0402124151509663e-05,
+      "loss": 1.2736,
+      "step": 6940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0390868588113107e-05,
+      "loss": 0.6725,
+      "step": 6941
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0379618411354951e-05,
+      "loss": 1.3057,
+      "step": 6942
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.036837362276517e-05,
+      "loss": 0.6554,
+      "step": 6943
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0357134223872982e-05,
+      "loss": 1.0912,
+      "step": 6944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.03459002162069e-05,
+      "loss": 1.1815,
+      "step": 6945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0334671601294681e-05,
+      "loss": 0.9524,
+      "step": 6946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0323448380663369e-05,
+      "loss": 1.1264,
+      "step": 6947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0312230555839264e-05,
+      "loss": 1.2313,
+      "step": 6948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0301018128347922e-05,
+      "loss": 1.0943,
+      "step": 6949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0289811099714187e-05,
+      "loss": 0.7022,
+      "step": 6950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0278609471462147e-05,
+      "loss": 1.2181,
+      "step": 6951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0267413245115177e-05,
+      "loss": 1.2235,
+      "step": 6952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0256222422195899e-05,
+      "loss": 1.1481,
+      "step": 6953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.024503700422621e-05,
+      "loss": 0.9855,
+      "step": 6954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.023385699272727e-05,
+      "loss": 1.0689,
+      "step": 6955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0222682389219495e-05,
+      "loss": 1.3591,
+      "step": 6956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0211513195222588e-05,
+      "loss": 0.5273,
+      "step": 6957
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.020034941225549e-05,
+      "loss": 1.122,
+      "step": 6958
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0189191041836427e-05,
+      "loss": 1.007,
+      "step": 6959
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0178038085482872e-05,
+      "loss": 1.127,
+      "step": 6960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.016689054471157e-05,
+      "loss": 0.9279,
+      "step": 6961
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0155748421038536e-05,
+      "loss": 1.1211,
+      "step": 6962
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0144611715979042e-05,
+      "loss": 1.767,
+      "step": 6963
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0133480431047615e-05,
+      "loss": 1.1775,
+      "step": 6964
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0122354567758069e-05,
+      "loss": 0.5794,
+      "step": 6965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0111234127623425e-05,
+      "loss": 1.2214,
+      "step": 6966
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0100119112156048e-05,
+      "loss": 1.0041,
+      "step": 6967
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.008900952286751e-05,
+      "loss": 1.1133,
+      "step": 6968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0077905361268664e-05,
+      "loss": 1.4311,
+      "step": 6969
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0066806628869586e-05,
+      "loss": 0.894,
+      "step": 6970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0055713327179683e-05,
+      "loss": 1.0584,
+      "step": 6971
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.004462545770759e-05,
+      "loss": 1.0649,
+      "step": 6972
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0033543021961173e-05,
+      "loss": 1.1408,
+      "step": 6973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0022466021447585e-05,
+      "loss": 1.0073,
+      "step": 6974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0011394457673268e-05,
+      "loss": 1.4289,
+      "step": 6975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0000328332143894e-05,
+      "loss": 0.7523,
+      "step": 6976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.989267646364382e-06,
+      "loss": 0.8409,
+      "step": 6977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.978212401838922e-06,
+      "loss": 1.5746,
+      "step": 6978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.967162600071006e-06,
+      "loss": 1.134,
+      "step": 6979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.956118242563323e-06,
+      "loss": 1.4138,
+      "step": 6980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.945079330817837e-06,
+      "loss": 1.1576,
+      "step": 6981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.934045866335817e-06,
+      "loss": 1.3837,
+      "step": 6982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.923017850617755e-06,
+      "loss": 1.1126,
+      "step": 6983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.911995285163378e-06,
+      "loss": 0.7296,
+      "step": 6984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.900978171471697e-06,
+      "loss": 0.889,
+      "step": 6985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.889966511041027e-06,
+      "loss": 0.9878,
+      "step": 6986
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.87896030536885e-06,
+      "loss": 1.1495,
+      "step": 6987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.867959555951978e-06,
+      "loss": 0.5698,
+      "step": 6988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.856964264286428e-06,
+      "loss": 0.8698,
+      "step": 6989
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.845974431867555e-06,
+      "loss": 1.3279,
+      "step": 6990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.834990060189874e-06,
+      "loss": 0.9973,
+      "step": 6991
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.824011150747226e-06,
+      "loss": 1.3436,
+      "step": 6992
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.813037705032658e-06,
+      "loss": 1.2049,
+      "step": 6993
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.802069724538553e-06,
+      "loss": 0.7812,
+      "step": 6994
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.791107210756455e-06,
+      "loss": 0.9813,
+      "step": 6995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.780150165177227e-06,
+      "loss": 0.7113,
+      "step": 6996
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.769198589290968e-06,
+      "loss": 1.1028,
+      "step": 6997
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.758252484587032e-06,
+      "loss": 0.6312,
+      "step": 6998
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.747311852554043e-06,
+      "loss": 1.3344,
+      "step": 6999
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.736376694679867e-06,
+      "loss": 0.5998,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.725447012451622e-06,
+      "loss": 1.2722,
+      "step": 7001
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.714522807355696e-06,
+      "loss": 0.5162,
+      "step": 7002
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.703604080877727e-06,
+      "loss": 0.9897,
+      "step": 7003
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.6926908345026e-06,
+      "loss": 1.3885,
+      "step": 7004
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.681783069714462e-06,
+      "loss": 1.1584,
+      "step": 7005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.67088078799671e-06,
+      "loss": 0.7974,
+      "step": 7006
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.659983990832005e-06,
+      "loss": 1.0603,
+      "step": 7007
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.649092679702255e-06,
+      "loss": 0.5831,
+      "step": 7008
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.63820685608861e-06,
+      "loss": 0.94,
+      "step": 7009
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.627326521471502e-06,
+      "loss": 1.4686,
+      "step": 7010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.616451677330595e-06,
+      "loss": 0.5509,
+      "step": 7011
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.605582325144807e-06,
+      "loss": 1.4936,
+      "step": 7012
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.594718466392322e-06,
+      "loss": 1.0538,
+      "step": 7013
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.583860102550562e-06,
+      "loss": 1.4818,
+      "step": 7014
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.573007235096216e-06,
+      "loss": 1.363,
+      "step": 7015
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.56215986550521e-06,
+      "loss": 1.3148,
+      "step": 7016
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.551317995252735e-06,
+      "loss": 0.9823,
+      "step": 7017
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.54048162581323e-06,
+      "loss": 0.9272,
+      "step": 7018
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.529650758660402e-06,
+      "loss": 1.2383,
+      "step": 7019
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.518825395267145e-06,
+      "loss": 1.265,
+      "step": 7020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.508005537105696e-06,
+      "loss": 1.3015,
+      "step": 7021
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.497191185647492e-06,
+      "loss": 0.9116,
+      "step": 7022
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.48638234236322e-06,
+      "loss": 0.6096,
+      "step": 7023
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.47557900872284e-06,
+      "loss": 0.7501,
+      "step": 7024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.464781186195537e-06,
+      "loss": 1.417,
+      "step": 7025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.453988876249775e-06,
+      "loss": 1.0561,
+      "step": 7026
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.44320208035322e-06,
+      "loss": 1.2353,
+      "step": 7027
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.432420799972858e-06,
+      "loss": 0.9637,
+      "step": 7028
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.421645036574872e-06,
+      "loss": 1.219,
+      "step": 7029
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.410874791624725e-06,
+      "loss": 1.5906,
+      "step": 7030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.40011006658707e-06,
+      "loss": 1.3823,
+      "step": 7031
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.389350862925905e-06,
+      "loss": 1.1493,
+      "step": 7032
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.37859718210442e-06,
+      "loss": 1.1508,
+      "step": 7033
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.36784902558503e-06,
+      "loss": 1.6872,
+      "step": 7034
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.35710639482944e-06,
+      "loss": 0.9847,
+      "step": 7035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.346369291298606e-06,
+      "loss": 1.5749,
+      "step": 7036
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.335637716452727e-06,
+      "loss": 1.0956,
+      "step": 7037
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.324911671751213e-06,
+      "loss": 1.0545,
+      "step": 7038
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.314191158652757e-06,
+      "loss": 1.2141,
+      "step": 7039
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.303476178615323e-06,
+      "loss": 0.8541,
+      "step": 7040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.292766733096064e-06,
+      "loss": 1.1349,
+      "step": 7041
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.282062823551413e-06,
+      "loss": 1.2053,
+      "step": 7042
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.271364451437042e-06,
+      "loss": 0.5885,
+      "step": 7043
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.260671618207906e-06,
+      "loss": 1.5015,
+      "step": 7044
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.249984325318146e-06,
+      "loss": 1.2564,
+      "step": 7045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.23930257422117e-06,
+      "loss": 1.2716,
+      "step": 7046
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.228626366369681e-06,
+      "loss": 1.1386,
+      "step": 7047
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.217955703215553e-06,
+      "loss": 1.068,
+      "step": 7048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.207290586209954e-06,
+      "loss": 0.6545,
+      "step": 7049
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.196631016803276e-06,
+      "loss": 0.7366,
+      "step": 7050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.185976996445194e-06,
+      "loss": 1.2913,
+      "step": 7051
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.175328526584576e-06,
+      "loss": 1.1844,
+      "step": 7052
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.164685608669554e-06,
+      "loss": 0.9946,
+      "step": 7053
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.154048244147524e-06,
+      "loss": 0.4941,
+      "step": 7054
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.143416434465108e-06,
+      "loss": 0.7139,
+      "step": 7055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.13279018106818e-06,
+      "loss": 1.1283,
+      "step": 7056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.122169485401848e-06,
+      "loss": 1.3757,
+      "step": 7057
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.111554348910484e-06,
+      "loss": 0.919,
+      "step": 7058
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.100944773037684e-06,
+      "loss": 1.5827,
+      "step": 7059
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.090340759226301e-06,
+      "loss": 0.849,
+      "step": 7060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.07974230891842e-06,
+      "loss": 1.1371,
+      "step": 7061
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.069149423555378e-06,
+      "loss": 1.1655,
+      "step": 7062
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.058562104577755e-06,
+      "loss": 0.7443,
+      "step": 7063
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.047980353425372e-06,
+      "loss": 0.4662,
+      "step": 7064
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.037404171537294e-06,
+      "loss": 0.8793,
+      "step": 7065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.026833560351822e-06,
+      "loss": 0.8679,
+      "step": 7066
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.016268521306508e-06,
+      "loss": 1.3839,
+      "step": 7067
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.005709055838147e-06,
+      "loss": 0.9272,
+      "step": 7068
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.995155165382763e-06,
+      "loss": 0.8473,
+      "step": 7069
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.984606851375638e-06,
+      "loss": 1.2067,
+      "step": 7070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.97406411525129e-06,
+      "loss": 0.6508,
+      "step": 7071
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.96352695844347e-06,
+      "loss": 0.6403,
+      "step": 7072
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.952995382385182e-06,
+      "loss": 1.1299,
+      "step": 7073
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.942469388508663e-06,
+      "loss": 0.6812,
+      "step": 7074
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.931948978245398e-06,
+      "loss": 1.0383,
+      "step": 7075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.92143415302612e-06,
+      "loss": 0.3993,
+      "step": 7076
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.91092491428075e-06,
+      "loss": 1.5472,
+      "step": 7077
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.90042126343853e-06,
+      "loss": 1.1161,
+      "step": 7078
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.889923201927897e-06,
+      "loss": 1.2251,
+      "step": 7079
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.879430731176535e-06,
+      "loss": 0.6771,
+      "step": 7080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.868943852611339e-06,
+      "loss": 1.2735,
+      "step": 7081
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.858462567658499e-06,
+      "loss": 0.7219,
+      "step": 7082
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.84798687774342e-06,
+      "loss": 0.9275,
+      "step": 7083
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.837516784290717e-06,
+      "loss": 1.0926,
+      "step": 7084
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.82705228872427e-06,
+      "loss": 1.5369,
+      "step": 7085
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.816593392467214e-06,
+      "loss": 1.4705,
+      "step": 7086
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.806140096941918e-06,
+      "loss": 1.126,
+      "step": 7087
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.795692403569933e-06,
+      "loss": 1.1856,
+      "step": 7088
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.785250313772126e-06,
+      "loss": 1.2668,
+      "step": 7089
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.77481382896857e-06,
+      "loss": 0.5869,
+      "step": 7090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.764382950578549e-06,
+      "loss": 1.4125,
+      "step": 7091
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.753957680020608e-06,
+      "loss": 0.684,
+      "step": 7092
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.743538018712554e-06,
+      "loss": 0.8884,
+      "step": 7093
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.73312396807141e-06,
+      "loss": 0.6146,
+      "step": 7094
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.722715529513409e-06,
+      "loss": 0.9643,
+      "step": 7095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.71231270445404e-06,
+      "loss": 0.6173,
+      "step": 7096
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.701915494308077e-06,
+      "loss": 1.4251,
+      "step": 7097
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.691523900489452e-06,
+      "loss": 1.195,
+      "step": 7098
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.68113792441137e-06,
+      "loss": 1.3575,
+      "step": 7099
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.670757567486265e-06,
+      "loss": 1.0538,
+      "step": 7100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.660382831125852e-06,
+      "loss": 1.421,
+      "step": 7101
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.650013716741002e-06,
+      "loss": 0.7575,
+      "step": 7102
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.63965022574187e-06,
+      "loss": 1.159,
+      "step": 7103
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.629292359537844e-06,
+      "loss": 1.16,
+      "step": 7104
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.618940119537532e-06,
+      "loss": 1.1546,
+      "step": 7105
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.608593507148794e-06,
+      "loss": 1.147,
+      "step": 7106
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.598252523778699e-06,
+      "loss": 1.2738,
+      "step": 7107
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.587917170833609e-06,
+      "loss": 1.0053,
+      "step": 7108
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.577587449719038e-06,
+      "loss": 1.3617,
+      "step": 7109
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.567263361839784e-06,
+      "loss": 1.3366,
+      "step": 7110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.556944908599879e-06,
+      "loss": 1.3391,
+      "step": 7111
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.546632091402573e-06,
+      "loss": 0.6234,
+      "step": 7112
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.536324911650362e-06,
+      "loss": 1.4888,
+      "step": 7113
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.526023370744963e-06,
+      "loss": 1.5637,
+      "step": 7114
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.515727470087332e-06,
+      "loss": 0.77,
+      "step": 7115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.505437211077666e-06,
+      "loss": 1.234,
+      "step": 7116
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.495152595115379e-06,
+      "loss": 1.0694,
+      "step": 7117
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.484873623599137e-06,
+      "loss": 1.2217,
+      "step": 7118
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.474600297926816e-06,
+      "loss": 1.2389,
+      "step": 7119
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.46433261949554e-06,
+      "loss": 1.4313,
+      "step": 7120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.454070589701662e-06,
+      "loss": 1.45,
+      "step": 7121
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.443814209940765e-06,
+      "loss": 0.5389,
+      "step": 7122
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.433563481607659e-06,
+      "loss": 1.1438,
+      "step": 7123
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.423318406096398e-06,
+      "loss": 1.1109,
+      "step": 7124
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.41307898480025e-06,
+      "loss": 0.5019,
+      "step": 7125
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.402845219111748e-06,
+      "loss": 1.0445,
+      "step": 7126
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.392617110422584e-06,
+      "loss": 1.4163,
+      "step": 7127
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.382394660123772e-06,
+      "loss": 0.9105,
+      "step": 7128
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.372177869605496e-06,
+      "loss": 1.5339,
+      "step": 7129
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.361966740257193e-06,
+      "loss": 0.9148,
+      "step": 7130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.351761273467518e-06,
+      "loss": 1.0521,
+      "step": 7131
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.341561470624365e-06,
+      "loss": 1.3352,
+      "step": 7132
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.331367333114864e-06,
+      "loss": 1.1722,
+      "step": 7133
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.321178862325335e-06,
+      "loss": 0.8825,
+      "step": 7134
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.310996059641385e-06,
+      "loss": 0.5109,
+      "step": 7135
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.300818926447823e-06,
+      "loss": 1.2826,
+      "step": 7136
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.290647464128686e-06,
+      "loss": 1.4449,
+      "step": 7137
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.280481674067215e-06,
+      "loss": 0.5769,
+      "step": 7138
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.270321557645933e-06,
+      "loss": 1.2279,
+      "step": 7139
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.260167116246564e-06,
+      "loss": 1.3166,
+      "step": 7140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.250018351250043e-06,
+      "loss": 1.3348,
+      "step": 7141
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.23987526403655e-06,
+      "loss": 1.0574,
+      "step": 7142
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.22973785598551e-06,
+      "loss": 1.1315,
+      "step": 7143
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.21960612847556e-06,
+      "loss": 1.0733,
+      "step": 7144
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.20948008288454e-06,
+      "loss": 0.5356,
+      "step": 7145
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.199359720589539e-06,
+      "loss": 1.7158,
+      "step": 7146
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.189245042966914e-06,
+      "loss": 0.7643,
+      "step": 7147
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.179136051392172e-06,
+      "loss": 1.54,
+      "step": 7148
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.169032747240079e-06,
+      "loss": 0.9258,
+      "step": 7149
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.158935131884664e-06,
+      "loss": 1.0556,
+      "step": 7150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.148843206699148e-06,
+      "loss": 1.0539,
+      "step": 7151
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.13875697305595e-06,
+      "loss": 1.4267,
+      "step": 7152
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.128676432326755e-06,
+      "loss": 1.317,
+      "step": 7153
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.118601585882496e-06,
+      "loss": 1.124,
+      "step": 7154
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.108532435093269e-06,
+      "loss": 0.6791,
+      "step": 7155
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.098468981328438e-06,
+      "loss": 1.4315,
+      "step": 7156
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.088411225956561e-06,
+      "loss": 1.0339,
+      "step": 7157
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.078359170345485e-06,
+      "loss": 1.2256,
+      "step": 7158
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.0683128158622e-06,
+      "loss": 1.2156,
+      "step": 7159
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.058272163872971e-06,
+      "loss": 1.2274,
+      "step": 7160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.048237215743276e-06,
+      "loss": 0.6095,
+      "step": 7161
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.038207972837814e-06,
+      "loss": 1.1722,
+      "step": 7162
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.028184436520508e-06,
+      "loss": 1.0169,
+      "step": 7163
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.018166608154515e-06,
+      "loss": 1.7076,
+      "step": 7164
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.008154489102204e-06,
+      "loss": 0.6715,
+      "step": 7165
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.99814808072517e-06,
+      "loss": 1.266,
+      "step": 7166
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.988147384384237e-06,
+      "loss": 1.1909,
+      "step": 7167
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.978152401439449e-06,
+      "loss": 1.2983,
+      "step": 7168
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.968163133250061e-06,
+      "loss": 1.0276,
+      "step": 7169
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.95817958117458e-06,
+      "loss": 0.9096,
+      "step": 7170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.9482017465707e-06,
+      "loss": 1.4028,
+      "step": 7171
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.938229630795369e-06,
+      "loss": 0.7009,
+      "step": 7172
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.928263235204736e-06,
+      "loss": 0.6696,
+      "step": 7173
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.918302561154183e-06,
+      "loss": 1.6381,
+      "step": 7174
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.908347609998313e-06,
+      "loss": 1.5039,
+      "step": 7175
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.898398383090944e-06,
+      "loss": 1.1565,
+      "step": 7176
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.888454881785118e-06,
+      "loss": 1.609,
+      "step": 7177
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.878517107433103e-06,
+      "loss": 0.5107,
+      "step": 7178
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.868585061386385e-06,
+      "loss": 0.7763,
+      "step": 7179
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.858658744995667e-06,
+      "loss": 1.1529,
+      "step": 7180
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.848738159610886e-06,
+      "loss": 0.9624,
+      "step": 7181
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.838823306581183e-06,
+      "loss": 1.2782,
+      "step": 7182
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.828914187254948e-06,
+      "loss": 1.1511,
+      "step": 7183
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.81901080297972e-06,
+      "loss": 1.167,
+      "step": 7184
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.80911315510236e-06,
+      "loss": 1.0224,
+      "step": 7185
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.799221244968875e-06,
+      "loss": 1.3413,
+      "step": 7186
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.789335073924525e-06,
+      "loss": 1.2975,
+      "step": 7187
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.77945464331375e-06,
+      "loss": 0.6504,
+      "step": 7188
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.769579954480266e-06,
+      "loss": 1.1999,
+      "step": 7189
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.75971100876699e-06,
+      "loss": 0.9367,
+      "step": 7190
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.749847807516014e-06,
+      "loss": 1.4146,
+      "step": 7191
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.739990352068689e-06,
+      "loss": 0.713,
+      "step": 7192
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.7301386437656e-06,
+      "loss": 1.2839,
+      "step": 7193
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.720292683946529e-06,
+      "loss": 1.4163,
+      "step": 7194
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.710452473950435e-06,
+      "loss": 0.6253,
+      "step": 7195
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.700618015115579e-06,
+      "loss": 0.7244,
+      "step": 7196
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.690789308779396e-06,
+      "loss": 1.1024,
+      "step": 7197
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.680966356278512e-06,
+      "loss": 0.9119,
+      "step": 7198
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.671149158948803e-06,
+      "loss": 1.0732,
+      "step": 7199
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.661337718125378e-06,
+      "loss": 1.5766,
+      "step": 7200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.651532035142545e-06,
+      "loss": 1.0423,
+      "step": 7201
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.641732111333793e-06,
+      "loss": 1.1518,
+      "step": 7202
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.631937948031875e-06,
+      "loss": 0.6087,
+      "step": 7203
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.622149546568763e-06,
+      "loss": 0.5571,
+      "step": 7204
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.612366908275636e-06,
+      "loss": 0.7699,
+      "step": 7205
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.6025900344828485e-06,
+      "loss": 1.263,
+      "step": 7206
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.592818926520012e-06,
+      "loss": 1.2445,
+      "step": 7207
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.5830535857159825e-06,
+      "loss": 1.4494,
+      "step": 7208
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.573294013398757e-06,
+      "loss": 0.9629,
+      "step": 7209
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.563540210895598e-06,
+      "loss": 1.3427,
+      "step": 7210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.55379217953297e-06,
+      "loss": 1.1676,
+      "step": 7211
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.544049920636581e-06,
+      "loss": 0.7884,
+      "step": 7212
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.534313435531299e-06,
+      "loss": 1.2128,
+      "step": 7213
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.524582725541229e-06,
+      "loss": 0.9908,
+      "step": 7214
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.514857791989743e-06,
+      "loss": 1.3844,
+      "step": 7215
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.5051386361993385e-06,
+      "loss": 0.59,
+      "step": 7216
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.4954252594917785e-06,
+      "loss": 1.2151,
+      "step": 7217
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.485717663188036e-06,
+      "loss": 0.7958,
+      "step": 7218
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.4760158486083145e-06,
+      "loss": 1.3243,
+      "step": 7219
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.466319817071987e-06,
+      "loss": 0.8797,
+      "step": 7220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.456629569897666e-06,
+      "loss": 0.9815,
+      "step": 7221
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.4469451084031804e-06,
+      "loss": 1.2626,
+      "step": 7222
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.437266433905571e-06,
+      "loss": 1.4805,
+      "step": 7223
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.4275935477210785e-06,
+      "loss": 0.6053,
+      "step": 7224
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.417926451165169e-06,
+      "loss": 0.6427,
+      "step": 7225
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.408265145552523e-06,
+      "loss": 0.8472,
+      "step": 7226
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.398609632197023e-06,
+      "loss": 1.6034,
+      "step": 7227
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.388959912411775e-06,
+      "loss": 1.1514,
+      "step": 7228
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.379315987509083e-06,
+      "loss": 0.6164,
+      "step": 7229
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.369677858800478e-06,
+      "loss": 1.0268,
+      "step": 7230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.360045527596693e-06,
+      "loss": 0.5929,
+      "step": 7231
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.350418995207681e-06,
+      "loss": 0.9327,
+      "step": 7232
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.340798262942588e-06,
+      "loss": 0.9634,
+      "step": 7233
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.331183332109798e-06,
+      "loss": 1.2068,
+      "step": 7234
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.321574204016884e-06,
+      "loss": 1.3881,
+      "step": 7235
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.311970879970642e-06,
+      "loss": 0.9679,
+      "step": 7236
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.302373361277076e-06,
+      "loss": 0.6246,
+      "step": 7237
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.292781649241392e-06,
+      "loss": 0.8478,
+      "step": 7238
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.28319574516802e-06,
+      "loss": 1.3991,
+      "step": 7239
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.273615650360594e-06,
+      "loss": 1.5677,
+      "step": 7240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.264041366121954e-06,
+      "loss": 1.2067,
+      "step": 7241
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.254472893754155e-06,
+      "loss": 0.6389,
+      "step": 7242
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.244910234558461e-06,
+      "loss": 0.6741,
+      "step": 7243
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.235353389835353e-06,
+      "loss": 0.6106,
+      "step": 7244
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.2258023608844816e-06,
+      "loss": 1.2114,
+      "step": 7245
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.216257149004774e-06,
+      "loss": 1.1133,
+      "step": 7246
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.206717755494313e-06,
+      "loss": 1.3552,
+      "step": 7247
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.197184181650418e-06,
+      "loss": 1.3032,
+      "step": 7248
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.187656428769579e-06,
+      "loss": 1.0571,
+      "step": 7249
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.178134498147554e-06,
+      "loss": 1.4037,
+      "step": 7250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.1686183910792685e-06,
+      "loss": 1.0888,
+      "step": 7251
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.159108108858853e-06,
+      "loss": 0.6812,
+      "step": 7252
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.149603652779646e-06,
+      "loss": 0.9287,
+      "step": 7253
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.140105024134236e-06,
+      "loss": 1.0962,
+      "step": 7254
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.1306122242143835e-06,
+      "loss": 0.5983,
+      "step": 7255
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.121125254311023e-06,
+      "loss": 1.2474,
+      "step": 7256
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.111644115714377e-06,
+      "loss": 0.8767,
+      "step": 7257
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.102168809713827e-06,
+      "loss": 0.9877,
+      "step": 7258
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.092699337597941e-06,
+      "loss": 1.2246,
+      "step": 7259
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.083235700654517e-06,
+      "loss": 1.3173,
+      "step": 7260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.073777900170591e-06,
+      "loss": 0.9891,
+      "step": 7261
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.064325937432375e-06,
+      "loss": 1.1091,
+      "step": 7262
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.054879813725257e-06,
+      "loss": 1.2362,
+      "step": 7263
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.0454395303338635e-06,
+      "loss": 0.7746,
+      "step": 7264
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.036005088542064e-06,
+      "loss": 0.7129,
+      "step": 7265
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.026576489632858e-06,
+      "loss": 1.0139,
+      "step": 7266
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.017153734888493e-06,
+      "loss": 1.1919,
+      "step": 7267
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.007736825590406e-06,
+      "loss": 0.6761,
+      "step": 7268
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.998325763019287e-06,
+      "loss": 1.1035,
+      "step": 7269
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.988920548454952e-06,
+      "loss": 0.5551,
+      "step": 7270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.9795211831764794e-06,
+      "loss": 1.1401,
+      "step": 7271
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.970127668462129e-06,
+      "loss": 1.2795,
+      "step": 7272
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.960740005589373e-06,
+      "loss": 1.0049,
+      "step": 7273
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.951358195834889e-06,
+      "loss": 0.7238,
+      "step": 7274
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.941982240474548e-06,
+      "loss": 0.9093,
+      "step": 7275
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.932612140783434e-06,
+      "loss": 1.5621,
+      "step": 7276
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.923247898035834e-06,
+      "loss": 0.7338,
+      "step": 7277
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.913889513505234e-06,
+      "loss": 0.9694,
+      "step": 7278
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.904536988464327e-06,
+      "loss": 0.6282,
+      "step": 7279
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.895190324185014e-06,
+      "loss": 1.1611,
+      "step": 7280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.885849521938387e-06,
+      "loss": 1.0344,
+      "step": 7281
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.876514582994753e-06,
+      "loss": 1.2077,
+      "step": 7282
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.867185508623608e-06,
+      "loss": 1.2864,
+      "step": 7283
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.857862300093665e-06,
+      "loss": 1.266,
+      "step": 7284
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.848544958672826e-06,
+      "loss": 1.4468,
+      "step": 7285
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8392334856282044e-06,
+      "loss": 1.3308,
+      "step": 7286
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.829927882226117e-06,
+      "loss": 1.4054,
+      "step": 7287
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.820628149732073e-06,
+      "loss": 0.9289,
+      "step": 7288
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.811334289410792e-06,
+      "loss": 0.5565,
+      "step": 7289
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.802046302526189e-06,
+      "loss": 1.1003,
+      "step": 7290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.7927641903413775e-06,
+      "loss": 0.8481,
+      "step": 7291
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.783487954118683e-06,
+      "loss": 1.1388,
+      "step": 7292
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.7742175951196306e-06,
+      "loss": 0.8247,
+      "step": 7293
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.764953114604944e-06,
+      "loss": 1.3329,
+      "step": 7294
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.755694513834515e-06,
+      "loss": 1.0982,
+      "step": 7295
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.7464417940674985e-06,
+      "loss": 1.094,
+      "step": 7296
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.737194956562204e-06,
+      "loss": 1.2552,
+      "step": 7297
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.727954002576159e-06,
+      "loss": 0.8918,
+      "step": 7298
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.7187189333660806e-06,
+      "loss": 0.9897,
+      "step": 7299
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.7094897501878916e-06,
+      "loss": 1.3823,
+      "step": 7300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.700266454296727e-06,
+      "loss": 1.3859,
+      "step": 7301
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.691049046946874e-06,
+      "loss": 0.7908,
+      "step": 7302
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.681837529391888e-06,
+      "loss": 0.6586,
+      "step": 7303
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.672631902884469e-06,
+      "loss": 0.8628,
+      "step": 7304
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.663432168676553e-06,
+      "loss": 1.298,
+      "step": 7305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.654238328019224e-06,
+      "loss": 1.0245,
+      "step": 7306
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.645050382162826e-06,
+      "loss": 1.2471,
+      "step": 7307
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.635868332356876e-06,
+      "loss": 1.4263,
+      "step": 7308
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.626692179850058e-06,
+      "loss": 0.7231,
+      "step": 7309
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.617521925890285e-06,
+      "loss": 0.9723,
+      "step": 7310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.608357571724694e-06,
+      "loss": 1.1187,
+      "step": 7311
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.599199118599575e-06,
+      "loss": 0.5731,
+      "step": 7312
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.590046567760422e-06,
+      "loss": 1.2261,
+      "step": 7313
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.580899920451927e-06,
+      "loss": 1.679,
+      "step": 7314
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.5717591779180185e-06,
+      "loss": 1.0259,
+      "step": 7315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.562624341401769e-06,
+      "loss": 1.5714,
+      "step": 7316
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.553495412145466e-06,
+      "loss": 1.3758,
+      "step": 7317
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.544372391390596e-06,
+      "loss": 0.7693,
+      "step": 7318
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.535255280377878e-06,
+      "loss": 1.1468,
+      "step": 7319
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.526144080347146e-06,
+      "loss": 1.3513,
+      "step": 7320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.517038792537489e-06,
+      "loss": 1.1124,
+      "step": 7321
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.507939418187203e-06,
+      "loss": 1.3195,
+      "step": 7322
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.498845958533734e-06,
+      "loss": 0.3894,
+      "step": 7323
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.489758414813746e-06,
+      "loss": 1.1295,
+      "step": 7324
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.480676788263096e-06,
+      "loss": 1.1434,
+      "step": 7325
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.471601080116862e-06,
+      "loss": 1.1523,
+      "step": 7326
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.46253129160927e-06,
+      "loss": 1.3488,
+      "step": 7327
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.453467423973775e-06,
+      "loss": 1.1196,
+      "step": 7328
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.444409478443009e-06,
+      "loss": 1.3831,
+      "step": 7329
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.435357456248808e-06,
+      "loss": 0.7163,
+      "step": 7330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.4263113586222115e-06,
+      "loss": 0.9114,
+      "step": 7331
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.4172711867934274e-06,
+      "loss": 1.2025,
+      "step": 7332
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.40823694199188e-06,
+      "loss": 1.4329,
+      "step": 7333
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.399208625446179e-06,
+      "loss": 1.3255,
+      "step": 7334
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.390186238384127e-06,
+      "loss": 1.5949,
+      "step": 7335
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.381169782032731e-06,
+      "loss": 0.6794,
+      "step": 7336
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.372159257618171e-06,
+      "loss": 1.342,
+      "step": 7337
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.3631546663658395e-06,
+      "loss": 1.1544,
+      "step": 7338
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.354156009500312e-06,
+      "loss": 0.4594,
+      "step": 7339
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.34516328824536e-06,
+      "loss": 0.8366,
+      "step": 7340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.336176503823943e-06,
+      "loss": 1.2499,
+      "step": 7341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.3271956574582296e-06,
+      "loss": 0.5708,
+      "step": 7342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.318220750369558e-06,
+      "loss": 0.8374,
+      "step": 7343
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.309251783778469e-06,
+      "loss": 0.8819,
+      "step": 7344
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.300288758904705e-06,
+      "loss": 0.6985,
+      "step": 7345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.291331676967182e-06,
+      "loss": 1.2262,
+      "step": 7346
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.282380539184024e-06,
+      "loss": 1.3056,
+      "step": 7347
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.273435346772533e-06,
+      "loss": 1.5757,
+      "step": 7348
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.264496100949219e-06,
+      "loss": 1.4244,
+      "step": 7349
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.2555628029297654e-06,
+      "loss": 1.6923,
+      "step": 7350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.2466354539290615e-06,
+      "loss": 1.3664,
+      "step": 7351
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.237714055161164e-06,
+      "loss": 0.8089,
+      "step": 7352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.228798607839353e-06,
+      "loss": 1.385,
+      "step": 7353
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.2198891131760864e-06,
+      "loss": 1.2928,
+      "step": 7354
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.21098557238301e-06,
+      "loss": 0.9022,
+      "step": 7355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.2020879866709345e-06,
+      "loss": 1.2868,
+      "step": 7356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.193196357249914e-06,
+      "loss": 1.3345,
+      "step": 7357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.1843106853291646e-06,
+      "loss": 1.4835,
+      "step": 7358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.175430972117075e-06,
+      "loss": 0.5616,
+      "step": 7359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.166557218821234e-06,
+      "loss": 1.0184,
+      "step": 7360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.157689426648449e-06,
+      "loss": 0.9509,
+      "step": 7361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.148827596804702e-06,
+      "loss": 1.2923,
+      "step": 7362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.139971730495115e-06,
+      "loss": 0.587,
+      "step": 7363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.13112182892408e-06,
+      "loss": 1.2998,
+      "step": 7364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.122277893295131e-06,
+      "loss": 0.925,
+      "step": 7365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.113439924810977e-06,
+      "loss": 0.9765,
+      "step": 7366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.104607924673544e-06,
+      "loss": 1.0055,
+      "step": 7367
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.095781894083952e-06,
+      "loss": 1.3897,
+      "step": 7368
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.086961834242499e-06,
+      "loss": 0.7054,
+      "step": 7369
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.078147746348644e-06,
+      "loss": 0.7165,
+      "step": 7370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.069339631601062e-06,
+      "loss": 1.2181,
+      "step": 7371
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.060537491197643e-06,
+      "loss": 0.765,
+      "step": 7372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.051741326335397e-06,
+      "loss": 1.3015,
+      "step": 7373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.042951138210567e-06,
+      "loss": 0.9703,
+      "step": 7374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0341669280185696e-06,
+      "loss": 0.6,
+      "step": 7375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.025388696954037e-06,
+      "loss": 1.5545,
+      "step": 7376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.016616446210743e-06,
+      "loss": 0.9283,
+      "step": 7377
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0078501769816665e-06,
+      "loss": 0.8287,
+      "step": 7378
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.999089890458975e-06,
+      "loss": 1.6828,
+      "step": 7379
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.990335587834034e-06,
+      "loss": 1.0825,
+      "step": 7380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.981587270297373e-06,
+      "loss": 1.2398,
+      "step": 7381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.972844939038724e-06,
+      "loss": 1.2547,
+      "step": 7382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9641085952469975e-06,
+      "loss": 1.3602,
+      "step": 7383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9553782401102935e-06,
+      "loss": 1.5629,
+      "step": 7384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.946653874815889e-06,
+      "loss": 0.8418,
+      "step": 7385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.937935500550257e-06,
+      "loss": 1.3906,
+      "step": 7386
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.929223118499056e-06,
+      "loss": 1.424,
+      "step": 7387
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.920516729847115e-06,
+      "loss": 1.122,
+      "step": 7388
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.911816335778469e-06,
+      "loss": 0.7458,
+      "step": 7389
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.903121937476319e-06,
+      "loss": 0.6032,
+      "step": 7390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8944335361230615e-06,
+      "loss": 0.66,
+      "step": 7391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.885751132900275e-06,
+      "loss": 1.2296,
+      "step": 7392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.877074728988724e-06,
+      "loss": 0.8741,
+      "step": 7393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.86840432556835e-06,
+      "loss": 1.389,
+      "step": 7394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8597399238182805e-06,
+      "loss": 1.4642,
+      "step": 7395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.85108152491684e-06,
+      "loss": 1.4896,
+      "step": 7396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.842429130041521e-06,
+      "loss": 1.2077,
+      "step": 7397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8337827403690036e-06,
+      "loss": 1.0072,
+      "step": 7398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.825142357075153e-06,
+      "loss": 1.4574,
+      "step": 7399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.816507981335012e-06,
+      "loss": 0.8461,
+      "step": 7400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.807879614322836e-06,
+      "loss": 1.8114,
+      "step": 7401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.799257257211993e-06,
+      "loss": 0.9584,
+      "step": 7402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7906409111751155e-06,
+      "loss": 0.5923,
+      "step": 7403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.782030577383974e-06,
+      "loss": 1.1913,
+      "step": 7404
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.773426257009529e-06,
+      "loss": 1.3187,
+      "step": 7405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.764827951221918e-06,
+      "loss": 1.6536,
+      "step": 7406
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.756235661190474e-06,
+      "loss": 0.878,
+      "step": 7407
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.747649388083704e-06,
+      "loss": 1.2332,
+      "step": 7408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.739069133069291e-06,
+      "loss": 1.1697,
+      "step": 7409
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7304948973141095e-06,
+      "loss": 1.1485,
+      "step": 7410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.721926681984213e-06,
+      "loss": 1.0371,
+      "step": 7411
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7133644882448376e-06,
+      "loss": 1.6197,
+      "step": 7412
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.704808317260379e-06,
+      "loss": 1.1915,
+      "step": 7413
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.696258170194451e-06,
+      "loss": 1.4207,
+      "step": 7414
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.68771404820983e-06,
+      "loss": 0.8057,
+      "step": 7415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.679175952468479e-06,
+      "loss": 0.6803,
+      "step": 7416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.670643884131499e-06,
+      "loss": 0.6334,
+      "step": 7417
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.662117844359244e-06,
+      "loss": 1.4657,
+      "step": 7418
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.653597834311214e-06,
+      "loss": 0.9164,
+      "step": 7419
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.645083855146055e-06,
+      "loss": 0.7375,
+      "step": 7420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.636575908021641e-06,
+      "loss": 0.8316,
+      "step": 7421
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.628073994095012e-06,
+      "loss": 0.4999,
+      "step": 7422
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.619578114522394e-06,
+      "loss": 0.7047,
+      "step": 7423
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.611088270459169e-06,
+      "loss": 1.3657,
+      "step": 7424
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.602604463059896e-06,
+      "loss": 1.1042,
+      "step": 7425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.594126693478374e-06,
+      "loss": 1.3106,
+      "step": 7426
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.585654962867498e-06,
+      "loss": 0.7866,
+      "step": 7427
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.577189272379385e-06,
+      "loss": 1.2759,
+      "step": 7428
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.568729623165342e-06,
+      "loss": 0.6817,
+      "step": 7429
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.560276016375831e-06,
+      "loss": 0.9271,
+      "step": 7430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5518284531604955e-06,
+      "loss": 0.7205,
+      "step": 7431
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5433869346681485e-06,
+      "loss": 0.4186,
+      "step": 7432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.534951462046828e-06,
+      "loss": 1.4416,
+      "step": 7433
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.526522036443682e-06,
+      "loss": 1.2577,
+      "step": 7434
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.518098659005078e-06,
+      "loss": 0.9941,
+      "step": 7435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5096813308765484e-06,
+      "loss": 1.4162,
+      "step": 7436
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.501270053202829e-06,
+      "loss": 1.1665,
+      "step": 7437
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.492864827127786e-06,
+      "loss": 1.3156,
+      "step": 7438
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.48446565379449e-06,
+      "loss": 0.5981,
+      "step": 7439
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4760725343452e-06,
+      "loss": 1.0421,
+      "step": 7440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4676854699213185e-06,
+      "loss": 1.3187,
+      "step": 7441
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.459304461663456e-06,
+      "loss": 1.0221,
+      "step": 7442
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4509295107113835e-06,
+      "loss": 1.1197,
+      "step": 7443
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.442560618204051e-06,
+      "loss": 1.1627,
+      "step": 7444
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.434197785279582e-06,
+      "loss": 1.5611,
+      "step": 7445
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.425841013075284e-06,
+      "loss": 1.3485,
+      "step": 7446
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.417490302727629e-06,
+      "loss": 0.8775,
+      "step": 7447
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4091456553722775e-06,
+      "loss": 1.2182,
+      "step": 7448
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.400807072144054e-06,
+      "loss": 1.1179,
+      "step": 7449
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.392474554176962e-06,
+      "loss": 1.1817,
+      "step": 7450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.384148102604186e-06,
+      "loss": 0.8059,
+      "step": 7451
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.37582771855808e-06,
+      "loss": 0.9642,
+      "step": 7452
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.367513403170166e-06,
+      "loss": 1.6047,
+      "step": 7453
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.359205157571156e-06,
+      "loss": 1.3482,
+      "step": 7454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.350902982890932e-06,
+      "loss": 0.8456,
+      "step": 7455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.342606880258533e-06,
+      "loss": 1.5586,
+      "step": 7456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.334316850802201e-06,
+      "loss": 1.3693,
+      "step": 7457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.32603289564933e-06,
+      "loss": 0.6926,
+      "step": 7458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.31775501592649e-06,
+      "loss": 0.7404,
+      "step": 7459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3094832127594415e-06,
+      "loss": 0.9384,
+      "step": 7460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.301217487273108e-06,
+      "loss": 1.3875,
+      "step": 7461
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.292957840591583e-06,
+      "loss": 1.2309,
+      "step": 7462
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.284704273838115e-06,
+      "loss": 1.0207,
+      "step": 7463
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.276456788135176e-06,
+      "loss": 1.0665,
+      "step": 7464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.268215384604369e-06,
+      "loss": 1.2023,
+      "step": 7465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2599800643664935e-06,
+      "loss": 1.5109,
+      "step": 7466
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2517508285414855e-06,
+      "loss": 0.967,
+      "step": 7467
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2435276782485e-06,
+      "loss": 1.6517,
+      "step": 7468
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.235310614605848e-06,
+      "loss": 1.6572,
+      "step": 7469
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.227099638730981e-06,
+      "loss": 1.0043,
+      "step": 7470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.218894751740577e-06,
+      "loss": 1.0399,
+      "step": 7471
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.210695954750444e-06,
+      "loss": 1.2728,
+      "step": 7472
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.202503248875595e-06,
+      "loss": 0.7124,
+      "step": 7473
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1943166352301565e-06,
+      "loss": 1.1068,
+      "step": 7474
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.186136114927503e-06,
+      "loss": 1.1505,
+      "step": 7475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.177961689080135e-06,
+      "loss": 1.1545,
+      "step": 7476
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1697933587997235e-06,
+      "loss": 1.2519,
+      "step": 7477
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.161631125197114e-06,
+      "loss": 0.6013,
+      "step": 7478
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.153474989382351e-06,
+      "loss": 1.0879,
+      "step": 7479
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.14532495246462e-06,
+      "loss": 1.3099,
+      "step": 7480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.137181015552267e-06,
+      "loss": 1.5687,
+      "step": 7481
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.12904317975283e-06,
+      "loss": 1.4011,
+      "step": 7482
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.120911446173038e-06,
+      "loss": 1.7395,
+      "step": 7483
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.112785815918741e-06,
+      "loss": 1.2509,
+      "step": 7484
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.104666290094984e-06,
+      "loss": 1.2107,
+      "step": 7485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.096552869805982e-06,
+      "loss": 0.7474,
+      "step": 7486
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.088445556155136e-06,
+      "loss": 1.1764,
+      "step": 7487
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.080344350244981e-06,
+      "loss": 1.0489,
+      "step": 7488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.07224925317723e-06,
+      "loss": 1.3549,
+      "step": 7489
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0641602660528085e-06,
+      "loss": 0.8287,
+      "step": 7490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.056077389971747e-06,
+      "loss": 1.141,
+      "step": 7491
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.048000626033289e-06,
+      "loss": 0.6942,
+      "step": 7492
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.039929975335816e-06,
+      "loss": 1.3626,
+      "step": 7493
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.031865438976924e-06,
+      "loss": 1.1637,
+      "step": 7494
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.023807018053328e-06,
+      "loss": 0.7173,
+      "step": 7495
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.015754713660936e-06,
+      "loss": 0.9429,
+      "step": 7496
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.007708526894822e-06,
+      "loss": 1.2052,
+      "step": 7497
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.999668458849221e-06,
+      "loss": 1.3218,
+      "step": 7498
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.991634510617544e-06,
+      "loss": 1.2039,
+      "step": 7499
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.983606683292369e-06,
+      "loss": 0.8758,
+      "step": 7500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.975584977965437e-06,
+      "loss": 1.238,
+      "step": 7501
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.967569395727661e-06,
+      "loss": 0.724,
+      "step": 7502
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.959559937669112e-06,
+      "loss": 0.9282,
+      "step": 7503
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.951556604879048e-06,
+      "loss": 1.5293,
+      "step": 7504
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.943559398445869e-06,
+      "loss": 1.3129,
+      "step": 7505
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.935568319457157e-06,
+      "loss": 0.5527,
+      "step": 7506
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9275833689996605e-06,
+      "loss": 0.8433,
+      "step": 7507
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.91960454815929e-06,
+      "loss": 0.585,
+      "step": 7508
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9116318580211265e-06,
+      "loss": 1.2632,
+      "step": 7509
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.903665299669413e-06,
+      "loss": 1.1219,
+      "step": 7510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.895704874187562e-06,
+      "loss": 1.4059,
+      "step": 7511
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.887750582658146e-06,
+      "loss": 1.4682,
+      "step": 7512
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.87980242616291e-06,
+      "loss": 0.9408,
+      "step": 7513
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.871860405782764e-06,
+      "loss": 0.6226,
+      "step": 7514
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.863924522597785e-06,
+      "loss": 1.1374,
+      "step": 7515
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.855994777687201e-06,
+      "loss": 1.1932,
+      "step": 7516
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.848071172129431e-06,
+      "loss": 0.9935,
+      "step": 7517
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.840153707002032e-06,
+      "loss": 0.9951,
+      "step": 7518
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.832242383381758e-06,
+      "loss": 1.2476,
+      "step": 7519
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.824337202344475e-06,
+      "loss": 0.8142,
+      "step": 7520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8164381649652755e-06,
+      "loss": 0.8004,
+      "step": 7521
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8085452723183824e-06,
+      "loss": 1.2779,
+      "step": 7522
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.800658525477192e-06,
+      "loss": 1.3374,
+      "step": 7523
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.792777925514236e-06,
+      "loss": 1.0001,
+      "step": 7524
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.784903473501262e-06,
+      "loss": 0.4278,
+      "step": 7525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.777035170509159e-06,
+      "loss": 1.1664,
+      "step": 7526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.769173017607953e-06,
+      "loss": 1.0904,
+      "step": 7527
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.761317015866856e-06,
+      "loss": 0.8463,
+      "step": 7528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.75346716635427e-06,
+      "loss": 0.9989,
+      "step": 7529
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.745623470137728e-06,
+      "loss": 1.3037,
+      "step": 7530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.737785928283911e-06,
+      "loss": 1.5802,
+      "step": 7531
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.729954541858689e-06,
+      "loss": 1.2774,
+      "step": 7532
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.72212931192712e-06,
+      "loss": 0.3999,
+      "step": 7533
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.71431023955336e-06,
+      "loss": 1.2081,
+      "step": 7534
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.706497325800774e-06,
+      "loss": 0.5532,
+      "step": 7535
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.698690571731884e-06,
+      "loss": 0.9821,
+      "step": 7536
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.690889978408375e-06,
+      "loss": 1.2603,
+      "step": 7537
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.68309554689107e-06,
+      "loss": 1.2832,
+      "step": 7538
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.675307278239971e-06,
+      "loss": 1.0771,
+      "step": 7539
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.66752517351427e-06,
+      "loss": 0.7451,
+      "step": 7540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.659749233772259e-06,
+      "loss": 1.2639,
+      "step": 7541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.651979460071448e-06,
+      "loss": 1.2953,
+      "step": 7542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6442158534684644e-06,
+      "loss": 1.0178,
+      "step": 7543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.636458415019157e-06,
+      "loss": 1.1499,
+      "step": 7544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.62870714577846e-06,
+      "loss": 1.0124,
+      "step": 7545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.620962046800526e-06,
+      "loss": 0.8415,
+      "step": 7546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.613223119138638e-06,
+      "loss": 0.6688,
+      "step": 7547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.605490363845261e-06,
+      "loss": 1.0069,
+      "step": 7548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5977637819719975e-06,
+      "loss": 1.1899,
+      "step": 7549
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.590043374569636e-06,
+      "loss": 1.2686,
+      "step": 7550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.582329142688102e-06,
+      "loss": 1.3898,
+      "step": 7551
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.574621087376501e-06,
+      "loss": 1.1585,
+      "step": 7552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5669192096830795e-06,
+      "loss": 1.363,
+      "step": 7553
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.559223510655258e-06,
+      "loss": 1.1538,
+      "step": 7554
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5515339913396125e-06,
+      "loss": 0.9614,
+      "step": 7555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.543850652781872e-06,
+      "loss": 0.8604,
+      "step": 7556
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.536173496026941e-06,
+      "loss": 0.6143,
+      "step": 7557
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.528502522118871e-06,
+      "loss": 0.8277,
+      "step": 7558
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.520837732100869e-06,
+      "loss": 0.9344,
+      "step": 7559
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.51317912701531e-06,
+      "loss": 1.1031,
+      "step": 7560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.505526707903729e-06,
+      "loss": 1.2351,
+      "step": 7561
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.497880475806809e-06,
+      "loss": 1.2573,
+      "step": 7562
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.490240431764398e-06,
+      "loss": 1.0369,
+      "step": 7563
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4826065768155125e-06,
+      "loss": 0.9636,
+      "step": 7564
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.474978911998312e-06,
+      "loss": 1.0938,
+      "step": 7565
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4673574383501115e-06,
+      "loss": 0.5315,
+      "step": 7566
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.459742156907409e-06,
+      "loss": 1.3085,
+      "step": 7567
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.452133068705827e-06,
+      "loss": 1.5617,
+      "step": 7568
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.444530174780182e-06,
+      "loss": 0.7227,
+      "step": 7569
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.436933476164401e-06,
+      "loss": 1.2953,
+      "step": 7570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.429342973891615e-06,
+      "loss": 1.3954,
+      "step": 7571
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.421758668994092e-06,
+      "loss": 0.7667,
+      "step": 7572
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.414180562503268e-06,
+      "loss": 1.4903,
+      "step": 7573
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.406608655449696e-06,
+      "loss": 1.4366,
+      "step": 7574
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.399042948863141e-06,
+      "loss": 0.8244,
+      "step": 7575
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.391483443772504e-06,
+      "loss": 0.7998,
+      "step": 7576
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.383930141205817e-06,
+      "loss": 1.2544,
+      "step": 7577
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3763830421903015e-06,
+      "loss": 1.1616,
+      "step": 7578
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.368842147752333e-06,
+      "loss": 0.6012,
+      "step": 7579
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.361307458917436e-06,
+      "loss": 1.2612,
+      "step": 7580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.35377897671026e-06,
+      "loss": 0.6812,
+      "step": 7581
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.346256702154667e-06,
+      "loss": 1.11,
+      "step": 7582
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.338740636273653e-06,
+      "loss": 0.7682,
+      "step": 7583
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.331230780089346e-06,
+      "loss": 1.1585,
+      "step": 7584
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.32372713462304e-06,
+      "loss": 1.3267,
+      "step": 7585
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.316229700895214e-06,
+      "loss": 1.3272,
+      "step": 7586
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.308738479925484e-06,
+      "loss": 1.1582,
+      "step": 7587
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.301253472732603e-06,
+      "loss": 0.4594,
+      "step": 7588
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.293774680334478e-06,
+      "loss": 0.6289,
+      "step": 7589
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2863021037482296e-06,
+      "loss": 1.1471,
+      "step": 7590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.278835743990056e-06,
+      "loss": 1.005,
+      "step": 7591
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.27137560207535e-06,
+      "loss": 1.3236,
+      "step": 7592
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.263921679018651e-06,
+      "loss": 1.3237,
+      "step": 7593
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.256473975833674e-06,
+      "loss": 1.0919,
+      "step": 7594
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.249032493533245e-06,
+      "loss": 1.3591,
+      "step": 7595
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.241597233129369e-06,
+      "loss": 1.1074,
+      "step": 7596
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.234168195633226e-06,
+      "loss": 0.5065,
+      "step": 7597
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.226745382055103e-06,
+      "loss": 1.197,
+      "step": 7598
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.21932879340447e-06,
+      "loss": 1.7964,
+      "step": 7599
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2119184306899376e-06,
+      "loss": 0.9417,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.204514294919309e-06,
+      "loss": 1.6535,
+      "step": 7601
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.197116387099476e-06,
+      "loss": 1.2085,
+      "step": 7602
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.197116387099476e-06,
+      "loss": 1.2211,
+      "step": 7603
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.189724708236526e-06,
+      "loss": 1.1806,
+      "step": 7604
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1823392593356825e-06,
+      "loss": 1.4066,
+      "step": 7605
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.174960041401355e-06,
+      "loss": 0.7187,
+      "step": 7606
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.167587055437044e-06,
+      "loss": 1.229,
+      "step": 7607
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.160220302445461e-06,
+      "loss": 1.0208,
+      "step": 7608
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.152859783428437e-06,
+      "loss": 1.3285,
+      "step": 7609
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1455054993869705e-06,
+      "loss": 1.5799,
+      "step": 7610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1381574513212e-06,
+      "loss": 1.0899,
+      "step": 7611
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.130815640230429e-06,
+      "loss": 0.6477,
+      "step": 7612
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.123480067113094e-06,
+      "loss": 1.22,
+      "step": 7613
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.116150732966811e-06,
+      "loss": 1.8855,
+      "step": 7614
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.108827638788326e-06,
+      "loss": 1.3682,
+      "step": 7615
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.101510785573537e-06,
+      "loss": 1.509,
+      "step": 7616
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.094200174317503e-06,
+      "loss": 1.2587,
+      "step": 7617
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.086895806014429e-06,
+      "loss": 1.1613,
+      "step": 7618
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0795976816576705e-06,
+      "loss": 0.5947,
+      "step": 7619
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.07230580223974e-06,
+      "loss": 1.1138,
+      "step": 7620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.065020168752287e-06,
+      "loss": 1.3396,
+      "step": 7621
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.057740782186125e-06,
+      "loss": 1.4307,
+      "step": 7622
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.050467643531214e-06,
+      "loss": 1.1078,
+      "step": 7623
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.043200753776666e-06,
+      "loss": 1.2048,
+      "step": 7624
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.03594011391073e-06,
+      "loss": 0.4927,
+      "step": 7625
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.028685724920828e-06,
+      "loss": 1.0037,
+      "step": 7626
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.021437587793514e-06,
+      "loss": 1.3295,
+      "step": 7627
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.014195703514495e-06,
+      "loss": 0.6919,
+      "step": 7628
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.006960073068633e-06,
+      "loss": 1.2456,
+      "step": 7629
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.99973069743994e-06,
+      "loss": 1.0212,
+      "step": 7630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.992507577611576e-06,
+      "loss": 1.3377,
+      "step": 7631
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.985290714565825e-06,
+      "loss": 0.5785,
+      "step": 7632
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.978080109284171e-06,
+      "loss": 1.4249,
+      "step": 7633
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.970875762747211e-06,
+      "loss": 1.3876,
+      "step": 7634
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.963677675934702e-06,
+      "loss": 1.1257,
+      "step": 7635
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.956485849825525e-06,
+      "loss": 1.2444,
+      "step": 7636
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.949300285397761e-06,
+      "loss": 0.6352,
+      "step": 7637
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9421209836286055e-06,
+      "loss": 1.504,
+      "step": 7638
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9349479454943775e-06,
+      "loss": 0.8924,
+      "step": 7639
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.927781171970607e-06,
+      "loss": 1.333,
+      "step": 7640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.920620664031921e-06,
+      "loss": 0.9946,
+      "step": 7641
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.913466422652129e-06,
+      "loss": 0.9247,
+      "step": 7642
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.906318448804142e-06,
+      "loss": 1.2773,
+      "step": 7643
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.89917674346007e-06,
+      "loss": 0.8174,
+      "step": 7644
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.892041307591155e-06,
+      "loss": 1.1519,
+      "step": 7645
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8849121421677516e-06,
+      "loss": 1.6656,
+      "step": 7646
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.877789248159403e-06,
+      "loss": 0.9962,
+      "step": 7647
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.870672626534793e-06,
+      "loss": 1.3555,
+      "step": 7648
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.863562278261756e-06,
+      "loss": 1.065,
+      "step": 7649
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.856458204307234e-06,
+      "loss": 0.8489,
+      "step": 7650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8493604056373535e-06,
+      "loss": 1.4203,
+      "step": 7651
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.842268883217398e-06,
+      "loss": 0.5276,
+      "step": 7652
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.835183638011758e-06,
+      "loss": 0.745,
+      "step": 7653
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.82810467098399e-06,
+      "loss": 0.7648,
+      "step": 7654
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8210319830968015e-06,
+      "loss": 0.65,
+      "step": 7655
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8139655753120562e-06,
+      "loss": 0.9024,
+      "step": 7656
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8069054485907307e-06,
+      "loss": 0.9668,
+      "step": 7657
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.799851603892973e-06,
+      "loss": 1.2662,
+      "step": 7658
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.792804042178066e-06,
+      "loss": 1.2763,
+      "step": 7659
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7857627644044437e-06,
+      "loss": 1.6932,
+      "step": 7660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7787277715296886e-06,
+      "loss": 1.1199,
+      "step": 7661
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7716990645105087e-06,
+      "loss": 1.3522,
+      "step": 7662
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7646766443028004e-06,
+      "loss": 0.9438,
+      "step": 7663
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7576605118615494e-06,
+      "loss": 1.2565,
+      "step": 7664
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.750650668140926e-06,
+      "loss": 0.8853,
+      "step": 7665
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7436471140942287e-06,
+      "loss": 1.6194,
+      "step": 7666
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7366498506739067e-06,
+      "loss": 1.1326,
+      "step": 7667
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.729658878831549e-06,
+      "loss": 0.9674,
+      "step": 7668
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7226741995179006e-06,
+      "loss": 0.5455,
+      "step": 7669
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7156958136828346e-06,
+      "loss": 1.2517,
+      "step": 7670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7087237222753755e-06,
+      "loss": 0.5918,
+      "step": 7671
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.701757926243693e-06,
+      "loss": 0.7887,
+      "step": 7672
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.694798426535101e-06,
+      "loss": 0.8636,
+      "step": 7673
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6878452240960538e-06,
+      "loss": 0.8385,
+      "step": 7674
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6808983198721557e-06,
+      "loss": 0.7849,
+      "step": 7675
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6739577148081406e-06,
+      "loss": 0.9018,
+      "step": 7676
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6670234098479083e-06,
+      "loss": 0.9583,
+      "step": 7677
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6600954059344773e-06,
+      "loss": 0.4151,
+      "step": 7678
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.653173704010021e-06,
+      "loss": 1.564,
+      "step": 7679
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.646258305015865e-06,
+      "loss": 1.2492,
+      "step": 7680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6393492098924675e-06,
+      "loss": 0.5778,
+      "step": 7681
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6324464195794104e-06,
+      "loss": 1.2164,
+      "step": 7682
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6255499350154544e-06,
+      "loss": 1.1627,
+      "step": 7683
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.618659757138487e-06,
+      "loss": 1.1618,
+      "step": 7684
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6117758868855377e-06,
+      "loss": 1.3065,
+      "step": 7685
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6048983251927683e-06,
+      "loss": 1.4264,
+      "step": 7686
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5980270729954967e-06,
+      "loss": 1.2987,
+      "step": 7687
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5911621312281873e-06,
+      "loss": 0.9295,
+      "step": 7688
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5843035008244097e-06,
+      "loss": 1.1036,
+      "step": 7689
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.577451182716929e-06,
+      "loss": 1.1004,
+      "step": 7690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.570605177837616e-06,
+      "loss": 0.9489,
+      "step": 7691
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.563765487117504e-06,
+      "loss": 0.8792,
+      "step": 7692
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.556932111486727e-06,
+      "loss": 1.2319,
+      "step": 7693
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.550105051874614e-06,
+      "loss": 1.2195,
+      "step": 7694
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.543284309209605e-06,
+      "loss": 0.8759,
+      "step": 7695
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.53646988441928e-06,
+      "loss": 0.5254,
+      "step": 7696
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5296617784303655e-06,
+      "loss": 0.9185,
+      "step": 7697
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5228599921687365e-06,
+      "loss": 1.4279,
+      "step": 7698
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.516064526559404e-06,
+      "loss": 0.5525,
+      "step": 7699
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.509275382526506e-06,
+      "loss": 1.4028,
+      "step": 7700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5024925609933202e-06,
+      "loss": 1.0446,
+      "step": 7701
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.495716062882309e-06,
+      "loss": 1.1914,
+      "step": 7702
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.488945889115014e-06,
+      "loss": 1.4198,
+      "step": 7703
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4821820406121473e-06,
+      "loss": 0.9829,
+      "step": 7704
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4754245182935685e-06,
+      "loss": 1.3435,
+      "step": 7705
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.46867332307827e-06,
+      "loss": 1.1908,
+      "step": 7706
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.461928455884361e-06,
+      "loss": 0.9403,
+      "step": 7707
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4551899176291093e-06,
+      "loss": 1.2304,
+      "step": 7708
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4484577092289415e-06,
+      "loss": 1.0829,
+      "step": 7709
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4417318315993876e-06,
+      "loss": 0.5104,
+      "step": 7710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.435012285655137e-06,
+      "loss": 1.5036,
+      "step": 7711
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4282990723100038e-06,
+      "loss": 1.0778,
+      "step": 7712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.42159219247698e-06,
+      "loss": 0.4502,
+      "step": 7713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4148916470681302e-06,
+      "loss": 1.0431,
+      "step": 7714
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4081974369947147e-06,
+      "loss": 1.1628,
+      "step": 7715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4015095631671103e-06,
+      "loss": 1.1994,
+      "step": 7716
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3948280264948285e-06,
+      "loss": 1.2941,
+      "step": 7717
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3881528278865205e-06,
+      "loss": 0.7628,
+      "step": 7718
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3814839682499933e-06,
+      "loss": 0.4982,
+      "step": 7719
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.37482144849216e-06,
+      "loss": 1.5743,
+      "step": 7720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3681652695191023e-06,
+      "loss": 1.0193,
+      "step": 7721
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.361515432236023e-06,
+      "loss": 1.1835,
+      "step": 7722
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.354871937547266e-06,
+      "loss": 0.8141,
+      "step": 7723
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.348234786356308e-06,
+      "loss": 0.8774,
+      "step": 7724
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3416039795657715e-06,
+      "loss": 1.2801,
+      "step": 7725
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.334979518077408e-06,
+      "loss": 0.9615,
+      "step": 7726
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3283614027921182e-06,
+      "loss": 1.0128,
+      "step": 7727
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3217496346099274e-06,
+      "loss": 1.202,
+      "step": 7728
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.315144214429999e-06,
+      "loss": 1.0067,
+      "step": 7729
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3085451431506377e-06,
+      "loss": 1.1848,
+      "step": 7730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.301952421669291e-06,
+      "loss": 1.323,
+      "step": 7731
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.29536605088252e-06,
+      "loss": 1.1991,
+      "step": 7732
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.288786031686053e-06,
+      "loss": 0.8529,
+      "step": 7733
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2822123649747296e-06,
+      "loss": 1.7365,
+      "step": 7734
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2756450516425343e-06,
+      "loss": 1.2963,
+      "step": 7735
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2690840925825974e-06,
+      "loss": 1.1073,
+      "step": 7736
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2625294886871605e-06,
+      "loss": 0.9079,
+      "step": 7737
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2559812408476385e-06,
+      "loss": 1.0252,
+      "step": 7738
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.249439349954525e-06,
+      "loss": 0.6522,
+      "step": 7739
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2429038168975133e-06,
+      "loss": 1.285,
+      "step": 7740
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2363746425653985e-06,
+      "loss": 1.6368,
+      "step": 7741
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.229851827846109e-06,
+      "loss": 1.2453,
+      "step": 7742
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2233353736267015e-06,
+      "loss": 0.7015,
+      "step": 7743
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.216825280793406e-06,
+      "loss": 1.3087,
+      "step": 7744
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2103215502315488e-06,
+      "loss": 1.3003,
+      "step": 7745
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2038241828255933e-06,
+      "loss": 0.7066,
+      "step": 7746
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1973331794591664e-06,
+      "loss": 1.3363,
+      "step": 7747
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.190848541015001e-06,
+      "loss": 1.4105,
+      "step": 7748
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1843702683749867e-06,
+      "loss": 1.3523,
+      "step": 7749
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1778983624201066e-06,
+      "loss": 1.1413,
+      "step": 7750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.17143282403054e-06,
+      "loss": 1.0749,
+      "step": 7751
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.164973654085557e-06,
+      "loss": 0.7431,
+      "step": 7752
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1585208534635537e-06,
+      "loss": 1.0582,
+      "step": 7753
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1520744230420894e-06,
+      "loss": 1.2619,
+      "step": 7754
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.145634363697858e-06,
+      "loss": 1.0642,
+      "step": 7755
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1392006763066694e-06,
+      "loss": 1.3993,
+      "step": 7756
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1327733617434575e-06,
+      "loss": 1.1313,
+      "step": 7757
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.126352420882306e-06,
+      "loss": 1.3663,
+      "step": 7758
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1199378545964495e-06,
+      "loss": 1.1966,
+      "step": 7759
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.113529663758219e-06,
+      "loss": 1.8169,
+      "step": 7760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1071278492391e-06,
+      "loss": 1.5915,
+      "step": 7761
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1007324119096958e-06,
+      "loss": 0.4925,
+      "step": 7762
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0943433526397838e-06,
+      "loss": 1.0112,
+      "step": 7763
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0879606722982123e-06,
+      "loss": 0.5098,
+      "step": 7764
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0815843717530047e-06,
+      "loss": 0.4843,
+      "step": 7765
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0752144518713053e-06,
+      "loss": 0.9848,
+      "step": 7766
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.068850913519389e-06,
+      "loss": 0.6255,
+      "step": 7767
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.062493757562668e-06,
+      "loss": 0.4853,
+      "step": 7768
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0561429848656676e-06,
+      "loss": 0.8176,
+      "step": 7769
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.049798596292086e-06,
+      "loss": 0.7641,
+      "step": 7770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.043460592704711e-06,
+      "loss": 1.1242,
+      "step": 7771
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0371289749654807e-06,
+      "loss": 1.5206,
+      "step": 7772
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0308037439354564e-06,
+      "loss": 1.4033,
+      "step": 7773
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.02448490047485e-06,
+      "loss": 1.3196,
+      "step": 7774
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0181724454429847e-06,
+      "loss": 1.1756,
+      "step": 7775
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.011866379698325e-06,
+      "loss": 0.9167,
+      "step": 7776
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0055667040984612e-06,
+      "loss": 0.7296,
+      "step": 7777
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.999273419500115e-06,
+      "loss": 1.3096,
+      "step": 7778
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9929865267591393e-06,
+      "loss": 1.0657,
+      "step": 7779
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9867060267305293e-06,
+      "loss": 1.5209,
+      "step": 7780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9804319202683896e-06,
+      "loss": 0.9816,
+      "step": 7781
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9741642082259715e-06,
+      "loss": 1.2582,
+      "step": 7782
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9679028914556482e-06,
+      "loss": 1.1255,
+      "step": 7783
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.961647970808934e-06,
+      "loss": 0.612,
+      "step": 7784
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9553994471364586e-06,
+      "loss": 1.3869,
+      "step": 7785
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9491573212879986e-06,
+      "loss": 1.2072,
+      "step": 7786
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9429215941124356e-06,
+      "loss": 1.3662,
+      "step": 7787
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9366922664578076e-06,
+      "loss": 1.005,
+      "step": 7788
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9304693391712713e-06,
+      "loss": 1.2567,
+      "step": 7789
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.92425281309911e-06,
+      "loss": 1.0498,
+      "step": 7790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9180426890867417e-06,
+      "loss": 0.8976,
+      "step": 7791
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9118389679787028e-06,
+      "loss": 0.5719,
+      "step": 7792
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.905641650618679e-06,
+      "loss": 0.5104,
+      "step": 7793
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8994507378494684e-06,
+      "loss": 1.6226,
+      "step": 7794
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.893266230513009e-06,
+      "loss": 1.444,
+      "step": 7795
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8870881294503436e-06,
+      "loss": 1.4042,
+      "step": 7796
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8809164355016847e-06,
+      "loss": 0.7385,
+      "step": 7797
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8747511495063382e-06,
+      "loss": 1.0193,
+      "step": 7798
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.868592272302761e-06,
+      "loss": 1.6456,
+      "step": 7799
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8624398047285117e-06,
+      "loss": 0.5855,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.856293747620309e-06,
+      "loss": 1.3048,
+      "step": 7801
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.85015410181399e-06,
+      "loss": 1.1221,
+      "step": 7802
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.844020868144498e-06,
+      "loss": 1.1696,
+      "step": 7803
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8378940474459214e-06,
+      "loss": 0.64,
+      "step": 7804
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8317736405514936e-06,
+      "loss": 1.4633,
+      "step": 7805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.825659648293555e-06,
+      "loss": 0.8355,
+      "step": 7806
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8195520715035674e-06,
+      "loss": 1.3055,
+      "step": 7807
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.813450911012122e-06,
+      "loss": 0.614,
+      "step": 7808
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8073561676489724e-06,
+      "loss": 1.3548,
+      "step": 7809
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8012678422429496e-06,
+      "loss": 0.6503,
+      "step": 7810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7951859356220354e-06,
+      "loss": 1.5692,
+      "step": 7811
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7891104486133525e-06,
+      "loss": 1.6889,
+      "step": 7812
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7830413820431333e-06,
+      "loss": 1.3466,
+      "step": 7813
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7769787367367296e-06,
+      "loss": 1.037,
+      "step": 7814
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.770922513518626e-06,
+      "loss": 0.4305,
+      "step": 7815
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7648727132124575e-06,
+      "loss": 1.2354,
+      "step": 7816
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7588293366409556e-06,
+      "loss": 1.055,
+      "step": 7817
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.752792384625985e-06,
+      "loss": 1.29,
+      "step": 7818
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7467618579885333e-06,
+      "loss": 1.4861,
+      "step": 7819
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7407377575487502e-06,
+      "loss": 1.6572,
+      "step": 7820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.734720084125847e-06,
+      "loss": 1.522,
+      "step": 7821
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7287088385382186e-06,
+      "loss": 1.2189,
+      "step": 7822
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7227040216033507e-06,
+      "loss": 1.5633,
+      "step": 7823
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.716705634137884e-06,
+      "loss": 0.4799,
+      "step": 7824
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.710713676957555e-06,
+      "loss": 1.1085,
+      "step": 7825
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7047281508772394e-06,
+      "loss": 1.078,
+      "step": 7826
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.698749056710942e-06,
+      "loss": 0.9192,
+      "step": 7827
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.69277639527179e-06,
+      "loss": 1.2218,
+      "step": 7828
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.686810167372034e-06,
+      "loss": 1.2432,
+      "step": 7829
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.680850373823035e-06,
+      "loss": 0.8483,
+      "step": 7830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.674897015435329e-06,
+      "loss": 0.6005,
+      "step": 7831
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.668950093018513e-06,
+      "loss": 1.413,
+      "step": 7832
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.663009607381345e-06,
+      "loss": 1.151,
+      "step": 7833
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6570755593317065e-06,
+      "loss": 0.6901,
+      "step": 7834
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6511479496765855e-06,
+      "loss": 1.5302,
+      "step": 7835
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6452267792221208e-06,
+      "loss": 1.2861,
+      "step": 7836
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.639312048773551e-06,
+      "loss": 0.7424,
+      "step": 7837
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6334037591352502e-06,
+      "loss": 1.4788,
+      "step": 7838
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6275019111107247e-06,
+      "loss": 1.5954,
+      "step": 7839
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.621606505502583e-06,
+      "loss": 0.6935,
+      "step": 7840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.615717543112578e-06,
+      "loss": 1.1251,
+      "step": 7841
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6098350247415748e-06,
+      "loss": 1.2732,
+      "step": 7842
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.603958951189561e-06,
+      "loss": 1.3054,
+      "step": 7843
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5980893232556647e-06,
+      "loss": 0.7756,
+      "step": 7844
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5922261417381187e-06,
+      "loss": 0.448,
+      "step": 7845
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.586369407434286e-06,
+      "loss": 0.9022,
+      "step": 7846
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5805191211406455e-06,
+      "loss": 1.1802,
+      "step": 7847
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5746752836528167e-06,
+      "loss": 0.75,
+      "step": 7848
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.568837895765536e-06,
+      "loss": 1.3889,
+      "step": 7849
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5630069582726344e-06,
+      "loss": 0.6607,
+      "step": 7850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5571824719671065e-06,
+      "loss": 1.4611,
+      "step": 7851
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5513644376410573e-06,
+      "loss": 0.5877,
+      "step": 7852
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5455528560856976e-06,
+      "loss": 1.7053,
+      "step": 7853
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5397477280913795e-06,
+      "loss": 0.9259,
+      "step": 7854
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5339490544475653e-06,
+      "loss": 1.1708,
+      "step": 7855
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5281568359428633e-06,
+      "loss": 0.909,
+      "step": 7856
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5223710733649486e-06,
+      "loss": 1.3275,
+      "step": 7857
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5165917675006867e-06,
+      "loss": 1.1067,
+      "step": 7858
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5108189191360266e-06,
+      "loss": 0.9931,
+      "step": 7859
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5050525290560456e-06,
+      "loss": 1.4242,
+      "step": 7860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.499292598044928e-06,
+      "loss": 1.0036,
+      "step": 7861
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4935391268860187e-06,
+      "loss": 0.7872,
+      "step": 7862
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4877921163617533e-06,
+      "loss": 1.4276,
+      "step": 7863
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4820515672536848e-06,
+      "loss": 1.0669,
+      "step": 7864
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4763174803424993e-06,
+      "loss": 1.2128,
+      "step": 7865
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.470589856408012e-06,
+      "loss": 1.3831,
+      "step": 7866
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4648686962291557e-06,
+      "loss": 1.3911,
+      "step": 7867
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4591540005839688e-06,
+      "loss": 1.0685,
+      "step": 7868
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4534457702496074e-06,
+      "loss": 1.3203,
+      "step": 7869
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.447744006002395e-06,
+      "loss": 0.5487,
+      "step": 7870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4420487086177167e-06,
+      "loss": 0.5287,
+      "step": 7871
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.436359878870109e-06,
+      "loss": 1.1695,
+      "step": 7872
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.43067751753322e-06,
+      "loss": 1.1234,
+      "step": 7873
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4250016253798358e-06,
+      "loss": 0.6331,
+      "step": 7874
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4193322031818346e-06,
+      "loss": 1.4156,
+      "step": 7875
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4136692517102323e-06,
+      "loss": 1.3359,
+      "step": 7876
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4080127717351686e-06,
+      "loss": 1.2712,
+      "step": 7877
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4023627640258837e-06,
+      "loss": 1.7596,
+      "step": 7878
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.396719229350758e-06,
+      "loss": 0.7951,
+      "step": 7879
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3910821684772768e-06,
+      "loss": 1.3877,
+      "step": 7880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3854515821720656e-06,
+      "loss": 1.3367,
+      "step": 7881
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3798274712008407e-06,
+      "loss": 0.5911,
+      "step": 7882
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3742098363284503e-06,
+      "loss": 0.7147,
+      "step": 7883
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3685986783188784e-06,
+      "loss": 1.3159,
+      "step": 7884
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.362993997935198e-06,
+      "loss": 0.8709,
+      "step": 7885
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3573957959396265e-06,
+      "loss": 1.2112,
+      "step": 7886
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3518040730934886e-06,
+      "loss": 1.2792,
+      "step": 7887
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3462188301572318e-06,
+      "loss": 0.5273,
+      "step": 7888
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3406400678904204e-06,
+      "loss": 1.2404,
+      "step": 7889
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3350677870517313e-06,
+      "loss": 1.3929,
+      "step": 7890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.329501988398969e-06,
+      "loss": 1.345,
+      "step": 7891
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3239426726890555e-06,
+      "loss": 1.4662,
+      "step": 7892
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3183898406780313e-06,
+      "loss": 1.4013,
+      "step": 7893
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3128434931210476e-06,
+      "loss": 0.4501,
+      "step": 7894
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3073036307723837e-06,
+      "loss": 1.4461,
+      "step": 7895
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.301770254385427e-06,
+      "loss": 1.0238,
+      "step": 7896
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2962433647126913e-06,
+      "loss": 1.2042,
+      "step": 7897
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.290722962505809e-06,
+      "loss": 1.0644,
+      "step": 7898
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.285209048515524e-06,
+      "loss": 1.4196,
+      "step": 7899
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2797016234916925e-06,
+      "loss": 1.153,
+      "step": 7900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.274200688183309e-06,
+      "loss": 1.3248,
+      "step": 7901
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.268706243338464e-06,
+      "loss": 1.5508,
+      "step": 7902
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2632182897043773e-06,
+      "loss": 1.0588,
+      "step": 7903
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2577368280273793e-06,
+      "loss": 1.5114,
+      "step": 7904
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2522618590529233e-06,
+      "loss": 1.6859,
+      "step": 7905
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2467933835255806e-06,
+      "loss": 0.9003,
+      "step": 7906
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.241331402189023e-06,
+      "loss": 1.4568,
+      "step": 7907
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.235875915786062e-06,
+      "loss": 1.1698,
+      "step": 7908
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.230426925058615e-06,
+      "loss": 1.2863,
+      "step": 7909
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.224984430747723e-06,
+      "loss": 0.8825,
+      "step": 7910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.219548433593516e-06,
+      "loss": 1.2698,
+      "step": 7911
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2141189343352798e-06,
+      "loss": 1.2371,
+      "step": 7912
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2086959337114023e-06,
+      "loss": 0.6399,
+      "step": 7913
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.203279432459371e-06,
+      "loss": 0.7874,
+      "step": 7914
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.197869431315802e-06,
+      "loss": 0.6364,
+      "step": 7915
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1924659310164354e-06,
+      "loss": 0.9371,
+      "step": 7916
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.187068932296121e-06,
+      "loss": 1.287,
+      "step": 7917
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.181678435888812e-06,
+      "loss": 0.7454,
+      "step": 7918
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.176294442527599e-06,
+      "loss": 1.0413,
+      "step": 7919
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1709169529446737e-06,
+      "loss": 1.3381,
+      "step": 7920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1655459678713464e-06,
+      "loss": 1.0197,
+      "step": 7921
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.160181488038038e-06,
+      "loss": 0.9728,
+      "step": 7922
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.154823514174298e-06,
+      "loss": 0.7384,
+      "step": 7923
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1494720470087882e-06,
+      "loss": 1.0689,
+      "step": 7924
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1441270872692655e-06,
+      "loss": 0.4889,
+      "step": 7925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1387886356826257e-06,
+      "loss": 1.2101,
+      "step": 7926
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1334566929748778e-06,
+      "loss": 1.7546,
+      "step": 7927
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1281312598711247e-06,
+      "loss": 0.9152,
+      "step": 7928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1228123370956033e-06,
+      "loss": 1.3372,
+      "step": 7929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.117499925371658e-06,
+      "loss": 1.3642,
+      "step": 7930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.11219402542176e-06,
+      "loss": 0.7353,
+      "step": 7931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1068946379674713e-06,
+      "loss": 0.7582,
+      "step": 7932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.101601763729488e-06,
+      "loss": 1.0808,
+      "step": 7933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.096315403427612e-06,
+      "loss": 1.4672,
+      "step": 7934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.091035557780763e-06,
+      "loss": 1.078,
+      "step": 7935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.085762227506971e-06,
+      "loss": 1.1226,
+      "step": 7936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0804954133233743e-06,
+      "loss": 1.4629,
+      "step": 7937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0752351159462557e-06,
+      "loss": 1.0432,
+      "step": 7938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0699813360909702e-06,
+      "loss": 0.7708,
+      "step": 7939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0647340744720024e-06,
+      "loss": 0.6467,
+      "step": 7940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0594933318029584e-06,
+      "loss": 1.6211,
+      "step": 7941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.054259108796558e-06,
+      "loss": 1.3449,
+      "step": 7942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.04903140616462e-06,
+      "loss": 1.5822,
+      "step": 7943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0438102246180868e-06,
+      "loss": 0.9457,
+      "step": 7944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.038595564867013e-06,
+      "loss": 0.7608,
+      "step": 7945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.03338742762057e-06,
+      "loss": 0.6585,
+      "step": 7946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0281858135870358e-06,
+      "loss": 1.3227,
+      "step": 7947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0229907234737945e-06,
+      "loss": 1.8095,
+      "step": 7948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.017802157987364e-06,
+      "loss": 0.9132,
+      "step": 7949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.012620117833358e-06,
+      "loss": 1.0478,
+      "step": 7950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.007444603716502e-06,
+      "loss": 0.7057,
+      "step": 7951
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.002275616340643e-06,
+      "loss": 0.6881,
+      "step": 7952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.997113156408742e-06,
+      "loss": 0.6306,
+      "step": 7953
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9919572246228644e-06,
+      "loss": 0.7534,
+      "step": 7954
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9868078216841833e-06,
+      "loss": 0.8345,
+      "step": 7955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.981664948293005e-06,
+      "loss": 1.1824,
+      "step": 7956
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9765286051487144e-06,
+      "loss": 0.6271,
+      "step": 7957
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9713987929498413e-06,
+      "loss": 1.5165,
+      "step": 7958
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.966275512394011e-06,
+      "loss": 1.3857,
+      "step": 7959
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.961158764177967e-06,
+      "loss": 1.1828,
+      "step": 7960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.956048548997558e-06,
+      "loss": 0.4836,
+      "step": 7961
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9509448675477436e-06,
+      "loss": 0.8291,
+      "step": 7962
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9458477205226087e-06,
+      "loss": 0.7033,
+      "step": 7963
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.940757108615315e-06,
+      "loss": 0.7081,
+      "step": 7964
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.935673032518187e-06,
+      "loss": 1.32,
+      "step": 7965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9305954929226158e-06,
+      "loss": 1.0435,
+      "step": 7966
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9255244905191373e-06,
+      "loss": 0.9711,
+      "step": 7967
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9204600259973616e-06,
+      "loss": 1.1161,
+      "step": 7968
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9154021000460433e-06,
+      "loss": 1.4609,
+      "step": 7969
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9103507133530317e-06,
+      "loss": 1.5044,
+      "step": 7970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9053058666052892e-06,
+      "loss": 1.4331,
+      "step": 7971
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9002675604888775e-06,
+      "loss": 1.2224,
+      "step": 7972
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8952357956889988e-06,
+      "loss": 0.7606,
+      "step": 7973
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8902105728899444e-06,
+      "loss": 1.2319,
+      "step": 7974
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8851918927751122e-06,
+      "loss": 1.3172,
+      "step": 7975
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8801797560270062e-06,
+      "loss": 1.1922,
+      "step": 7976
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8751741633272813e-06,
+      "loss": 1.5207,
+      "step": 7977
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8701751153566427e-06,
+      "loss": 0.5435,
+      "step": 7978
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8651826127949467e-06,
+      "loss": 0.5113,
+      "step": 7979
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8601966563211504e-06,
+      "loss": 1.571,
+      "step": 7980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8552172466133223e-06,
+      "loss": 1.4136,
+      "step": 7981
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8502443843486262e-06,
+      "loss": 0.7704,
+      "step": 7982
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.845278070203349e-06,
+      "loss": 1.4383,
+      "step": 7983
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8403183048528894e-06,
+      "loss": 0.6222,
+      "step": 7984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8353650889717467e-06,
+      "loss": 1.6387,
+      "step": 7985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8304184232335265e-06,
+      "loss": 1.2913,
+      "step": 7986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8254783083109517e-06,
+      "loss": 0.801,
+      "step": 7987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8205447448758684e-06,
+      "loss": 1.2033,
+      "step": 7988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8156177335991952e-06,
+      "loss": 1.3114,
+      "step": 7989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8106972751509909e-06,
+      "loss": 0.6972,
+      "step": 7990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.805783370200409e-06,
+      "loss": 1.408,
+      "step": 7991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8008760194157203e-06,
+      "loss": 1.4063,
+      "step": 7992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7959752234642912e-06,
+      "loss": 0.7856,
+      "step": 7993
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7910809830126107e-06,
+      "loss": 1.2629,
+      "step": 7994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7861932987262687e-06,
+      "loss": 0.8123,
+      "step": 7995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7813121712699665e-06,
+      "loss": 1.3631,
+      "step": 7996
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7764376013075123e-06,
+      "loss": 1.1081,
+      "step": 7997
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7715695895018202e-06,
+      "loss": 1.042,
+      "step": 7998
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7667081365149162e-06,
+      "loss": 0.8727,
+      "step": 7999
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7618532430079326e-06,
+      "loss": 0.6365,
+      "step": 8000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7570049096411135e-06,
+      "loss": 1.3108,
+      "step": 8001
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7521631370738034e-06,
+      "loss": 1.208,
+      "step": 8002
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7473279259644537e-06,
+      "loss": 0.5022,
+      "step": 8003
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7424992769706383e-06,
+      "loss": 1.2161,
+      "step": 8004
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7376771907490263e-06,
+      "loss": 1.4523,
+      "step": 8005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7328616679553877e-06,
+      "loss": 1.0405,
+      "step": 8006
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7280527092446209e-06,
+      "loss": 1.4805,
+      "step": 8007
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7232503152707136e-06,
+      "loss": 0.6997,
+      "step": 8008
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7184544866867658e-06,
+      "loss": 1.0297,
+      "step": 8009
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.713665224144989e-06,
+      "loss": 1.2437,
+      "step": 8010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7088825282966957e-06,
+      "loss": 0.8132,
+      "step": 8011
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7041063997923101e-06,
+      "loss": 1.1059,
+      "step": 8012
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.699336839281368e-06,
+      "loss": 0.5892,
+      "step": 8013
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6945738474124783e-06,
+      "loss": 1.2827,
+      "step": 8014
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.689817424833412e-06,
+      "loss": 0.9297,
+      "step": 8015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.685067572191007e-06,
+      "loss": 0.6834,
+      "step": 8016
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.68032429013123e-06,
+      "loss": 1.1215,
+      "step": 8017
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6755875792991204e-06,
+      "loss": 1.3807,
+      "step": 8018
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6708574403388633e-06,
+      "loss": 1.1461,
+      "step": 8019
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.666133873893727e-06,
+      "loss": 1.2344,
+      "step": 8020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6614168806060982e-06,
+      "loss": 1.1786,
+      "step": 8021
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6567064611174576e-06,
+      "loss": 1.0306,
+      "step": 8022
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6520026160683987e-06,
+      "loss": 1.326,
+      "step": 8023
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6473053460986321e-06,
+      "loss": 1.0913,
+      "step": 8024
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6426146518469409e-06,
+      "loss": 1.3224,
+      "step": 8025
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6379305339512484e-06,
+      "loss": 0.3482,
+      "step": 8026
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6332529930485728e-06,
+      "loss": 0.9734,
+      "step": 8027
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6285820297750331e-06,
+      "loss": 0.8688,
+      "step": 8028
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623917644765849e-06,
+      "loss": 1.3402,
+      "step": 8029
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6192598386553626e-06,
+      "loss": 1.4506,
+      "step": 8030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.614608612077012e-06,
+      "loss": 1.2309,
+      "step": 8031
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.609963965663336e-06,
+      "loss": 1.4531,
+      "step": 8032
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.605325900045973e-06,
+      "loss": 1.0488,
+      "step": 8033
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6006944158556913e-06,
+      "loss": 0.9309,
+      "step": 8034
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5960695137223481e-06,
+      "loss": 1.2614,
+      "step": 8035
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5914511942748956e-06,
+      "loss": 0.7588,
+      "step": 8036
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5868394581414092e-06,
+      "loss": 0.5951,
+      "step": 8037
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5822343059490652e-06,
+      "loss": 1.3736,
+      "step": 8038
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5776357383241292e-06,
+      "loss": 0.6056,
+      "step": 8039
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5730437558919897e-06,
+      "loss": 1.0091,
+      "step": 8040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5684583592771252e-06,
+      "loss": 1.111,
+      "step": 8041
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5638795491031477e-06,
+      "loss": 0.9579,
+      "step": 8042
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.559307325992726e-06,
+      "loss": 1.0909,
+      "step": 8043
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.554741690567668e-06,
+      "loss": 1.3835,
+      "step": 8044
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5501826434488887e-06,
+      "loss": 0.9361,
+      "step": 8045
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5456301852563803e-06,
+      "loss": 1.6854,
+      "step": 8046
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5410843166092592e-06,
+      "loss": 0.872,
+      "step": 8047
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5365450381257307e-06,
+      "loss": 1.4399,
+      "step": 8048
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5320123504231343e-06,
+      "loss": 1.1696,
+      "step": 8049
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.527486254117877e-06,
+      "loss": 1.2743,
+      "step": 8050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.522966749825483e-06,
+      "loss": 1.0136,
+      "step": 8051
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5184538381605883e-06,
+      "loss": 0.579,
+      "step": 8052
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5139475197369245e-06,
+      "loss": 1.141,
+      "step": 8053
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5094477951673292e-06,
+      "loss": 1.1658,
+      "step": 8054
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5049546650637348e-06,
+      "loss": 1.059,
+      "step": 8055
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5004681300371914e-06,
+      "loss": 0.4969,
+      "step": 8056
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.495988190697839e-06,
+      "loss": 1.0256,
+      "step": 8057
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4915148476549346e-06,
+      "loss": 1.2746,
+      "step": 8058
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.487048101516819e-06,
+      "loss": 1.5067,
+      "step": 8059
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.482587952890957e-06,
+      "loss": 1.2362,
+      "step": 8060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.478134402383896e-06,
+      "loss": 1.7183,
+      "step": 8061
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.473687450601302e-06,
+      "loss": 1.1454,
+      "step": 8062
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.469247098147941e-06,
+      "loss": 0.561,
+      "step": 8063
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4648133456276692e-06,
+      "loss": 0.9025,
+      "step": 8064
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.460386193643465e-06,
+      "loss": 0.9685,
+      "step": 8065
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4559656427973856e-06,
+      "loss": 1.7374,
+      "step": 8066
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.451551693690617e-06,
+      "loss": 1.2094,
+      "step": 8067
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.447144346923418e-06,
+      "loss": 1.2416,
+      "step": 8068
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4427436030951814e-06,
+      "loss": 1.1623,
+      "step": 8069
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4383494628043725e-06,
+      "loss": 0.8582,
+      "step": 8070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4339619266485804e-06,
+      "loss": 1.1542,
+      "step": 8071
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4295809952244832e-06,
+      "loss": 0.8928,
+      "step": 8072
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4252066691278654e-06,
+      "loss": 1.2566,
+      "step": 8073
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4208389489536177e-06,
+      "loss": 1.3551,
+      "step": 8074
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4164778352957154e-06,
+      "loss": 1.3174,
+      "step": 8075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4121233287472558e-06,
+      "loss": 1.0553,
+      "step": 8076
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4077754299004375e-06,
+      "loss": 0.7517,
+      "step": 8077
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4034341393465432e-06,
+      "loss": 1.158,
+      "step": 8078
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.399099457675962e-06,
+      "loss": 1.0371,
+      "step": 8079
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3947713854781942e-06,
+      "loss": 1.2163,
+      "step": 8080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3904499233418411e-06,
+      "loss": 1.6825,
+      "step": 8081
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3861350718545885e-06,
+      "loss": 0.9604,
+      "step": 8082
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.381826831603228e-06,
+      "loss": 1.183,
+      "step": 8083
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3775252031736796e-06,
+      "loss": 1.5184,
+      "step": 8084
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3732301871509368e-06,
+      "loss": 0.774,
+      "step": 8085
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3689417841190821e-06,
+      "loss": 1.1803,
+      "step": 8086
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.364659994661338e-06,
+      "loss": 1.2142,
+      "step": 8087
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3603848193599943e-06,
+      "loss": 0.9302,
+      "step": 8088
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3561162587964582e-06,
+      "loss": 1.009,
+      "step": 8089
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3518543135512208e-06,
+      "loss": 1.486,
+      "step": 8090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.347598984203896e-06,
+      "loss": 0.6401,
+      "step": 8091
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.343350271333188e-06,
+      "loss": 0.876,
+      "step": 8092
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3391081755168954e-06,
+      "loss": 0.8448,
+      "step": 8093
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3348726973319125e-06,
+      "loss": 0.6501,
+      "step": 8094
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3306438373542617e-06,
+      "loss": 1.3476,
+      "step": 8095
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.326421596159033e-06,
+      "loss": 1.1629,
+      "step": 8096
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.322205974320434e-06,
+      "loss": 1.143,
+      "step": 8097
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3179969724117613e-06,
+      "loss": 1.3928,
+      "step": 8098
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3137945910054295e-06,
+      "loss": 1.0794,
+      "step": 8099
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3095988306729312e-06,
+      "loss": 0.676,
+      "step": 8100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3054096919848714e-06,
+      "loss": 1.1259,
+      "step": 8101
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3012271755109495e-06,
+      "loss": 1.0478,
+      "step": 8102
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2970512818199665e-06,
+      "loss": 1.1317,
+      "step": 8103
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.292882011479829e-06,
+      "loss": 0.3536,
+      "step": 8104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.288719365057528e-06,
+      "loss": 0.9417,
+      "step": 8105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2845633431191662e-06,
+      "loss": 1.3453,
+      "step": 8106
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2804139462299413e-06,
+      "loss": 1.1988,
+      "step": 8107
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2762711749541466e-06,
+      "loss": 0.9061,
+      "step": 8108
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2721350298551816e-06,
+      "loss": 1.1082,
+      "step": 8109
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2680055114955348e-06,
+      "loss": 0.8043,
+      "step": 8110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.263882620436807e-06,
+      "loss": 0.7487,
+      "step": 8111
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.259766357239689e-06,
+      "loss": 0.6456,
+      "step": 8112
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2556567224639714e-06,
+      "loss": 0.5959,
+      "step": 8113
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2515537166685354e-06,
+      "loss": 1.2694,
+      "step": 8114
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2474573404113787e-06,
+      "loss": 1.8185,
+      "step": 8115
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2433675942495838e-06,
+      "loss": 1.3347,
+      "step": 8116
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.239284478739333e-06,
+      "loss": 0.9982,
+      "step": 8117
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2352079944359107e-06,
+      "loss": 1.2261,
+      "step": 8118
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2311381418937008e-06,
+      "loss": 0.8428,
+      "step": 8119
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.227074921666177e-06,
+      "loss": 0.5106,
+      "step": 8120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.22301833430592e-06,
+      "loss": 0.9254,
+      "step": 8121
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2189683803646047e-06,
+      "loss": 1.2854,
+      "step": 8122
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2149250603930073e-06,
+      "loss": 1.1224,
+      "step": 8123
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2108883749409938e-06,
+      "loss": 0.8627,
+      "step": 8124
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.206858324557525e-06,
+      "loss": 0.9057,
+      "step": 8125
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2028349097906788e-06,
+      "loss": 0.6685,
+      "step": 8126
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1988181311876179e-06,
+      "loss": 0.8373,
+      "step": 8127
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1948079892945996e-06,
+      "loss": 1.2622,
+      "step": 8128
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1908044846569876e-06,
+      "loss": 1.3896,
+      "step": 8129
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.18680761781923e-06,
+      "loss": 1.1861,
+      "step": 8130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1828173893248918e-06,
+      "loss": 0.8904,
+      "step": 8131
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1788337997166055e-06,
+      "loss": 0.984,
+      "step": 8132
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1748568495361378e-06,
+      "loss": 0.871,
+      "step": 8133
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1708865393243228e-06,
+      "loss": 1.534,
+      "step": 8134
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1669228696211122e-06,
+      "loss": 0.5388,
+      "step": 8135
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1629658409655243e-06,
+      "loss": 1.7632,
+      "step": 8136
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1590154538957177e-06,
+      "loss": 0.9026,
+      "step": 8137
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1550717089489126e-06,
+      "loss": 1.2764,
+      "step": 8138
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1511346066614415e-06,
+      "loss": 0.4696,
+      "step": 8139
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1472041475687256e-06,
+      "loss": 1.1147,
+      "step": 8140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1432803322052877e-06,
+      "loss": 1.277,
+      "step": 8141
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1393631611047618e-06,
+      "loss": 1.4626,
+      "step": 8142
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.135452634799844e-06,
+      "loss": 1.2112,
+      "step": 8143
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1315487538223424e-06,
+      "loss": 0.6609,
+      "step": 8144
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1276515187031933e-06,
+      "loss": 1.0255,
+      "step": 8145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1237609299723728e-06,
+      "loss": 1.3405,
+      "step": 8146
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1198769881589854e-06,
+      "loss": 1.0675,
+      "step": 8147
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1159996937912364e-06,
+      "loss": 0.8453,
+      "step": 8148
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1121290473964152e-06,
+      "loss": 0.9717,
+      "step": 8149
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1082650495009118e-06,
+      "loss": 1.0667,
+      "step": 8150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1044077006301946e-06,
+      "loss": 0.7848,
+      "step": 8151
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1005570013088662e-06,
+      "loss": 0.9893,
+      "step": 8152
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0967129520605912e-06,
+      "loss": 1.3644,
+      "step": 8153
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0928755534081347e-06,
+      "loss": 1.0661,
+      "step": 8154
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0890448058733626e-06,
+      "loss": 1.3891,
+      "step": 8155
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0852207099772582e-06,
+      "loss": 0.9707,
+      "step": 8156
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0814032662398555e-06,
+      "loss": 1.3733,
+      "step": 8157
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0775924751803168e-06,
+      "loss": 1.237,
+      "step": 8158
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.073788337316889e-06,
+      "loss": 0.6876,
+      "step": 8159
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0699908531669134e-06,
+      "loss": 0.7229,
+      "step": 8160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0662000232468328e-06,
+      "loss": 0.5045,
+      "step": 8161
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0624158480721792e-06,
+      "loss": 1.1696,
+      "step": 8162
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.058638328157574e-06,
+      "loss": 1.0485,
+      "step": 8163
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0548674640167511e-06,
+      "loss": 0.5003,
+      "step": 8164
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0511032561625223e-06,
+      "loss": 1.046,
+      "step": 8165
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0473457051068003e-06,
+      "loss": 1.6029,
+      "step": 8166
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0435948113605931e-06,
+      "loss": 1.5455,
+      "step": 8167
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0398505754340093e-06,
+      "loss": 1.5328,
+      "step": 8168
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.036112997836236e-06,
+      "loss": 1.0886,
+      "step": 8169
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0323820790755722e-06,
+      "loss": 1.5325,
+      "step": 8170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.028657819659401e-06,
+      "loss": 1.1005,
+      "step": 8171
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0249402200942004e-06,
+      "loss": 1.5509,
+      "step": 8172
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.021229280885544e-06,
+      "loss": 1.1318,
+      "step": 8173
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0175250025381112e-06,
+      "loss": 1.1736,
+      "step": 8174
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.013827385555649e-06,
+      "loss": 1.0801,
+      "step": 8175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0101364304410277e-06,
+      "loss": 1.226,
+      "step": 8176
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0064521376961899e-06,
+      "loss": 0.8765,
+      "step": 8177
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0027745078221851e-06,
+      "loss": 1.1221,
+      "step": 8178
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.991035413191463e-07,
+      "loss": 1.0717,
+      "step": 8179
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.954392386863132e-07,
+      "loss": 1.0456,
+      "step": 8180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.917816004220148e-07,
+      "loss": 1.1952,
+      "step": 8181
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.881306270236535e-07,
+      "loss": 0.6084,
+      "step": 8182
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.844863189877596e-07,
+      "loss": 1.1707,
+      "step": 8183
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.808486768099367e-07,
+      "loss": 1.1906,
+      "step": 8184
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.772177009848837e-07,
+      "loss": 1.3223,
+      "step": 8185
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.735933920063945e-07,
+      "loss": 1.1269,
+      "step": 8186
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.699757503673524e-07,
+      "loss": 1.2705,
+      "step": 8187
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.663647765597527e-07,
+      "loss": 0.8181,
+      "step": 8188
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.627604710746585e-07,
+      "loss": 1.3176,
+      "step": 8189
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.591628344022275e-07,
+      "loss": 0.9897,
+      "step": 8190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.555718670317349e-07,
+      "loss": 1.1685,
+      "step": 8191
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.519875694515346e-07,
+      "loss": 1.1814,
+      "step": 8192
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.484099421490589e-07,
+      "loss": 1.1959,
+      "step": 8193
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.448389856108575e-07,
+      "loss": 1.4068,
+      "step": 8194
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.412747003225641e-07,
+      "loss": 0.4744,
+      "step": 8195
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.377170867688967e-07,
+      "loss": 0.8315,
+      "step": 8196
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.341661454336681e-07,
+      "loss": 0.767,
+      "step": 8197
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.306218767997976e-07,
+      "loss": 1.5385,
+      "step": 8198
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.270842813492886e-07,
+      "loss": 1.6263,
+      "step": 8199
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.235533595632285e-07,
+      "loss": 0.9703,
+      "step": 8200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.200291119217996e-07,
+      "loss": 1.1899,
+      "step": 8201
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.165115389043022e-07,
+      "loss": 1.2699,
+      "step": 8202
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.130006409890923e-07,
+      "loss": 1.1497,
+      "step": 8203
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.094964186536325e-07,
+      "loss": 0.5546,
+      "step": 8204
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.059988723744806e-07,
+      "loss": 0.6984,
+      "step": 8205
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.025080026273003e-07,
+      "loss": 1.3768,
+      "step": 8206
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.99023809886812e-07,
+      "loss": 0.8051,
+      "step": 8207
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.955462946268533e-07,
+      "loss": 0.7705,
+      "step": 8208
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.920754573203571e-07,
+      "loss": 0.5512,
+      "step": 8209
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.886112984393292e-07,
+      "loss": 1.1143,
+      "step": 8210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.851538184548814e-07,
+      "loss": 1.0025,
+      "step": 8211
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.817030178372099e-07,
+      "loss": 1.2118,
+      "step": 8212
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.782588970556116e-07,
+      "loss": 0.9689,
+      "step": 8213
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.748214565784674e-07,
+      "loss": 1.5589,
+      "step": 8214
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.713906968732532e-07,
+      "loss": 0.5876,
+      "step": 8215
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.679666184065239e-07,
+      "loss": 0.8472,
+      "step": 8216
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.645492216439566e-07,
+      "loss": 0.4587,
+      "step": 8217
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.611385070502798e-07,
+      "loss": 0.9144,
+      "step": 8218
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.577344750893446e-07,
+      "loss": 0.8354,
+      "step": 8219
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.543371262240807e-07,
+      "loss": 0.8522,
+      "step": 8220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.509464609165019e-07,
+      "loss": 0.6856,
+      "step": 8221
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.475624796277337e-07,
+      "loss": 1.25,
+      "step": 8222
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.441851828179692e-07,
+      "loss": 0.9523,
+      "step": 8223
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.408145709465131e-07,
+      "loss": 0.8055,
+      "step": 8224
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.374506444717434e-07,
+      "loss": 0.823,
+      "step": 8225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.340934038511383e-07,
+      "loss": 0.6105,
+      "step": 8226
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.307428495412717e-07,
+      "loss": 0.4861,
+      "step": 8227
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.273989819977901e-07,
+      "loss": 1.3651,
+      "step": 8228
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.240618016754575e-07,
+      "loss": 0.7602,
+      "step": 8229
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.207313090280999e-07,
+      "loss": 1.4008,
+      "step": 8230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.174075045086493e-07,
+      "loss": 1.3813,
+      "step": 8231
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.140903885691331e-07,
+      "loss": 0.7059,
+      "step": 8232
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.107799616606571e-07,
+      "loss": 1.2263,
+      "step": 8233
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.074762242334222e-07,
+      "loss": 0.5636,
+      "step": 8234
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.041791767367191e-07,
+      "loss": 0.5638,
+      "step": 8235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.008888196189335e-07,
+      "loss": 1.2696,
+      "step": 8236
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.976051533275353e-07,
+      "loss": 1.2213,
+      "step": 8237
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.943281783090839e-07,
+      "loss": 1.509,
+      "step": 8238
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.910578950092284e-07,
+      "loss": 1.2973,
+      "step": 8239
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.877943038727187e-07,
+      "loss": 1.2028,
+      "step": 8240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.845374053433829e-07,
+      "loss": 1.4333,
+      "step": 8241
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.812871998641447e-07,
+      "loss": 0.8139,
+      "step": 8242
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.780436878770059e-07,
+      "loss": 1.5365,
+      "step": 8243
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.748068698230748e-07,
+      "loss": 0.5565,
+      "step": 8244
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.715767461425439e-07,
+      "loss": 0.468,
+      "step": 8245
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.683533172746949e-07,
+      "loss": 1.2908,
+      "step": 8246
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.651365836578828e-07,
+      "loss": 1.2204,
+      "step": 8247
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.619265457295855e-07,
+      "loss": 0.8623,
+      "step": 8248
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.587232039263481e-07,
+      "loss": 1.5045,
+      "step": 8249
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.555265586837946e-07,
+      "loss": 1.1703,
+      "step": 8250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.523366104366658e-07,
+      "loss": 1.0989,
+      "step": 8251
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.491533596187706e-07,
+      "loss": 1.4045,
+      "step": 8252
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.459768066630291e-07,
+      "loss": 1.1273,
+      "step": 8253
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.428069520014181e-07,
+      "loss": 1.346,
+      "step": 8254
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.39643796065026e-07,
+      "loss": 1.3868,
+      "step": 8255
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.364873392840366e-07,
+      "loss": 0.4535,
+      "step": 8256
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.333375820877008e-07,
+      "loss": 0.9432,
+      "step": 8257
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.301945249043707e-07,
+      "loss": 0.7524,
+      "step": 8258
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.270581681614874e-07,
+      "loss": 1.6673,
+      "step": 8259
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.239285122855877e-07,
+      "loss": 0.729,
+      "step": 8260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.208055577022754e-07,
+      "loss": 1.6038,
+      "step": 8261
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.176893048362554e-07,
+      "loss": 1.2537,
+      "step": 8262
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.145797541113386e-07,
+      "loss": 0.7945,
+      "step": 8263
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.114769059503978e-07,
+      "loss": 0.6727,
+      "step": 8264
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.083807607754011e-07,
+      "loss": 1.06,
+      "step": 8265
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.052913190074061e-07,
+      "loss": 1.4023,
+      "step": 8266
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.022085810665768e-07,
+      "loss": 1.0799,
+      "step": 8267
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.991325473721332e-07,
+      "loss": 1.4137,
+      "step": 8268
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.960632183424076e-07,
+      "loss": 0.5972,
+      "step": 8269
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.930005943948103e-07,
+      "loss": 1.3442,
+      "step": 8270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.899446759458472e-07,
+      "loss": 0.852,
+      "step": 8271
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.868954634111025e-07,
+      "loss": 0.7642,
+      "step": 8272
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.838529572052554e-07,
+      "loss": 1.2043,
+      "step": 8273
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.808171577420697e-07,
+      "loss": 1.0735,
+      "step": 8274
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.777880654344037e-07,
+      "loss": 1.2917,
+      "step": 8275
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.747656806941893e-07,
+      "loss": 0.9115,
+      "step": 8276
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.717500039324642e-07,
+      "loss": 0.6182,
+      "step": 8277
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.687410355593338e-07,
+      "loss": 1.06,
+      "step": 8278
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.65738775984015e-07,
+      "loss": 0.4637,
+      "step": 8279
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.62743225614798e-07,
+      "loss": 1.1442,
+      "step": 8280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.597543848590515e-07,
+      "loss": 1.2084,
+      "step": 8281
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.567722541232557e-07,
+      "loss": 1.2632,
+      "step": 8282
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.537968338129529e-07,
+      "loss": 1.4,
+      "step": 8283
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.508281243327974e-07,
+      "loss": 0.8734,
+      "step": 8284
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.478661260865105e-07,
+      "loss": 1.0413,
+      "step": 8285
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.449108394769088e-07,
+      "loss": 0.9921,
+      "step": 8286
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.419622649058931e-07,
+      "loss": 0.7493,
+      "step": 8287
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.39020402774465e-07,
+      "loss": 1.536,
+      "step": 8288
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.360852534826933e-07,
+      "loss": 1.5489,
+      "step": 8289
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.331568174297475e-07,
+      "loss": 0.8673,
+      "step": 8290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.302350950138813e-07,
+      "loss": 1.2393,
+      "step": 8291
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.273200866324325e-07,
+      "loss": 0.9265,
+      "step": 8292
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.244117926818227e-07,
+      "loss": 0.5954,
+      "step": 8293
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.21510213557569e-07,
+      "loss": 1.3145,
+      "step": 8294
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.186153496542669e-07,
+      "loss": 1.3811,
+      "step": 8295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.157272013656179e-07,
+      "loss": 1.1896,
+      "step": 8296
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.128457690843692e-07,
+      "loss": 0.8505,
+      "step": 8297
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.099710532024072e-07,
+      "loss": 0.6193,
+      "step": 8298
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.071030541106637e-07,
+      "loss": 1.4913,
+      "step": 8299
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.042417721991766e-07,
+      "loss": 1.1562,
+      "step": 8300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.013872078570626e-07,
+      "loss": 0.5297,
+      "step": 8301
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.985393614725276e-07,
+      "loss": 1.3518,
+      "step": 8302
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.956982334328731e-07,
+      "loss": 1.6322,
+      "step": 8303
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.928638241244622e-07,
+      "loss": 1.3037,
+      "step": 8304
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.900361339327698e-07,
+      "loss": 1.4306,
+      "step": 8305
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.872151632423495e-07,
+      "loss": 0.5946,
+      "step": 8306
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.844009124368388e-07,
+      "loss": 1.2957,
+      "step": 8307
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.815933818989483e-07,
+      "loss": 1.4135,
+      "step": 8308
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.787925720105003e-07,
+      "loss": 0.8913,
+      "step": 8309
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.759984831523957e-07,
+      "loss": 0.5895,
+      "step": 8310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.732111157045972e-07,
+      "loss": 1.5802,
+      "step": 8311
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.704304700461849e-07,
+      "loss": 1.1037,
+      "step": 8312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.676565465553175e-07,
+      "loss": 1.0353,
+      "step": 8313
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.648893456092263e-07,
+      "loss": 1.2383,
+      "step": 8314
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.621288675842329e-07,
+      "loss": 1.2789,
+      "step": 8315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.593751128557534e-07,
+      "loss": 1.3674,
+      "step": 8316
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.566280817982883e-07,
+      "loss": 1.1218,
+      "step": 8317
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.53887774785411e-07,
+      "loss": 1.2283,
+      "step": 8318
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.5115419218979e-07,
+      "loss": 1.6134,
+      "step": 8319
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.484273343831892e-07,
+      "loss": 0.9007,
+      "step": 8320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.457072017364396e-07,
+      "loss": 1.1406,
+      "step": 8321
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.429937946194618e-07,
+      "loss": 1.357,
+      "step": 8322
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.402871134012666e-07,
+      "loss": 1.198,
+      "step": 8323
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.375871584499592e-07,
+      "loss": 1.3066,
+      "step": 8324
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.348939301327071e-07,
+      "loss": 0.7114,
+      "step": 8325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.322074288157841e-07,
+      "loss": 1.1485,
+      "step": 8326
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.295276548645312e-07,
+      "loss": 1.2071,
+      "step": 8327
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.268546086433901e-07,
+      "loss": 1.1598,
+      "step": 8328
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.241882905158813e-07,
+      "loss": 1.303,
+      "step": 8329
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.21528700844609e-07,
+      "loss": 1.5657,
+      "step": 8330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.188758399912675e-07,
+      "loss": 0.9181,
+      "step": 8331
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.162297083166235e-07,
+      "loss": 1.2276,
+      "step": 8332
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.135903061805503e-07,
+      "loss": 0.7014,
+      "step": 8333
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.109576339419775e-07,
+      "loss": 0.5123,
+      "step": 8334
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.083316919589465e-07,
+      "loss": 0.714,
+      "step": 8335
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.057124805885716e-07,
+      "loss": 1.168,
+      "step": 8336
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.031000001870456e-07,
+      "loss": 0.763,
+      "step": 8337
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.004942511096567e-07,
+      "loss": 0.9937,
+      "step": 8338
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.978952337107712e-07,
+      "loss": 1.1417,
+      "step": 8339
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.953029483438398e-07,
+      "loss": 1.0708,
+      "step": 8340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.927173953614084e-07,
+      "loss": 1.1831,
+      "step": 8341
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.901385751150955e-07,
+      "loss": 1.2579,
+      "step": 8342
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.875664879556041e-07,
+      "loss": 0.8103,
+      "step": 8343
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.850011342327265e-07,
+      "loss": 1.4703,
+      "step": 8344
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.824425142953393e-07,
+      "loss": 1.2135,
+      "step": 8345
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.798906284913973e-07,
+      "loss": 1.0204,
+      "step": 8346
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.773454771679453e-07,
+      "loss": 1.215,
+      "step": 8347
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7480706067111747e-07,
+      "loss": 1.3388,
+      "step": 8348
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7227537934612097e-07,
+      "loss": 1.0382,
+      "step": 8349
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6975043353724157e-07,
+      "loss": 1.1427,
+      "step": 8350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6723222358787676e-07,
+      "loss": 1.1216,
+      "step": 8351
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.647207498404804e-07,
+      "loss": 0.6677,
+      "step": 8352
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6221601263660154e-07,
+      "loss": 0.5425,
+      "step": 8353
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5971801231686765e-07,
+      "loss": 0.9833,
+      "step": 8354
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5722674922099586e-07,
+      "loss": 1.3377,
+      "step": 8355
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5474222368779295e-07,
+      "loss": 1.349,
+      "step": 8356
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.522644360551331e-07,
+      "loss": 1.0388,
+      "step": 8357
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.497933866599802e-07,
+      "loss": 1.0089,
+      "step": 8358
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.47329075838393e-07,
+      "loss": 0.5123,
+      "step": 8359
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4487150392550916e-07,
+      "loss": 1.0772,
+      "step": 8360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.424206712555223e-07,
+      "loss": 1.2817,
+      "step": 8361
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.399765781617604e-07,
+      "loss": 1.1935,
+      "step": 8362
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3753922497659085e-07,
+      "loss": 0.961,
+      "step": 8363
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.351086120314873e-07,
+      "loss": 0.8386,
+      "step": 8364
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.326847396569966e-07,
+      "loss": 1.44,
+      "step": 8365
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.302676081827606e-07,
+      "loss": 0.9111,
+      "step": 8366
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.27857217937494e-07,
+      "loss": 1.4829,
+      "step": 8367
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.254535692489903e-07,
+      "loss": 1.0044,
+      "step": 8368
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2305666244413235e-07,
+      "loss": 0.9159,
+      "step": 8369
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.206664978489039e-07,
+      "loss": 0.6264,
+      "step": 8370
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.182830757883449e-07,
+      "loss": 1.2501,
+      "step": 8371
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.159063965865795e-07,
+      "loss": 1.3475,
+      "step": 8372
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1353646056683795e-07,
+      "loss": 1.1769,
+      "step": 8373
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1117326805141246e-07,
+      "loss": 0.9592,
+      "step": 8374
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0881681936169035e-07,
+      "loss": 0.6163,
+      "step": 8375
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.064671148181265e-07,
+      "loss": 1.6507,
+      "step": 8376
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.041241547402819e-07,
+      "loss": 0.9394,
+      "step": 8377
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.017879394467738e-07,
+      "loss": 0.8552,
+      "step": 8378
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9945846925532046e-07,
+      "loss": 0.9788,
+      "step": 8379
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9713574448272396e-07,
+      "loss": 1.467,
+      "step": 8380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.948197654448593e-07,
+      "loss": 1.0979,
+      "step": 8381
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9251053245668023e-07,
+      "loss": 1.2366,
+      "step": 8382
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.90208045832241e-07,
+      "loss": 1.0552,
+      "step": 8383
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.879123058846634e-07,
+      "loss": 1.3252,
+      "step": 8384
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.856233129261588e-07,
+      "loss": 0.7186,
+      "step": 8385
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8334106726801154e-07,
+      "loss": 1.1783,
+      "step": 8386
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.810655692206011e-07,
+      "loss": 0.8485,
+      "step": 8387
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7879681909338547e-07,
+      "loss": 1.1845,
+      "step": 8388
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.765348171948957e-07,
+      "loss": 1.3276,
+      "step": 8389
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.742795638327634e-07,
+      "loss": 1.5058,
+      "step": 8390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.720310593136822e-07,
+      "loss": 1.1638,
+      "step": 8391
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6978930394343524e-07,
+      "loss": 1.3276,
+      "step": 8392
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6755429802690087e-07,
+      "loss": 1.0247,
+      "step": 8393
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6532604186801934e-07,
+      "loss": 0.7871,
+      "step": 8394
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.631045357698204e-07,
+      "loss": 0.9677,
+      "step": 8395
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.608897800344235e-07,
+      "loss": 1.3353,
+      "step": 8396
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.586817749630267e-07,
+      "loss": 1.3748,
+      "step": 8397
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5648052085590075e-07,
+      "loss": 1.3692,
+      "step": 8398
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5428601801240637e-07,
+      "loss": 1.2704,
+      "step": 8399
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5209826673098244e-07,
+      "loss": 1.4159,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4991726730915775e-07,
+      "loss": 0.8704,
+      "step": 8401
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4774302004353385e-07,
+      "loss": 1.4214,
+      "step": 8402
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4557552522980207e-07,
+      "loss": 1.4305,
+      "step": 8403
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4341478316271547e-07,
+      "loss": 1.2115,
+      "step": 8404
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4126079413614455e-07,
+      "loss": 1.4038,
+      "step": 8405
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3911355844300497e-07,
+      "loss": 1.0106,
+      "step": 8406
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3697307637531317e-07,
+      "loss": 1.3317,
+      "step": 8407
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.348393482241696e-07,
+      "loss": 0.9494,
+      "step": 8408
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3271237427974777e-07,
+      "loss": 0.5532,
+      "step": 8409
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.305921548313051e-07,
+      "loss": 1.211,
+      "step": 8410
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.284786901671721e-07,
+      "loss": 0.6178,
+      "step": 8411
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2637198057478e-07,
+      "loss": 1.2609,
+      "step": 8412
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.242720263406274e-07,
+      "loss": 1.2161,
+      "step": 8413
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2217882775029683e-07,
+      "loss": 0.7267,
+      "step": 8414
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2009238508844964e-07,
+      "loss": 0.6061,
+      "step": 8415
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1801269863883656e-07,
+      "loss": 1.2756,
+      "step": 8416
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1593976868428686e-07,
+      "loss": 1.3114,
+      "step": 8417
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.138735955066974e-07,
+      "loss": 1.1901,
+      "step": 8418
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1181417938706547e-07,
+      "loss": 0.7251,
+      "step": 8419
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.09761520605456e-07,
+      "loss": 1.3671,
+      "step": 8420
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.077156194410236e-07,
+      "loss": 0.4634,
+      "step": 8421
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.05676476172001e-07,
+      "loss": 1.8856,
+      "step": 8422
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.036440910756999e-07,
+      "loss": 1.2273,
+      "step": 8423
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0161846442851583e-07,
+      "loss": 1.1893,
+      "step": 8424
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.995995965059173e-07,
+      "loss": 1.4449,
+      "step": 8425
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9758748758246247e-07,
+      "loss": 1.408,
+      "step": 8426
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9558213793179355e-07,
+      "loss": 0.7378,
+      "step": 8427
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9358354782662577e-07,
+      "loss": 1.1129,
+      "step": 8428
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9159171753875283e-07,
+      "loss": 1.2871,
+      "step": 8429
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8960664733905794e-07,
+      "loss": 1.2837,
+      "step": 8430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.876283374974975e-07,
+      "loss": 1.16,
+      "step": 8431
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8565678828311184e-07,
+      "loss": 0.9473,
+      "step": 8432
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.836919999640253e-07,
+      "loss": 1.2723,
+      "step": 8433
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8173397280742974e-07,
+      "loss": 1.2369,
+      "step": 8434
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.797827070796177e-07,
+      "loss": 0.5354,
+      "step": 8435
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7783820304594365e-07,
+      "loss": 1.3491,
+      "step": 8436
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.759004609708571e-07,
+      "loss": 0.9676,
+      "step": 8437
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.73969481117875e-07,
+      "loss": 1.3613,
+      "step": 8438
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.720452637495985e-07,
+      "loss": 0.6409,
+      "step": 8439
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.701278091277237e-07,
+      "loss": 0.5064,
+      "step": 8440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.682171175129977e-07,
+      "loss": 1.2896,
+      "step": 8441
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6631318916527924e-07,
+      "loss": 1.1652,
+      "step": 8442
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6441602434348346e-07,
+      "loss": 1.174,
+      "step": 8443
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6252562330562056e-07,
+      "loss": 1.0319,
+      "step": 8444
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.606419863087739e-07,
+      "loss": 1.46,
+      "step": 8445
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5876511360910514e-07,
+      "loss": 1.1053,
+      "step": 8446
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.568950054618602e-07,
+      "loss": 0.7591,
+      "step": 8447
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.550316621213689e-07,
+      "loss": 1.6483,
+      "step": 8448
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5317508384103405e-07,
+      "loss": 0.5684,
+      "step": 8449
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.513252708733427e-07,
+      "loss": 1.5383,
+      "step": 8450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4948222346985463e-07,
+      "loss": 1.1311,
+      "step": 8451
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.476459418812138e-07,
+      "loss": 1.221,
+      "step": 8452
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4581642635715384e-07,
+      "loss": 1.1472,
+      "step": 8453
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.439936771464757e-07,
+      "loss": 0.6464,
+      "step": 8454
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4217769449705885e-07,
+      "loss": 0.9498,
+      "step": 8455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4036847865587245e-07,
+      "loss": 1.5675,
+      "step": 8456
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.385660298689585e-07,
+      "loss": 0.7229,
+      "step": 8457
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.367703483814432e-07,
+      "loss": 1.186,
+      "step": 8458
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3498143443752562e-07,
+      "loss": 1.1609,
+      "step": 8459
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3319928828049455e-07,
+      "loss": 1.1662,
+      "step": 8460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3142391015271158e-07,
+      "loss": 1.0811,
+      "step": 8461
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2965530029561143e-07,
+      "loss": 1.0105,
+      "step": 8462
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2789345894972945e-07,
+      "loss": 0.8573,
+      "step": 8463
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2613838635465733e-07,
+      "loss": 1.4527,
+      "step": 8464
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2439008274907636e-07,
+      "loss": 1.2789,
+      "step": 8465
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2264854837074635e-07,
+      "loss": 1.2006,
+      "step": 8466
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.209137834565167e-07,
+      "loss": 1.452,
+      "step": 8467
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1918578824228765e-07,
+      "loss": 1.0744,
+      "step": 8468
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1746456296307672e-07,
+      "loss": 0.7807,
+      "step": 8469
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1575010785295778e-07,
+      "loss": 1.0351,
+      "step": 8470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1404242314507772e-07,
+      "loss": 0.85,
+      "step": 8471
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.123415090716785e-07,
+      "loss": 1.3861,
+      "step": 8472
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1064736586408062e-07,
+      "loss": 1.1604,
+      "step": 8473
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0895999375267762e-07,
+      "loss": 0.7599,
+      "step": 8474
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.072793929669359e-07,
+      "loss": 1.107,
+      "step": 8475
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.056055637354115e-07,
+      "loss": 1.1413,
+      "step": 8476
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0393850628574462e-07,
+      "loss": 1.1449,
+      "step": 8477
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.022782208446372e-07,
+      "loss": 1.3289,
+      "step": 8478
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0062470763788088e-07,
+      "loss": 0.9922,
+      "step": 8479
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9897796689034574e-07,
+      "loss": 1.0077,
+      "step": 8480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9733799882598601e-07,
+      "loss": 1.2453,
+      "step": 8481
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.957048036678233e-07,
+      "loss": 1.5584,
+      "step": 8482
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9407838163796322e-07,
+      "loss": 1.573,
+      "step": 8483
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9245873295759e-07,
+      "loss": 0.7405,
+      "step": 8484
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9084585784697185e-07,
+      "loss": 0.4982,
+      "step": 8485
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8923975652545001e-07,
+      "loss": 0.4983,
+      "step": 8486
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8764042921143864e-07,
+      "loss": 1.4364,
+      "step": 8487
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8604787612245267e-07,
+      "loss": 0.7125,
+      "step": 8488
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8446209747505772e-07,
+      "loss": 0.7045,
+      "step": 8489
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8288309348492017e-07,
+      "loss": 1.3955,
+      "step": 8490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8131086436676824e-07,
+      "loss": 1.2501,
+      "step": 8491
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7974541033443093e-07,
+      "loss": 1.3042,
+      "step": 8492
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.781867316007879e-07,
+      "loss": 1.0739,
+      "step": 8493
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.766348283778141e-07,
+      "loss": 1.3275,
+      "step": 8494
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7508970087656286e-07,
+      "loss": 1.4279,
+      "step": 8495
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.735513493071661e-07,
+      "loss": 0.8689,
+      "step": 8496
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7201977387882874e-07,
+      "loss": 0.5269,
+      "step": 8497
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7049497479983966e-07,
+      "loss": 1.3274,
+      "step": 8498
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6897695227755528e-07,
+      "loss": 1.2797,
+      "step": 8499
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6746570651843262e-07,
+      "loss": 1.5081,
+      "step": 8500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6596123772797955e-07,
+      "loss": 1.3638,
+      "step": 8501
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6446354611081015e-07,
+      "loss": 1.0944,
+      "step": 8502
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.629726318705893e-07,
+      "loss": 1.0637,
+      "step": 8503
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6148849521007703e-07,
+      "loss": 1.5114,
+      "step": 8504
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6001113633111741e-07,
+      "loss": 1.2029,
+      "step": 8505
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5854055543461088e-07,
+      "loss": 1.047,
+      "step": 8506
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5707675272055854e-07,
+      "loss": 1.0611,
+      "step": 8507
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5561972838802898e-07,
+      "loss": 1.5525,
+      "step": 8508
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5416948263516363e-07,
+      "loss": 0.8443,
+      "step": 8509
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5272601565919364e-07,
+      "loss": 0.6349,
+      "step": 8510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5128932765642313e-07,
+      "loss": 1.2422,
+      "step": 8511
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.498594188222291e-07,
+      "loss": 1.5109,
+      "step": 8512
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4843628935107823e-07,
+      "loss": 0.697,
+      "step": 8513
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.470199394365046e-07,
+      "loss": 0.5391,
+      "step": 8514
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4561036927113192e-07,
+      "loss": 0.8338,
+      "step": 8515
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4420757904664017e-07,
+      "loss": 1.3384,
+      "step": 8516
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4281156895382119e-07,
+      "loss": 1.2078,
+      "step": 8517
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4142233918250648e-07,
+      "loss": 0.4898,
+      "step": 8518
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.400398899216282e-07,
+      "loss": 0.5998,
+      "step": 8519
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3866422135920265e-07,
+      "loss": 1.0427,
+      "step": 8520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3729533368230796e-07,
+      "loss": 1.2706,
+      "step": 8521
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3593322707710076e-07,
+      "loss": 1.2138,
+      "step": 8522
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3457790172882734e-07,
+      "loss": 1.0672,
+      "step": 8523
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3322935782180135e-07,
+      "loss": 0.7005,
+      "step": 8524
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3188759553942053e-07,
+      "loss": 1.443,
+      "step": 8525
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3055261506415562e-07,
+      "loss": 1.091,
+      "step": 8526
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.292244165775558e-07,
+      "loss": 1.055,
+      "step": 8527
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.279030002602488e-07,
+      "loss": 1.435,
+      "step": 8528
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2658836629194647e-07,
+      "loss": 1.504,
+      "step": 8529
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.25280514851428e-07,
+      "loss": 1.4028,
+      "step": 8530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2397944611655666e-07,
+      "loss": 1.3047,
+      "step": 8531
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2268516026426312e-07,
+      "loss": 1.2209,
+      "step": 8532
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.213976574705733e-07,
+      "loss": 1.1066,
+      "step": 8533
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2011693791058043e-07,
+      "loss": 0.7562,
+      "step": 8534
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1884300175845076e-07,
+      "loss": 1.6003,
+      "step": 8535
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1757584918743459e-07,
+      "loss": 1.6623,
+      "step": 8536
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1631548036985518e-07,
+      "loss": 1.0313,
+      "step": 8537
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1506189547712543e-07,
+      "loss": 0.638,
+      "step": 8538
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1381509467972007e-07,
+      "loss": 1.5514,
+      "step": 8539
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1257507814719238e-07,
+      "loss": 1.0926,
+      "step": 8540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1134184604819075e-07,
+      "loss": 0.6513,
+      "step": 8541
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1011539855041997e-07,
+      "loss": 1.1537,
+      "step": 8542
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0889573582067436e-07,
+      "loss": 1.3038,
+      "step": 8543
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.076828580248157e-07,
+      "loss": 0.3944,
+      "step": 8544
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0647676532779537e-07,
+      "loss": 1.3372,
+      "step": 8545
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0527745789363219e-07,
+      "loss": 1.3926,
+      "step": 8546
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.04084935885429e-07,
+      "loss": 0.7857,
+      "step": 8547
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0289919946536164e-07,
+      "loss": 1.2207,
+      "step": 8548
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.017202487946789e-07,
+      "loss": 0.9652,
+      "step": 8549
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0054808403372473e-07,
+      "loss": 0.9455,
+      "step": 8550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.938270534189386e-08,
+      "loss": 0.597,
+      "step": 8551
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.822411287767619e-08,
+      "loss": 1.0496,
+      "step": 8552
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.70723067986401e-08,
+      "loss": 1.1392,
+      "step": 8553
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.592728726141587e-08,
+      "loss": 0.5085,
+      "step": 8554
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.478905442172892e-08,
+      "loss": 1.3265,
+      "step": 8555
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.365760843436655e-08,
+      "loss": 1.1347,
+      "step": 8556
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.253294945320568e-08,
+      "loss": 0.9732,
+      "step": 8557
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.141507763118507e-08,
+      "loss": 1.548,
+      "step": 8558
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.03039931203442e-08,
+      "loss": 1.3655,
+      "step": 8559
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.919969607176782e-08,
+      "loss": 1.5253,
+      "step": 8560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.810218663564684e-08,
+      "loss": 1.1771,
+      "step": 8561
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.701146496122858e-08,
+      "loss": 1.1353,
+      "step": 8562
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.592753119685548e-08,
+      "loss": 1.3635,
+      "step": 8563
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.48503854899263e-08,
+      "loss": 0.6336,
+      "step": 8564
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.378002798693496e-08,
+      "loss": 0.5639,
+      "step": 8565
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.271645883344281e-08,
+      "loss": 0.9086,
+      "step": 8566
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.16596781740897e-08,
+      "loss": 1.029,
+      "step": 8567
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.060968615258846e-08,
+      "loss": 0.6512,
+      "step": 8568
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.956648291174151e-08,
+      "loss": 0.8594,
+      "step": 8569
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.853006859340762e-08,
+      "loss": 0.425,
+      "step": 8570
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.750044333854068e-08,
+      "loss": 1.4414,
+      "step": 8571
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.647760728716202e-08,
+      "loss": 0.5885,
+      "step": 8572
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.546156057837706e-08,
+      "loss": 1.1184,
+      "step": 8573
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.445230335035858e-08,
+      "loss": 1.3635,
+      "step": 8574
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.34498357403579e-08,
+      "loss": 1.1933,
+      "step": 8575
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.245415788471044e-08,
+      "loss": 1.313,
+      "step": 8576
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.146526991882452e-08,
+      "loss": 1.0521,
+      "step": 8577
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.048317197717591e-08,
+      "loss": 1.115,
+      "step": 8578
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.950786419333555e-08,
+      "loss": 1.0807,
+      "step": 8579
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.853934669993067e-08,
+      "loss": 0.3731,
+      "step": 8580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.757761962867814e-08,
+      "loss": 1.4953,
+      "step": 8581
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.662268311036779e-08,
+      "loss": 1.3028,
+      "step": 8582
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.567453727486795e-08,
+      "loss": 1.2976,
+      "step": 8583
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.473318225111991e-08,
+      "loss": 1.452,
+      "step": 8584
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.37986181671435e-08,
+      "loss": 1.4685,
+      "step": 8585
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.287084515003705e-08,
+      "loss": 1.2207,
+      "step": 8586
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.194986332596631e-08,
+      "loss": 0.7363,
+      "step": 8587
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.103567282018663e-08,
+      "loss": 1.44,
+      "step": 8588
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.012827375702079e-08,
+      "loss": 1.194,
+      "step": 8589
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.922766625986453e-08,
+      "loss": 0.9812,
+      "step": 8590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.833385045120876e-08,
+      "loss": 1.3054,
+      "step": 8591
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.744682645260069e-08,
+      "loss": 0.5712,
+      "step": 8592
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.656659438466605e-08,
+      "loss": 1.0107,
+      "step": 8593
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.569315436712019e-08,
+      "loss": 1.1951,
+      "step": 8594
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.482650651874588e-08,
+      "loss": 1.156,
+      "step": 8595
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3966650957393284e-08,
+      "loss": 0.6187,
+      "step": 8596
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.31135878000133e-08,
+      "loss": 1.1809,
+      "step": 8597
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.226731716260202e-08,
+      "loss": 1.4519,
+      "step": 8598
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.142783916026184e-08,
+      "loss": 1.0777,
+      "step": 8599
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0595153907145866e-08,
+      "loss": 1.0519,
+      "step": 8600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.976926151650796e-08,
+      "loss": 1.4438,
+      "step": 8601
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.895016210065273e-08,
+      "loss": 1.1377,
+      "step": 8602
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.813785577097996e-08,
+      "loss": 1.4678,
+      "step": 8603
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.733234263795683e-08,
+      "loss": 1.0463,
+      "step": 8604
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.65336228111346e-08,
+      "loss": 1.1422,
+      "step": 8605
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.57416963991264e-08,
+      "loss": 1.2065,
+      "step": 8606
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4956563509634955e-08,
+      "loss": 1.1022,
+      "step": 8607
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.417822424943041e-08,
+      "loss": 1.3329,
+      "step": 8608
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.340667872437254e-08,
+      "loss": 1.0826,
+      "step": 8609
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2641927039377416e-08,
+      "loss": 1.034,
+      "step": 8610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.188396929845073e-08,
+      "loss": 1.3063,
+      "step": 8611
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.113280560467114e-08,
+      "loss": 0.9984,
+      "step": 8612
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.038843606019582e-08,
+      "loss": 1.4912,
+      "step": 8613
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9650860766249354e-08,
+      "loss": 0.4452,
+      "step": 8614
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.892007982314039e-08,
+      "loss": 1.3919,
+      "step": 8615
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.819609333025609e-08,
+      "loss": 1.3602,
+      "step": 8616
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7478901386051034e-08,
+      "loss": 1.3816,
+      "step": 8617
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.676850408805277e-08,
+      "loss": 1.3212,
+      "step": 8618
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6064901532884e-08,
+      "loss": 1.2066,
+      "step": 8619
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.53680938162293e-08,
+      "loss": 0.7724,
+      "step": 8620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.467808103284065e-08,
+      "loss": 1.0102,
+      "step": 8621
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3994863276565204e-08,
+      "loss": 1.3435,
+      "step": 8622
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.331844064031753e-08,
+      "loss": 0.7087,
+      "step": 8623
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.264881321608515e-08,
+      "loss": 0.5721,
+      "step": 8624
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.198598109492856e-08,
+      "loss": 0.576,
+      "step": 8625
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1329944366997875e-08,
+      "loss": 1.0304,
+      "step": 8626
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.068070312150506e-08,
+      "loss": 0.5605,
+      "step": 8627
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0038257446751706e-08,
+      "loss": 1.1887,
+      "step": 8628
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.940260743009571e-08,
+      "loss": 1.3456,
+      "step": 8629
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8773753157990135e-08,
+      "loss": 1.1316,
+      "step": 8630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8151694715949918e-08,
+      "loss": 1.1433,
+      "step": 8631
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.75364321885796e-08,
+      "loss": 1.0794,
+      "step": 8632
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6927965659551135e-08,
+      "loss": 1.1855,
+      "step": 8633
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6326295211603903e-08,
+      "loss": 1.2263,
+      "step": 8634
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5731420926566885e-08,
+      "loss": 1.4019,
+      "step": 8635
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5143342885347588e-08,
+      "loss": 0.6641,
+      "step": 8636
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.456206116790982e-08,
+      "loss": 1.2223,
+      "step": 8637
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.398757585330702e-08,
+      "loss": 1.3563,
+      "step": 8638
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3419887019671127e-08,
+      "loss": 1.3429,
+      "step": 8639
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.28589947442015e-08,
+      "loss": 1.203,
+      "step": 8640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2304899103181564e-08,
+      "loss": 1.4533,
+      "step": 8641
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1757600171962154e-08,
+      "loss": 1.0621,
+      "step": 8642
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1217098024967076e-08,
+      "loss": 1.9216,
+      "step": 8643
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.068339273570974e-08,
+      "loss": 1.2781,
+      "step": 8644
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.015648437677098e-08,
+      "loss": 0.5578,
+      "step": 8645
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9636373019804588e-08,
+      "loss": 1.0539,
+      "step": 8646
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9123058735542875e-08,
+      "loss": 1.6458,
+      "step": 8647
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8616541593796665e-08,
+      "loss": 0.9384,
+      "step": 8648
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8116821663449745e-08,
+      "loss": 0.6372,
+      "step": 8649
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7623899012458866e-08,
+      "loss": 1.4124,
+      "step": 8650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7137773707859294e-08,
+      "loss": 1.092,
+      "step": 8651
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.665844581575926e-08,
+      "loss": 0.8942,
+      "step": 8652
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.618591540135661e-08,
+      "loss": 1.2259,
+      "step": 8653
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.57201825288944e-08,
+      "loss": 1.3256,
+      "step": 8654
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5261247261727506e-08,
+      "loss": 0.3783,
+      "step": 8655
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4809109662256016e-08,
+      "loss": 1.6999,
+      "step": 8656
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4363769791975178e-08,
+      "loss": 0.918,
+      "step": 8657
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3925227711447664e-08,
+      "loss": 1.511,
+      "step": 8658
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.349348348031465e-08,
+      "loss": 1.1793,
+      "step": 8659
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3068537157284733e-08,
+      "loss": 0.6349,
+      "step": 8660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2650388800156121e-08,
+      "loss": 1.2398,
+      "step": 8661
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2239038465788887e-08,
+      "loss": 1.1381,
+      "step": 8662
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1834486210127172e-08,
+      "loss": 1.1542,
+      "step": 8663
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1436732088188073e-08,
+      "loss": 0.8644,
+      "step": 8664
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1045776154067211e-08,
+      "loss": 1.284,
+      "step": 8665
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0661618460927613e-08,
+      "loss": 1.2115,
+      "step": 8666
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0284259061010826e-08,
+      "loss": 1.1251,
+      "step": 8667
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.913698005642457e-09,
+      "loss": 1.1196,
+      "step": 8668
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.549935345215532e-09,
+      "loss": 1.2997,
+      "step": 8669
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.192971129196037e-09,
+      "loss": 1.307,
+      "step": 8670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.842805406134025e-09,
+      "loss": 1.314,
+      "step": 8671
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.499438223641409e-09,
+      "loss": 1.3029,
+      "step": 8672
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.162869628425274e-09,
+      "loss": 0.9497,
+      "step": 8673
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.833099666254561e-09,
+      "loss": 1.4899,
+      "step": 8674
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.510128381971182e-09,
+      "loss": 1.0536,
+      "step": 8675
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.193955819501108e-09,
+      "loss": 1.3174,
+      "step": 8676
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.884582021843277e-09,
+      "loss": 1.3721,
+      "step": 8677
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.582007031069592e-09,
+      "loss": 1.3003,
+      "step": 8678
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.286230888324918e-09,
+      "loss": 1.2965,
+      "step": 8679
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.997253633838185e-09,
+      "loss": 1.2512,
+      "step": 8680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.715075306911289e-09,
+      "loss": 1.0706,
+      "step": 8681
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.4396959459079856e-09,
+      "loss": 1.3557,
+      "step": 8682
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.171115588292752e-09,
+      "loss": 1.0058,
+      "step": 8683
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.909334270580823e-09,
+      "loss": 1.1414,
+      "step": 8684
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.654352028377052e-09,
+      "loss": 0.81,
+      "step": 8685
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.406168896359253e-09,
+      "loss": 0.7749,
+      "step": 8686
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.164784908272656e-09,
+      "loss": 1.2965,
+      "step": 8687
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.930200096952108e-09,
+      "loss": 1.7893,
+      "step": 8688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.702414494294315e-09,
+      "loss": 1.418,
+      "step": 8689
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.481428131280051e-09,
+      "loss": 0.5097,
+      "step": 8690
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2672410379630537e-09,
+      "loss": 0.5035,
+      "step": 8691
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.059853243464472e-09,
+      "loss": 1.2915,
+      "step": 8692
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8592647760006253e-09,
+      "loss": 0.6263,
+      "step": 8693
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.66547566283859e-09,
+      "loss": 1.4625,
+      "step": 8694
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4784859303350616e-09,
+      "loss": 1.7072,
+      "step": 8695
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.298295603925249e-09,
+      "loss": 1.4709,
+      "step": 8696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1249047081117746e-09,
+      "loss": 1.2394,
+      "step": 8697
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.958313266470224e-09,
+      "loss": 1.5047,
+      "step": 8698
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7985213016658008e-09,
+      "loss": 0.735,
+      "step": 8699
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6455288354200182e-09,
+      "loss": 1.0856,
+      "step": 8700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4993358885384556e-09,
+      "loss": 0.4474,
+      "step": 8701
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3599424809107586e-09,
+      "loss": 1.1495,
+      "step": 8702
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2273486314939852e-09,
+      "loss": 1.5953,
+      "step": 8703
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1015543583070553e-09,
+      "loss": 0.4727,
+      "step": 8704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.82559678475159e-10,
+      "loss": 1.2584,
+      "step": 8705
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.703646081631433e-10,
+      "loss": 1.1282,
+      "step": 8706
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.649691626421262e-10,
+      "loss": 0.4987,
+      "step": 8707
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.663733562450869e-10,
+      "loss": 1.3108,
+      "step": 8708
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.745772023724172e-10,
+      "loss": 1.0791,
+      "step": 8709
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.89580713508575e-10,
+      "loss": 1.3227,
+      "step": 8710
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1138390122208437e-10,
+      "loss": 1.5111,
+      "step": 8711
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3998677613777953e-10,
+      "loss": 0.6488,
+      "step": 8712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.75389347970112e-10,
+      "loss": 0.5683,
+      "step": 8713
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.175916255009458e-10,
+      "loss": 0.9735,
+      "step": 8714
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.665936165962112e-10,
+      "loss": 1.0074,
+      "step": 8715
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.223953281836998e-10,
+      "loss": 0.5721,
+      "step": 8716
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.499676628082042e-11,
+      "loss": 1.2824,
+      "step": 8717
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.4397935966843394e-11,
+      "loss": 1.3168,
+      "step": 8718
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.059884141065617e-11,
+      "loss": 0.5996,
+      "step": 8719
+    },
+    {
+      "epoch": 1.0,
+      "step": 8719,
+      "total_flos": 1.4738059356587164e+19,
+      "train_loss": 1.1088871026168328,
+      "train_runtime": 132755.6737,
+      "train_samples_per_second": 0.066,
+      "train_steps_per_second": 0.066
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 8719,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 200,
+  "total_flos": 1.4738059356587164e+19,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}