diff --git "a/checkpoint-12345/trainer_state.json" "b/checkpoint-12345/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-12345/trainer_state.json"
@@ -0,0 +1,74123 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 3087,
+  "global_step": 12345,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 2.0547,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 2.032890558242798,
+      "eval_runtime": 20.2818,
+      "eval_samples_per_second": 140.668,
+      "eval_steps_per_second": 140.668,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 2.1238,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 2.084,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.2058,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 1.9004,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 1.6004,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 1.7267,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.7358,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-06,
+      "loss": 1.7764,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 2.1887,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2e-06,
+      "loss": 1.9318,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.9538,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6e-06,
+      "loss": 2.2747,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 2.2759,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 1.9374,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 1.7258,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 1.7104,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-06,
+      "loss": 1.6783,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8e-06,
+      "loss": 1.7593,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.6427,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 2.1587,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4e-06,
+      "loss": 2.1035,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6e-06,
+      "loss": 2.0779,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.8098,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 1.8985,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2e-06,
+      "loss": 1.9149,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4e-06,
+      "loss": 2.2817,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 2.1256,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.8e-06,
+      "loss": 1.972,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 2.043,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.2e-06,
+      "loss": 1.656,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 2.0033,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.6e-06,
+      "loss": 1.9015,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 2.0402,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 1.8836,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2e-06,
+      "loss": 1.9023,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.4e-06,
+      "loss": 1.8028,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.6e-06,
+      "loss": 2.0918,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.8e-06,
+      "loss": 2.2415,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 2.0609,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.200000000000001e-06,
+      "loss": 2.3361,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 1.9019,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.599999999999999e-06,
+      "loss": 1.8774,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.8e-06,
+      "loss": 1.8814,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-06,
+      "loss": 1.6173,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.2e-06,
+      "loss": 2.2005,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.4e-06,
+      "loss": 2.0755,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 2.312,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 1.7521,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 2.1226,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.02e-05,
+      "loss": 2.2262,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.04e-05,
+      "loss": 2.0589,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.06e-05,
+      "loss": 1.9286,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.08e-05,
+      "loss": 1.9837,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 6.3914,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 1.581,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 1.7532,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.16e-05,
+      "loss": 1.5707,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.18e-05,
+      "loss": 2.1914,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 1.6615,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.22e-05,
+      "loss": 1.8925,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.24e-05,
+      "loss": 1.6665,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 1.7238,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 1.6182,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 1.961,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.32e-05,
+      "loss": 1.8425,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 1.8268,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 1.8723,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 1.6913,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 1.5972,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.42e-05,
+      "loss": 2.0436,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.44e-05,
+      "loss": 2.1502,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4599999999999999e-05,
+      "loss": 1.8821,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.48e-05,
+      "loss": 1.9766,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-05,
+      "loss": 1.868,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.52e-05,
+      "loss": 1.8861,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.54e-05,
+      "loss": 1.9063,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.56e-05,
+      "loss": 1.8481,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.58e-05,
+      "loss": 1.7799,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.8899,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.62e-05,
+      "loss": 1.189,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6400000000000002e-05,
+      "loss": 1.6904,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.66e-05,
+      "loss": 1.5551,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 1.5708,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 1.9263,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7199999999999998e-05,
+      "loss": 1.6307,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.74e-05,
+      "loss": 1.7866,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.76e-05,
+      "loss": 1.586,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.78e-05,
+      "loss": 1.7984,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-05,
+      "loss": 1.8674,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 1.9193,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.84e-05,
+      "loss": 1.6426,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.86e-05,
+      "loss": 1.5341,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.88e-05,
+      "loss": 1.4254,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9e-05,
+      "loss": 1.7267,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 1.4526,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.94e-05,
+      "loss": 1.5522,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 1.6558,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 1.8961,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 1.94,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0200000000000003e-05,
+      "loss": 1.6231,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.04e-05,
+      "loss": 1.8323,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.06e-05,
+      "loss": 1.9875,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.08e-05,
+      "loss": 1.872,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1e-05,
+      "loss": 1.7737,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.12e-05,
+      "loss": 2.1316,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1400000000000002e-05,
+      "loss": 1.6055,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.16e-05,
+      "loss": 1.5318,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.18e-05,
+      "loss": 1.5154,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 1.6614,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.22e-05,
+      "loss": 1.8519,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 1.8016,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.26e-05,
+      "loss": 1.6742,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 1.767,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 1.9138,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.32e-05,
+      "loss": 1.8979,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 1.9283,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.36e-05,
+      "loss": 1.3711,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.38e-05,
+      "loss": 1.5904,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-05,
+      "loss": 1.7512,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4200000000000002e-05,
+      "loss": 2.2549,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.44e-05,
+      "loss": 1.9787,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.46e-05,
+      "loss": 1.9389,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.48e-05,
+      "loss": 1.7391,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-05,
+      "loss": 1.8997,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5200000000000003e-05,
+      "loss": 1.3632,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.54e-05,
+      "loss": 1.7334,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 1.6878,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.58e-05,
+      "loss": 1.274,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 1.5435,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6200000000000003e-05,
+      "loss": 1.5112,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.64e-05,
+      "loss": 1.7326,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6600000000000003e-05,
+      "loss": 1.2891,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6800000000000004e-05,
+      "loss": 1.7933,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 1.7482,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 1.5952,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7400000000000002e-05,
+      "loss": 1.4617,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 1.8901,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 1.8179,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 2.0113,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 1.2497,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.84e-05,
+      "loss": 1.6295,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.86e-05,
+      "loss": 1.3429,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.88e-05,
+      "loss": 1.7611,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9e-05,
+      "loss": 1.5193,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9199999999999998e-05,
+      "loss": 1.6044,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.94e-05,
+      "loss": 1.4793,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.96e-05,
+      "loss": 1.764,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.98e-05,
+      "loss": 1.5408,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 1.6566,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.02e-05,
+      "loss": 1.6846,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.04e-05,
+      "loss": 1.6184,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.06e-05,
+      "loss": 1.4184,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.08e-05,
+      "loss": 1.5008,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1e-05,
+      "loss": 1.3783,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.12e-05,
+      "loss": 1.7596,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1400000000000004e-05,
+      "loss": 2.1059,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.16e-05,
+      "loss": 1.5143,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.18e-05,
+      "loss": 1.4822,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.7857,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2200000000000003e-05,
+      "loss": 1.7087,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.24e-05,
+      "loss": 1.4287,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.26e-05,
+      "loss": 1.679,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2800000000000004e-05,
+      "loss": 1.4188,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3e-05,
+      "loss": 1.7147,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.32e-05,
+      "loss": 1.6419,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3400000000000005e-05,
+      "loss": 1.6317,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 1.6363,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.38e-05,
+      "loss": 1.6637,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 1.7056,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4200000000000005e-05,
+      "loss": 1.6169,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 1.5603,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.46e-05,
+      "loss": 1.7586,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.48e-05,
+      "loss": 1.512,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5e-05,
+      "loss": 1.6072,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.52e-05,
+      "loss": 1.6307,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.54e-05,
+      "loss": 2.0672,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.56e-05,
+      "loss": 1.707,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.58e-05,
+      "loss": 1.5668,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-05,
+      "loss": 1.4658,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.62e-05,
+      "loss": 1.4409,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6400000000000004e-05,
+      "loss": 1.4298,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.66e-05,
+      "loss": 1.4506,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.68e-05,
+      "loss": 1.5356,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7e-05,
+      "loss": 1.0953,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.72e-05,
+      "loss": 1.561,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.74e-05,
+      "loss": 1.5869,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.76e-05,
+      "loss": 1.536,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 1.7751,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8e-05,
+      "loss": 1.4346,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.82e-05,
+      "loss": 1.4953,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 1.7136,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.86e-05,
+      "loss": 1.7467,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.88e-05,
+      "loss": 1.5167,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 1.3456,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9200000000000004e-05,
+      "loss": 1.7196,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.94e-05,
+      "loss": 1.634,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.960000000000001e-05,
+      "loss": 1.2937,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 1.4624,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4e-05,
+      "loss": 1.5253,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.02e-05,
+      "loss": 1.3867,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0400000000000006e-05,
+      "loss": 1.6412,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0600000000000004e-05,
+      "loss": 1.3812,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.08e-05,
+      "loss": 1.8283,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1e-05,
+      "loss": 1.4616,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.12e-05,
+      "loss": 1.4339,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.14e-05,
+      "loss": 1.6135,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.16e-05,
+      "loss": 2.0432,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.18e-05,
+      "loss": 1.4172,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2e-05,
+      "loss": 1.5665,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.22e-05,
+      "loss": 1.3772,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.24e-05,
+      "loss": 1.5019,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.26e-05,
+      "loss": 1.5439,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2800000000000004e-05,
+      "loss": 1.4076,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3e-05,
+      "loss": 1.283,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.32e-05,
+      "loss": 1.2356,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3400000000000005e-05,
+      "loss": 1.5674,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.36e-05,
+      "loss": 1.502,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.38e-05,
+      "loss": 1.5364,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 1.9097,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4200000000000004e-05,
+      "loss": 1.4706,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.44e-05,
+      "loss": 1.5675,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.46e-05,
+      "loss": 1.6833,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4800000000000005e-05,
+      "loss": 1.7794,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5e-05,
+      "loss": 1.834,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.52e-05,
+      "loss": 1.2674,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5400000000000006e-05,
+      "loss": 1.6993,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 1.4372,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.58e-05,
+      "loss": 1.2937,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 1.9018,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6200000000000005e-05,
+      "loss": 1.7445,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.64e-05,
+      "loss": 1.7906,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.660000000000001e-05,
+      "loss": 1.285,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6800000000000006e-05,
+      "loss": 1.441,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7e-05,
+      "loss": 1.4632,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.72e-05,
+      "loss": 1.4486,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.74e-05,
+      "loss": 1.5735,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.76e-05,
+      "loss": 1.6958,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.78e-05,
+      "loss": 1.6234,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8e-05,
+      "loss": 1.5115,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.82e-05,
+      "loss": 1.2801,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8400000000000004e-05,
+      "loss": 1.6176,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.86e-05,
+      "loss": 1.5616,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.88e-05,
+      "loss": 1.7993,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9e-05,
+      "loss": 1.6863,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.92e-05,
+      "loss": 1.1479,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.94e-05,
+      "loss": 1.6661,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.96e-05,
+      "loss": 1.3231,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 1.6112,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-05,
+      "loss": 1.4472,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.02e-05,
+      "loss": 1.3909,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.0400000000000005e-05,
+      "loss": 1.7184,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.0600000000000003e-05,
+      "loss": 1.441,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.08e-05,
+      "loss": 1.4289,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1000000000000006e-05,
+      "loss": 1.335,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1200000000000004e-05,
+      "loss": 1.4251,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.14e-05,
+      "loss": 1.4391,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.16e-05,
+      "loss": 1.1559,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1800000000000005e-05,
+      "loss": 1.3411,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 1.618,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.22e-05,
+      "loss": 1.5593,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2400000000000007e-05,
+      "loss": 1.633,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2600000000000005e-05,
+      "loss": 1.5671,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.28e-05,
+      "loss": 1.4871,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.300000000000001e-05,
+      "loss": 1.3152,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3200000000000006e-05,
+      "loss": 1.6148,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3400000000000004e-05,
+      "loss": 2.0055,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.360000000000001e-05,
+      "loss": 1.093,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.380000000000001e-05,
+      "loss": 1.4497,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 1.0888,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.420000000000001e-05,
+      "loss": 1.8853,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.440000000000001e-05,
+      "loss": 1.5032,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4600000000000006e-05,
+      "loss": 1.7524,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4800000000000004e-05,
+      "loss": 1.5074,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 1.2902,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.520000000000001e-05,
+      "loss": 1.4917,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5400000000000005e-05,
+      "loss": 1.589,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.560000000000001e-05,
+      "loss": 1.6595,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.580000000000001e-05,
+      "loss": 1.4476,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 1.6573,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.620000000000001e-05,
+      "loss": 1.5567,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 1.2408,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.66e-05,
+      "loss": 1.4575,
+      "step": 283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.68e-05,
+      "loss": 1.6787,
+      "step": 284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 1.0178,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.72e-05,
+      "loss": 1.8453,
+      "step": 286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.74e-05,
+      "loss": 1.5197,
+      "step": 287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.76e-05,
+      "loss": 1.3573,
+      "step": 288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7799999999999995e-05,
+      "loss": 1.1354,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8e-05,
+      "loss": 1.5171,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.82e-05,
+      "loss": 1.2755,
+      "step": 291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8399999999999997e-05,
+      "loss": 1.4802,
+      "step": 292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.86e-05,
+      "loss": 1.4177,
+      "step": 293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.88e-05,
+      "loss": 2.0432,
+      "step": 294
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9e-05,
+      "loss": 1.6314,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.92e-05,
+      "loss": 1.9263,
+      "step": 296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.94e-05,
+      "loss": 1.6432,
+      "step": 297
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.96e-05,
+      "loss": 1.6659,
+      "step": 298
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9800000000000003e-05,
+      "loss": 1.2489,
+      "step": 299
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6e-05,
+      "loss": 1.2108,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.02e-05,
+      "loss": 1.6832,
+      "step": 301
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.04e-05,
+      "loss": 1.652,
+      "step": 302
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.06e-05,
+      "loss": 1.5827,
+      "step": 303
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.08e-05,
+      "loss": 1.4894,
+      "step": 304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.1e-05,
+      "loss": 1.4859,
+      "step": 305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.12e-05,
+      "loss": 1.8552,
+      "step": 306
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.14e-05,
+      "loss": 1.5457,
+      "step": 307
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.16e-05,
+      "loss": 1.1685,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.18e-05,
+      "loss": 1.3348,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.2e-05,
+      "loss": 1.4138,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.220000000000001e-05,
+      "loss": 1.3308,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.24e-05,
+      "loss": 1.3738,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.26e-05,
+      "loss": 1.964,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.280000000000001e-05,
+      "loss": 1.4511,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.3e-05,
+      "loss": 1.7097,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.32e-05,
+      "loss": 1.5004,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.340000000000001e-05,
+      "loss": 1.6308,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.36e-05,
+      "loss": 1.4926,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.38e-05,
+      "loss": 1.6042,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 1.6722,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.42e-05,
+      "loss": 1.3271,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.440000000000001e-05,
+      "loss": 1.305,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.460000000000001e-05,
+      "loss": 1.2941,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.48e-05,
+      "loss": 1.21,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 1.6485,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.52e-05,
+      "loss": 1.385,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.54e-05,
+      "loss": 1.9986,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.560000000000001e-05,
+      "loss": 1.268,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.58e-05,
+      "loss": 1.8174,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.6e-05,
+      "loss": 1.182,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.620000000000001e-05,
+      "loss": 2.1805,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.64e-05,
+      "loss": 1.2734,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.66e-05,
+      "loss": 1.4622,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.680000000000001e-05,
+      "loss": 1.2889,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.7e-05,
+      "loss": 1.2408,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.720000000000001e-05,
+      "loss": 1.5586,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.740000000000001e-05,
+      "loss": 1.5067,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.76e-05,
+      "loss": 1.6171,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.780000000000001e-05,
+      "loss": 1.2487,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 1.018,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.82e-05,
+      "loss": 1.1409,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.840000000000001e-05,
+      "loss": 1.1521,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.860000000000001e-05,
+      "loss": 0.9494,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.879999999999999e-05,
+      "loss": 1.2111,
+      "step": 344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.9e-05,
+      "loss": 1.2757,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.92e-05,
+      "loss": 0.9577,
+      "step": 346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.939999999999999e-05,
+      "loss": 1.677,
+      "step": 347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.96e-05,
+      "loss": 1.7494,
+      "step": 348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.98e-05,
+      "loss": 1.356,
+      "step": 349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7e-05,
+      "loss": 1.3482,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.02e-05,
+      "loss": 1.395,
+      "step": 351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.04e-05,
+      "loss": 1.1863,
+      "step": 352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.06e-05,
+      "loss": 1.7247,
+      "step": 353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.08e-05,
+      "loss": 1.518,
+      "step": 354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.1e-05,
+      "loss": 1.4258,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.12e-05,
+      "loss": 1.3597,
+      "step": 356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.14e-05,
+      "loss": 1.4732,
+      "step": 357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.16e-05,
+      "loss": 1.3704,
+      "step": 358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.18e-05,
+      "loss": 1.803,
+      "step": 359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.2e-05,
+      "loss": 1.5873,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.22e-05,
+      "loss": 1.6076,
+      "step": 361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.24e-05,
+      "loss": 0.858,
+      "step": 362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.26e-05,
+      "loss": 1.5569,
+      "step": 363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.280000000000001e-05,
+      "loss": 1.9969,
+      "step": 364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.3e-05,
+      "loss": 1.7282,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.32e-05,
+      "loss": 1.7829,
+      "step": 366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.340000000000001e-05,
+      "loss": 1.6164,
+      "step": 367
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.36e-05,
+      "loss": 1.4056,
+      "step": 368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.38e-05,
+      "loss": 2.0958,
+      "step": 369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.4e-05,
+      "loss": 1.6465,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.42e-05,
+      "loss": 1.7758,
+      "step": 371
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.44e-05,
+      "loss": 1.2432,
+      "step": 372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.46e-05,
+      "loss": 1.6467,
+      "step": 373
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.48e-05,
+      "loss": 1.3182,
+      "step": 374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 1.6811,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.52e-05,
+      "loss": 1.5549,
+      "step": 376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.54e-05,
+      "loss": 2.3156,
+      "step": 377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.560000000000001e-05,
+      "loss": 1.4064,
+      "step": 378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.58e-05,
+      "loss": 1.5309,
+      "step": 379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.6e-05,
+      "loss": 1.2805,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.620000000000001e-05,
+      "loss": 1.4004,
+      "step": 381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.64e-05,
+      "loss": 1.553,
+      "step": 382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.66e-05,
+      "loss": 1.7702,
+      "step": 383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.680000000000001e-05,
+      "loss": 1.7585,
+      "step": 384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.7e-05,
+      "loss": 1.4616,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.72e-05,
+      "loss": 1.2726,
+      "step": 386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.740000000000001e-05,
+      "loss": 1.5926,
+      "step": 387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.76e-05,
+      "loss": 0.7802,
+      "step": 388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.780000000000001e-05,
+      "loss": 1.0099,
+      "step": 389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 1.4434,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.82e-05,
+      "loss": 1.7696,
+      "step": 391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.840000000000001e-05,
+      "loss": 1.5141,
+      "step": 392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.860000000000001e-05,
+      "loss": 1.6265,
+      "step": 393
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.88e-05,
+      "loss": 1.3492,
+      "step": 394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.900000000000001e-05,
+      "loss": 1.5901,
+      "step": 395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.920000000000001e-05,
+      "loss": 1.6346,
+      "step": 396
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.94e-05,
+      "loss": 1.4424,
+      "step": 397
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.960000000000001e-05,
+      "loss": 1.2514,
+      "step": 398
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.98e-05,
+      "loss": 1.3189,
+      "step": 399
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8e-05,
+      "loss": 1.3855,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.020000000000001e-05,
+      "loss": 1.4823,
+      "step": 401
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.04e-05,
+      "loss": 1.5709,
+      "step": 402
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.060000000000001e-05,
+      "loss": 1.2782,
+      "step": 403
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.080000000000001e-05,
+      "loss": 1.398,
+      "step": 404
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.1e-05,
+      "loss": 1.0744,
+      "step": 405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.120000000000001e-05,
+      "loss": 1.5305,
+      "step": 406
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.14e-05,
+      "loss": 1.6311,
+      "step": 407
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.16e-05,
+      "loss": 1.3434,
+      "step": 408
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.18e-05,
+      "loss": 1.5952,
+      "step": 409
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.2e-05,
+      "loss": 1.5664,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.22e-05,
+      "loss": 1.4026,
+      "step": 411
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.24e-05,
+      "loss": 1.544,
+      "step": 412
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.26e-05,
+      "loss": 1.5422,
+      "step": 413
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.28e-05,
+      "loss": 1.3202,
+      "step": 414
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.3e-05,
+      "loss": 1.4598,
+      "step": 415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.32e-05,
+      "loss": 1.3091,
+      "step": 416
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.34e-05,
+      "loss": 1.704,
+      "step": 417
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.36e-05,
+      "loss": 1.5211,
+      "step": 418
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.38e-05,
+      "loss": 1.6799,
+      "step": 419
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.4e-05,
+      "loss": 1.7912,
+      "step": 420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.42e-05,
+      "loss": 1.2058,
+      "step": 421
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.44e-05,
+      "loss": 1.5586,
+      "step": 422
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.46e-05,
+      "loss": 1.9113,
+      "step": 423
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.48e-05,
+      "loss": 1.4113,
+      "step": 424
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.5e-05,
+      "loss": 1.5128,
+      "step": 425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.52e-05,
+      "loss": 1.4342,
+      "step": 426
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.54e-05,
+      "loss": 1.7574,
+      "step": 427
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.560000000000001e-05,
+      "loss": 1.5988,
+      "step": 428
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.58e-05,
+      "loss": 1.3769,
+      "step": 429
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.6e-05,
+      "loss": 1.0799,
+      "step": 430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.620000000000001e-05,
+      "loss": 1.4777,
+      "step": 431
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.64e-05,
+      "loss": 1.1064,
+      "step": 432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.66e-05,
+      "loss": 1.4138,
+      "step": 433
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.680000000000001e-05,
+      "loss": 1.447,
+      "step": 434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.7e-05,
+      "loss": 1.6312,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.72e-05,
+      "loss": 2.0039,
+      "step": 436
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.740000000000001e-05,
+      "loss": 1.5865,
+      "step": 437
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.76e-05,
+      "loss": 1.8403,
+      "step": 438
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.78e-05,
+      "loss": 1.5404,
+      "step": 439
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 1.2334,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.82e-05,
+      "loss": 0.8699,
+      "step": 441
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.840000000000001e-05,
+      "loss": 1.1705,
+      "step": 442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.86e-05,
+      "loss": 1.4708,
+      "step": 443
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.88e-05,
+      "loss": 1.6027,
+      "step": 444
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.900000000000001e-05,
+      "loss": 1.4132,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.92e-05,
+      "loss": 1.1757,
+      "step": 446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.94e-05,
+      "loss": 1.508,
+      "step": 447
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.960000000000001e-05,
+      "loss": 1.5797,
+      "step": 448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.98e-05,
+      "loss": 1.8008,
+      "step": 449
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9e-05,
+      "loss": 1.3872,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.020000000000001e-05,
+      "loss": 1.5995,
+      "step": 451
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.04e-05,
+      "loss": 1.4805,
+      "step": 452
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.06e-05,
+      "loss": 1.365,
+      "step": 453
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.080000000000001e-05,
+      "loss": 1.1289,
+      "step": 454
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.1e-05,
+      "loss": 1.4652,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.120000000000001e-05,
+      "loss": 1.5237,
+      "step": 456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.140000000000001e-05,
+      "loss": 1.7387,
+      "step": 457
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.16e-05,
+      "loss": 1.3458,
+      "step": 458
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.180000000000001e-05,
+      "loss": 1.6596,
+      "step": 459
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 1.1005,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.22e-05,
+      "loss": 1.2414,
+      "step": 461
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.240000000000001e-05,
+      "loss": 1.1917,
+      "step": 462
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.260000000000001e-05,
+      "loss": 1.4902,
+      "step": 463
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.28e-05,
+      "loss": 1.2343,
+      "step": 464
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.300000000000001e-05,
+      "loss": 1.4145,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.320000000000002e-05,
+      "loss": 1.5181,
+      "step": 466
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.340000000000001e-05,
+      "loss": 1.5795,
+      "step": 467
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.360000000000001e-05,
+      "loss": 1.6768,
+      "step": 468
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.38e-05,
+      "loss": 1.4173,
+      "step": 469
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.4e-05,
+      "loss": 1.7437,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.42e-05,
+      "loss": 1.5802,
+      "step": 471
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.44e-05,
+      "loss": 1.6308,
+      "step": 472
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.46e-05,
+      "loss": 1.4224,
+      "step": 473
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.48e-05,
+      "loss": 1.6391,
+      "step": 474
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.5e-05,
+      "loss": 1.303,
+      "step": 475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.52e-05,
+      "loss": 1.723,
+      "step": 476
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.54e-05,
+      "loss": 1.1372,
+      "step": 477
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.56e-05,
+      "loss": 1.5179,
+      "step": 478
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.58e-05,
+      "loss": 1.549,
+      "step": 479
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.6e-05,
+      "loss": 1.8439,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.620000000000001e-05,
+      "loss": 1.571,
+      "step": 481
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.64e-05,
+      "loss": 2.0604,
+      "step": 482
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.66e-05,
+      "loss": 1.2386,
+      "step": 483
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.680000000000001e-05,
+      "loss": 1.5211,
+      "step": 484
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.7e-05,
+      "loss": 1.6248,
+      "step": 485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.72e-05,
+      "loss": 1.6244,
+      "step": 486
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.74e-05,
+      "loss": 1.6124,
+      "step": 487
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.76e-05,
+      "loss": 1.5272,
+      "step": 488
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.78e-05,
+      "loss": 1.5976,
+      "step": 489
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.8e-05,
+      "loss": 1.267,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.82e-05,
+      "loss": 1.4788,
+      "step": 491
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.84e-05,
+      "loss": 0.823,
+      "step": 492
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.86e-05,
+      "loss": 1.2987,
+      "step": 493
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.88e-05,
+      "loss": 1.3437,
+      "step": 494
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.900000000000001e-05,
+      "loss": 1.7767,
+      "step": 495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.92e-05,
+      "loss": 1.2853,
+      "step": 496
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.94e-05,
+      "loss": 1.4452,
+      "step": 497
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960000000000001e-05,
+      "loss": 1.2413,
+      "step": 498
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.98e-05,
+      "loss": 1.4011,
+      "step": 499
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001,
+      "loss": 1.3972,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010020000000000001,
+      "loss": 1.546,
+      "step": 501
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001004,
+      "loss": 1.2974,
+      "step": 502
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001006,
+      "loss": 1.4225,
+      "step": 503
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010080000000000001,
+      "loss": 1.4523,
+      "step": 504
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000101,
+      "loss": 1.178,
+      "step": 505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010120000000000001,
+      "loss": 1.5364,
+      "step": 506
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010140000000000001,
+      "loss": 1.4528,
+      "step": 507
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001016,
+      "loss": 1.8245,
+      "step": 508
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010180000000000001,
+      "loss": 1.2151,
+      "step": 509
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010200000000000001,
+      "loss": 1.6755,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001022,
+      "loss": 1.5159,
+      "step": 511
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010240000000000001,
+      "loss": 1.2306,
+      "step": 512
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010260000000000001,
+      "loss": 1.241,
+      "step": 513
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001028,
+      "loss": 1.4732,
+      "step": 514
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010300000000000001,
+      "loss": 1.6475,
+      "step": 515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001032,
+      "loss": 1.3318,
+      "step": 516
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001034,
+      "loss": 1.4072,
+      "step": 517
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010360000000000001,
+      "loss": 1.543,
+      "step": 518
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001038,
+      "loss": 1.3554,
+      "step": 519
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010400000000000001,
+      "loss": 1.5854,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010420000000000001,
+      "loss": 1.5154,
+      "step": 521
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001044,
+      "loss": 1.6702,
+      "step": 522
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010460000000000001,
+      "loss": 1.4039,
+      "step": 523
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010480000000000001,
+      "loss": 1.4432,
+      "step": 524
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000105,
+      "loss": 1.3841,
+      "step": 525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010520000000000001,
+      "loss": 1.3568,
+      "step": 526
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010540000000000001,
+      "loss": 1.1084,
+      "step": 527
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001056,
+      "loss": 1.5071,
+      "step": 528
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010580000000000001,
+      "loss": 1.1379,
+      "step": 529
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010600000000000002,
+      "loss": 1.7507,
+      "step": 530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001062,
+      "loss": 1.558,
+      "step": 531
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010640000000000001,
+      "loss": 1.7871,
+      "step": 532
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010660000000000002,
+      "loss": 1.5901,
+      "step": 533
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010680000000000001,
+      "loss": 1.3884,
+      "step": 534
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010700000000000001,
+      "loss": 1.4066,
+      "step": 535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010720000000000002,
+      "loss": 1.333,
+      "step": 536
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010740000000000001,
+      "loss": 1.1743,
+      "step": 537
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010760000000000001,
+      "loss": 1.2642,
+      "step": 538
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010780000000000002,
+      "loss": 1.728,
+      "step": 539
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010800000000000001,
+      "loss": 1.4403,
+      "step": 540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010820000000000001,
+      "loss": 1.3877,
+      "step": 541
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010840000000000002,
+      "loss": 1.7758,
+      "step": 542
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010860000000000001,
+      "loss": 1.2633,
+      "step": 543
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010880000000000002,
+      "loss": 1.3916,
+      "step": 544
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000109,
+      "loss": 1.4934,
+      "step": 545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010920000000000001,
+      "loss": 1.1001,
+      "step": 546
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010940000000000002,
+      "loss": 2.0889,
+      "step": 547
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010960000000000001,
+      "loss": 1.5629,
+      "step": 548
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010980000000000001,
+      "loss": 1.4997,
+      "step": 549
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011000000000000002,
+      "loss": 1.3677,
+      "step": 550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011020000000000001,
+      "loss": 1.4175,
+      "step": 551
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011040000000000001,
+      "loss": 1.5248,
+      "step": 552
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011060000000000002,
+      "loss": 1.3296,
+      "step": 553
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011080000000000001,
+      "loss": 1.3801,
+      "step": 554
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011100000000000001,
+      "loss": 1.1336,
+      "step": 555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011120000000000002,
+      "loss": 1.5654,
+      "step": 556
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011140000000000001,
+      "loss": 1.6585,
+      "step": 557
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011160000000000002,
+      "loss": 0.9196,
+      "step": 558
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011180000000000002,
+      "loss": 1.6131,
+      "step": 559
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011200000000000001,
+      "loss": 1.5301,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011220000000000002,
+      "loss": 1.2874,
+      "step": 561
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011240000000000002,
+      "loss": 1.7624,
+      "step": 562
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001126,
+      "loss": 1.6631,
+      "step": 563
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011279999999999999,
+      "loss": 1.5735,
+      "step": 564
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000113,
+      "loss": 1.5961,
+      "step": 565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001132,
+      "loss": 1.4755,
+      "step": 566
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011339999999999999,
+      "loss": 1.706,
+      "step": 567
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001136,
+      "loss": 1.5176,
+      "step": 568
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001138,
+      "loss": 1.5672,
+      "step": 569
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 1.4458,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001142,
+      "loss": 1.6121,
+      "step": 571
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001144,
+      "loss": 1.1258,
+      "step": 572
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001146,
+      "loss": 1.1178,
+      "step": 573
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001148,
+      "loss": 1.8393,
+      "step": 574
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011499999999999999,
+      "loss": 1.023,
+      "step": 575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001152,
+      "loss": 1.4485,
+      "step": 576
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001154,
+      "loss": 1.4775,
+      "step": 577
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011559999999999999,
+      "loss": 0.8042,
+      "step": 578
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001158,
+      "loss": 1.714,
+      "step": 579
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000116,
+      "loss": 1.7546,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011619999999999999,
+      "loss": 1.0704,
+      "step": 581
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001164,
+      "loss": 1.648,
+      "step": 582
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001166,
+      "loss": 1.1592,
+      "step": 583
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011679999999999999,
+      "loss": 1.3606,
+      "step": 584
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000117,
+      "loss": 1.7044,
+      "step": 585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001172,
+      "loss": 1.1941,
+      "step": 586
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001174,
+      "loss": 1.4298,
+      "step": 587
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001176,
+      "loss": 1.3192,
+      "step": 588
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001178,
+      "loss": 1.0779,
+      "step": 589
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000118,
+      "loss": 0.9457,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001182,
+      "loss": 1.2931,
+      "step": 591
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001184,
+      "loss": 1.462,
+      "step": 592
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001186,
+      "loss": 1.4907,
+      "step": 593
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001188,
+      "loss": 1.7805,
+      "step": 594
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000119,
+      "loss": 1.4701,
+      "step": 595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001192,
+      "loss": 0.9648,
+      "step": 596
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001194,
+      "loss": 1.2587,
+      "step": 597
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011960000000000001,
+      "loss": 1.8042,
+      "step": 598
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001198,
+      "loss": 1.5476,
+      "step": 599
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012,
+      "loss": 1.6051,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012020000000000001,
+      "loss": 1.2883,
+      "step": 601
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001204,
+      "loss": 1.6611,
+      "step": 602
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001206,
+      "loss": 1.2966,
+      "step": 603
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001208,
+      "loss": 1.6653,
+      "step": 604
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000121,
+      "loss": 1.1712,
+      "step": 605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001212,
+      "loss": 1.267,
+      "step": 606
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001214,
+      "loss": 1.2151,
+      "step": 607
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001216,
+      "loss": 1.3627,
+      "step": 608
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001218,
+      "loss": 1.2695,
+      "step": 609
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000122,
+      "loss": 1.6126,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012220000000000002,
+      "loss": 1.301,
+      "step": 611
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001224,
+      "loss": 1.8187,
+      "step": 612
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001226,
+      "loss": 1.5568,
+      "step": 613
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001228,
+      "loss": 1.2592,
+      "step": 614
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000123,
+      "loss": 1.3312,
+      "step": 615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001232,
+      "loss": 1.6488,
+      "step": 616
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012340000000000002,
+      "loss": 1.4315,
+      "step": 617
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001236,
+      "loss": 1.5581,
+      "step": 618
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001238,
+      "loss": 1.4216,
+      "step": 619
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000124,
+      "loss": 1.6793,
+      "step": 620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001242,
+      "loss": 1.1311,
+      "step": 621
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012440000000000002,
+      "loss": 1.256,
+      "step": 622
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001246,
+      "loss": 1.1003,
+      "step": 623
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001248,
+      "loss": 1.1129,
+      "step": 624
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000125,
+      "loss": 1.4271,
+      "step": 625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001252,
+      "loss": 0.7565,
+      "step": 626
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001254,
+      "loss": 1.7288,
+      "step": 627
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012560000000000002,
+      "loss": 1.1002,
+      "step": 628
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001258,
+      "loss": 1.3936,
+      "step": 629
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000126,
+      "loss": 1.0093,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001262,
+      "loss": 1.5043,
+      "step": 631
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001264,
+      "loss": 1.8093,
+      "step": 632
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012660000000000001,
+      "loss": 1.4558,
+      "step": 633
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012680000000000002,
+      "loss": 1.4166,
+      "step": 634
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000127,
+      "loss": 1.3696,
+      "step": 635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001272,
+      "loss": 1.9203,
+      "step": 636
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001274,
+      "loss": 1.8572,
+      "step": 637
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001276,
+      "loss": 1.4041,
+      "step": 638
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012780000000000002,
+      "loss": 1.353,
+      "step": 639
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012800000000000002,
+      "loss": 1.6167,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001282,
+      "loss": 1.5847,
+      "step": 641
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001284,
+      "loss": 1.5189,
+      "step": 642
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001286,
+      "loss": 1.7151,
+      "step": 643
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012880000000000001,
+      "loss": 1.451,
+      "step": 644
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012900000000000002,
+      "loss": 1.5162,
+      "step": 645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012920000000000002,
+      "loss": 1.2337,
+      "step": 646
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001294,
+      "loss": 1.1972,
+      "step": 647
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001296,
+      "loss": 1.5025,
+      "step": 648
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001298,
+      "loss": 1.3985,
+      "step": 649
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 1.3171,
+      "step": 650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013020000000000002,
+      "loss": 1.4803,
+      "step": 651
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001304,
+      "loss": 1.529,
+      "step": 652
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001306,
+      "loss": 1.4434,
+      "step": 653
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001308,
+      "loss": 1.0806,
+      "step": 654
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000131,
+      "loss": 1.1537,
+      "step": 655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013120000000000002,
+      "loss": 1.3307,
+      "step": 656
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013140000000000002,
+      "loss": 1.6359,
+      "step": 657
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001316,
+      "loss": 1.3696,
+      "step": 658
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001318,
+      "loss": 1.4737,
+      "step": 659
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000132,
+      "loss": 1.9183,
+      "step": 660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013220000000000001,
+      "loss": 1.7965,
+      "step": 661
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013240000000000002,
+      "loss": 1.3467,
+      "step": 662
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013260000000000002,
+      "loss": 1.7211,
+      "step": 663
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001328,
+      "loss": 1.7873,
+      "step": 664
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000133,
+      "loss": 1.5823,
+      "step": 665
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001332,
+      "loss": 1.4827,
+      "step": 666
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013340000000000002,
+      "loss": 1.8043,
+      "step": 667
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013360000000000002,
+      "loss": 1.442,
+      "step": 668
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013380000000000003,
+      "loss": 1.3361,
+      "step": 669
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000134,
+      "loss": 1.4699,
+      "step": 670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001342,
+      "loss": 1.8613,
+      "step": 671
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013440000000000001,
+      "loss": 1.7948,
+      "step": 672
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013460000000000002,
+      "loss": 1.6537,
+      "step": 673
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013480000000000002,
+      "loss": 1.3819,
+      "step": 674
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013500000000000003,
+      "loss": 1.4638,
+      "step": 675
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001352,
+      "loss": 1.7048,
+      "step": 676
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001354,
+      "loss": 1.4222,
+      "step": 677
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013560000000000002,
+      "loss": 1.6548,
+      "step": 678
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00013580000000000002,
+      "loss": 1.424,
+      "step": 679
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00013600000000000003,
+      "loss": 1.7807,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001362,
+      "loss": 1.7791,
+      "step": 681
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001364,
+      "loss": 1.2475,
+      "step": 682
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001366,
+      "loss": 1.5399,
+      "step": 683
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00013680000000000002,
+      "loss": 1.8002,
+      "step": 684
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00013700000000000002,
+      "loss": 1.3886,
+      "step": 685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00013720000000000003,
+      "loss": 1.2532,
+      "step": 686
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001374,
+      "loss": 1.377,
+      "step": 687
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00013759999999999998,
+      "loss": 1.5554,
+      "step": 688
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001378,
+      "loss": 1.2965,
+      "step": 689
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000138,
+      "loss": 1.5544,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001382,
+      "loss": 1.6136,
+      "step": 691
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001384,
+      "loss": 1.9085,
+      "step": 692
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001386,
+      "loss": 1.2809,
+      "step": 693
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00013879999999999999,
+      "loss": 1.5892,
+      "step": 694
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000139,
+      "loss": 1.9711,
+      "step": 695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001392,
+      "loss": 2.023,
+      "step": 696
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001394,
+      "loss": 0.9195,
+      "step": 697
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001396,
+      "loss": 1.5298,
+      "step": 698
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001398,
+      "loss": 1.6246,
+      "step": 699
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014,
+      "loss": 1.3674,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001402,
+      "loss": 1.2625,
+      "step": 701
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001404,
+      "loss": 1.1216,
+      "step": 702
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001406,
+      "loss": 1.384,
+      "step": 703
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001408,
+      "loss": 1.297,
+      "step": 704
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000141,
+      "loss": 1.3197,
+      "step": 705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001412,
+      "loss": 1.5897,
+      "step": 706
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001414,
+      "loss": 0.9781,
+      "step": 707
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001416,
+      "loss": 1.6137,
+      "step": 708
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001418,
+      "loss": 1.0798,
+      "step": 709
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000142,
+      "loss": 1.4615,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001422,
+      "loss": 1.4394,
+      "step": 711
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001424,
+      "loss": 1.4126,
+      "step": 712
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001426,
+      "loss": 1.419,
+      "step": 713
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001428,
+      "loss": 1.9973,
+      "step": 714
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000143,
+      "loss": 1.3262,
+      "step": 715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001432,
+      "loss": 1.4252,
+      "step": 716
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001434,
+      "loss": 1.468,
+      "step": 717
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001436,
+      "loss": 1.2148,
+      "step": 718
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001438,
+      "loss": 1.6929,
+      "step": 719
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000144,
+      "loss": 1.395,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001442,
+      "loss": 1.6186,
+      "step": 721
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001444,
+      "loss": 1.3492,
+      "step": 722
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001446,
+      "loss": 1.2352,
+      "step": 723
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001448,
+      "loss": 1.358,
+      "step": 724
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000145,
+      "loss": 1.7877,
+      "step": 725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001452,
+      "loss": 1.3352,
+      "step": 726
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001454,
+      "loss": 1.5035,
+      "step": 727
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014560000000000002,
+      "loss": 1.1398,
+      "step": 728
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001458,
+      "loss": 1.4126,
+      "step": 729
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000146,
+      "loss": 1.8767,
+      "step": 730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001462,
+      "loss": 1.6502,
+      "step": 731
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001464,
+      "loss": 1.5629,
+      "step": 732
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001466,
+      "loss": 1.1927,
+      "step": 733
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014680000000000002,
+      "loss": 1.4369,
+      "step": 734
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000147,
+      "loss": 1.6228,
+      "step": 735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001472,
+      "loss": 0.6199,
+      "step": 736
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001474,
+      "loss": 1.4311,
+      "step": 737
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001476,
+      "loss": 1.7825,
+      "step": 738
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014780000000000001,
+      "loss": 1.6303,
+      "step": 739
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000148,
+      "loss": 1.5027,
+      "step": 740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001482,
+      "loss": 1.7188,
+      "step": 741
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001484,
+      "loss": 1.3928,
+      "step": 742
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001486,
+      "loss": 1.1591,
+      "step": 743
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001488,
+      "loss": 1.4022,
+      "step": 744
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014900000000000002,
+      "loss": 1.0907,
+      "step": 745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001492,
+      "loss": 1.4968,
+      "step": 746
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001494,
+      "loss": 1.5342,
+      "step": 747
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001496,
+      "loss": 1.5563,
+      "step": 748
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001498,
+      "loss": 1.2384,
+      "step": 749
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.9154,
+      "step": 750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015020000000000002,
+      "loss": 1.8682,
+      "step": 751
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001504,
+      "loss": 1.4887,
+      "step": 752
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001506,
+      "loss": 1.6015,
+      "step": 753
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001508,
+      "loss": 1.7943,
+      "step": 754
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000151,
+      "loss": 1.1466,
+      "step": 755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015120000000000002,
+      "loss": 1.3393,
+      "step": 756
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015140000000000002,
+      "loss": 1.5637,
+      "step": 757
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001516,
+      "loss": 1.3269,
+      "step": 758
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001518,
+      "loss": 1.44,
+      "step": 759
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000152,
+      "loss": 1.4095,
+      "step": 760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001522,
+      "loss": 0.972,
+      "step": 761
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015240000000000002,
+      "loss": 1.6002,
+      "step": 762
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015260000000000002,
+      "loss": 1.618,
+      "step": 763
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001528,
+      "loss": 1.3481,
+      "step": 764
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000153,
+      "loss": 1.4666,
+      "step": 765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001532,
+      "loss": 1.6578,
+      "step": 766
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015340000000000002,
+      "loss": 1.8441,
+      "step": 767
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015360000000000002,
+      "loss": 1.0506,
+      "step": 768
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001538,
+      "loss": 1.4459,
+      "step": 769
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000154,
+      "loss": 2.0034,
+      "step": 770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001542,
+      "loss": 1.9728,
+      "step": 771
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001544,
+      "loss": 1.3435,
+      "step": 772
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015460000000000002,
+      "loss": 1.2247,
+      "step": 773
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015480000000000002,
+      "loss": 1.5204,
+      "step": 774
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000155,
+      "loss": 1.0024,
+      "step": 775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001552,
+      "loss": 1.6894,
+      "step": 776
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001554,
+      "loss": 1.3371,
+      "step": 777
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015560000000000001,
+      "loss": 1.5917,
+      "step": 778
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015580000000000002,
+      "loss": 1.3432,
+      "step": 779
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015600000000000002,
+      "loss": 1.5958,
+      "step": 780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001562,
+      "loss": 1.5246,
+      "step": 781
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001564,
+      "loss": 1.119,
+      "step": 782
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001566,
+      "loss": 1.4479,
+      "step": 783
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015680000000000002,
+      "loss": 1.7507,
+      "step": 784
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015700000000000002,
+      "loss": 1.078,
+      "step": 785
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015720000000000003,
+      "loss": 1.485,
+      "step": 786
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001574,
+      "loss": 1.1164,
+      "step": 787
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001576,
+      "loss": 1.8051,
+      "step": 788
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015780000000000001,
+      "loss": 1.2503,
+      "step": 789
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015800000000000002,
+      "loss": 1.4041,
+      "step": 790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015820000000000002,
+      "loss": 1.2593,
+      "step": 791
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015840000000000003,
+      "loss": 1.1401,
+      "step": 792
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001586,
+      "loss": 1.2461,
+      "step": 793
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001588,
+      "loss": 1.7059,
+      "step": 794
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015900000000000002,
+      "loss": 1.251,
+      "step": 795
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015920000000000002,
+      "loss": 1.9167,
+      "step": 796
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015940000000000003,
+      "loss": 1.7818,
+      "step": 797
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001596,
+      "loss": 1.4416,
+      "step": 798
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001598,
+      "loss": 1.8941,
+      "step": 799
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016,
+      "loss": 1.1635,
+      "step": 800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016020000000000002,
+      "loss": 1.5663,
+      "step": 801
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016040000000000002,
+      "loss": 1.3,
+      "step": 802
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016060000000000003,
+      "loss": 1.3644,
+      "step": 803
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001608,
+      "loss": 1.4951,
+      "step": 804
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000161,
+      "loss": 1.3978,
+      "step": 805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016120000000000002,
+      "loss": 1.643,
+      "step": 806
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016140000000000002,
+      "loss": 1.5616,
+      "step": 807
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016160000000000002,
+      "loss": 1.7618,
+      "step": 808
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016180000000000003,
+      "loss": 1.7519,
+      "step": 809
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000162,
+      "loss": 1.4218,
+      "step": 810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001622,
+      "loss": 1.8203,
+      "step": 811
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016240000000000002,
+      "loss": 1.9702,
+      "step": 812
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001626,
+      "loss": 1.7396,
+      "step": 813
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001628,
+      "loss": 1.8649,
+      "step": 814
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000163,
+      "loss": 1.8828,
+      "step": 815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001632,
+      "loss": 1.5457,
+      "step": 816
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001634,
+      "loss": 1.2831,
+      "step": 817
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001636,
+      "loss": 1.5317,
+      "step": 818
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001638,
+      "loss": 1.7262,
+      "step": 819
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000164,
+      "loss": 1.492,
+      "step": 820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001642,
+      "loss": 1.8386,
+      "step": 821
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001644,
+      "loss": 1.4252,
+      "step": 822
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001646,
+      "loss": 1.5431,
+      "step": 823
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001648,
+      "loss": 1.6406,
+      "step": 824
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000165,
+      "loss": 1.8041,
+      "step": 825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001652,
+      "loss": 1.9322,
+      "step": 826
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001654,
+      "loss": 0.9531,
+      "step": 827
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001656,
+      "loss": 1.2036,
+      "step": 828
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001658,
+      "loss": 1.6786,
+      "step": 829
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000166,
+      "loss": 1.8316,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001662,
+      "loss": 1.514,
+      "step": 831
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001664,
+      "loss": 1.2779,
+      "step": 832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001666,
+      "loss": 1.5057,
+      "step": 833
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001668,
+      "loss": 1.254,
+      "step": 834
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000167,
+      "loss": 1.5101,
+      "step": 835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001672,
+      "loss": 1.3411,
+      "step": 836
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001674,
+      "loss": 1.7435,
+      "step": 837
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001676,
+      "loss": 1.7571,
+      "step": 838
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001678,
+      "loss": 1.598,
+      "step": 839
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000168,
+      "loss": 1.7759,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001682,
+      "loss": 1.8047,
+      "step": 841
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001684,
+      "loss": 1.3423,
+      "step": 842
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001686,
+      "loss": 1.6953,
+      "step": 843
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001688,
+      "loss": 1.8372,
+      "step": 844
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016900000000000002,
+      "loss": 1.0665,
+      "step": 845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001692,
+      "loss": 1.6637,
+      "step": 846
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001694,
+      "loss": 1.4009,
+      "step": 847
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001696,
+      "loss": 1.3269,
+      "step": 848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001698,
+      "loss": 1.4418,
+      "step": 849
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017,
+      "loss": 1.2336,
+      "step": 850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017020000000000002,
+      "loss": 1.7316,
+      "step": 851
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001704,
+      "loss": 1.442,
+      "step": 852
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001706,
+      "loss": 1.4138,
+      "step": 853
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001708,
+      "loss": 1.6622,
+      "step": 854
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000171,
+      "loss": 1.0241,
+      "step": 855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017120000000000001,
+      "loss": 0.8718,
+      "step": 856
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001714,
+      "loss": 1.642,
+      "step": 857
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001716,
+      "loss": 1.4068,
+      "step": 858
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001718,
+      "loss": 1.5546,
+      "step": 859
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000172,
+      "loss": 1.4875,
+      "step": 860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001722,
+      "loss": 1.7402,
+      "step": 861
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017240000000000002,
+      "loss": 1.5206,
+      "step": 862
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001726,
+      "loss": 2.182,
+      "step": 863
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001728,
+      "loss": 1.154,
+      "step": 864
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000173,
+      "loss": 1.5995,
+      "step": 865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001732,
+      "loss": 1.3551,
+      "step": 866
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001734,
+      "loss": 1.6101,
+      "step": 867
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017360000000000002,
+      "loss": 1.447,
+      "step": 868
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001738,
+      "loss": 0.9601,
+      "step": 869
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000174,
+      "loss": 1.8979,
+      "step": 870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001742,
+      "loss": 1.2361,
+      "step": 871
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001744,
+      "loss": 1.5328,
+      "step": 872
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017460000000000002,
+      "loss": 1.5379,
+      "step": 873
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017480000000000002,
+      "loss": 1.5336,
+      "step": 874
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000175,
+      "loss": 1.5414,
+      "step": 875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001752,
+      "loss": 1.4737,
+      "step": 876
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001754,
+      "loss": 1.2206,
+      "step": 877
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001756,
+      "loss": 0.9643,
+      "step": 878
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017580000000000002,
+      "loss": 1.4217,
+      "step": 879
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017600000000000002,
+      "loss": 1.1601,
+      "step": 880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001762,
+      "loss": 1.3268,
+      "step": 881
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001764,
+      "loss": 1.5891,
+      "step": 882
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001766,
+      "loss": 1.7093,
+      "step": 883
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017680000000000001,
+      "loss": 1.2247,
+      "step": 884
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017700000000000002,
+      "loss": 1.7031,
+      "step": 885
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001772,
+      "loss": 1.5478,
+      "step": 886
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001774,
+      "loss": 1.1641,
+      "step": 887
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001776,
+      "loss": 1.3217,
+      "step": 888
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001778,
+      "loss": 0.4299,
+      "step": 889
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017800000000000002,
+      "loss": 1.4023,
+      "step": 890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017820000000000002,
+      "loss": 1.322,
+      "step": 891
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001784,
+      "loss": 1.9407,
+      "step": 892
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001786,
+      "loss": 1.6132,
+      "step": 893
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001788,
+      "loss": 1.4445,
+      "step": 894
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017900000000000001,
+      "loss": 1.7767,
+      "step": 895
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017920000000000002,
+      "loss": 1.6405,
+      "step": 896
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017940000000000002,
+      "loss": 1.4915,
+      "step": 897
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001796,
+      "loss": 1.8788,
+      "step": 898
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001798,
+      "loss": 1.6118,
+      "step": 899
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018,
+      "loss": 1.2953,
+      "step": 900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018020000000000002,
+      "loss": 1.355,
+      "step": 901
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018040000000000002,
+      "loss": 1.4189,
+      "step": 902
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018060000000000003,
+      "loss": 1.604,
+      "step": 903
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001808,
+      "loss": 1.226,
+      "step": 904
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000181,
+      "loss": 1.4836,
+      "step": 905
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001812,
+      "loss": 0.7573,
+      "step": 906
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018140000000000002,
+      "loss": 1.7299,
+      "step": 907
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018160000000000002,
+      "loss": 1.4491,
+      "step": 908
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018180000000000003,
+      "loss": 1.1456,
+      "step": 909
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000182,
+      "loss": 1.8538,
+      "step": 910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001822,
+      "loss": 1.973,
+      "step": 911
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018240000000000002,
+      "loss": 1.5405,
+      "step": 912
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018260000000000002,
+      "loss": 1.82,
+      "step": 913
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018280000000000003,
+      "loss": 1.8022,
+      "step": 914
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000183,
+      "loss": 1.7486,
+      "step": 915
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001832,
+      "loss": 1.5596,
+      "step": 916
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001834,
+      "loss": 1.2348,
+      "step": 917
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018360000000000002,
+      "loss": 1.7729,
+      "step": 918
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018380000000000002,
+      "loss": 1.4933,
+      "step": 919
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018400000000000003,
+      "loss": 1.4018,
+      "step": 920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001842,
+      "loss": 1.6991,
+      "step": 921
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001844,
+      "loss": 1.6084,
+      "step": 922
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018460000000000001,
+      "loss": 1.8682,
+      "step": 923
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018480000000000002,
+      "loss": 1.6047,
+      "step": 924
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018500000000000002,
+      "loss": 1.8456,
+      "step": 925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018520000000000003,
+      "loss": 1.3492,
+      "step": 926
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001854,
+      "loss": 1.4943,
+      "step": 927
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001856,
+      "loss": 1.419,
+      "step": 928
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018580000000000002,
+      "loss": 1.9021,
+      "step": 929
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018600000000000002,
+      "loss": 1.4237,
+      "step": 930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018620000000000003,
+      "loss": 1.5694,
+      "step": 931
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018640000000000003,
+      "loss": 1.5519,
+      "step": 932
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001866,
+      "loss": 1.5052,
+      "step": 933
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018680000000000001,
+      "loss": 0.9964,
+      "step": 934
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018700000000000002,
+      "loss": 1.4911,
+      "step": 935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018720000000000002,
+      "loss": 1.5602,
+      "step": 936
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018740000000000003,
+      "loss": 1.6938,
+      "step": 937
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001876,
+      "loss": 1.6021,
+      "step": 938
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001878,
+      "loss": 1.7421,
+      "step": 939
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000188,
+      "loss": 1.6676,
+      "step": 940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001882,
+      "loss": 1.518,
+      "step": 941
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001884,
+      "loss": 1.503,
+      "step": 942
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001886,
+      "loss": 1.6147,
+      "step": 943
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001888,
+      "loss": 1.6029,
+      "step": 944
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018899999999999999,
+      "loss": 2.3909,
+      "step": 945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001892,
+      "loss": 1.184,
+      "step": 946
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001894,
+      "loss": 1.4272,
+      "step": 947
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001896,
+      "loss": 1.3491,
+      "step": 948
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001898,
+      "loss": 1.3652,
+      "step": 949
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019,
+      "loss": 1.7427,
+      "step": 950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001902,
+      "loss": 1.5645,
+      "step": 951
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001904,
+      "loss": 1.2984,
+      "step": 952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001906,
+      "loss": 1.8424,
+      "step": 953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001908,
+      "loss": 1.3921,
+      "step": 954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000191,
+      "loss": 1.7446,
+      "step": 955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001912,
+      "loss": 1.4909,
+      "step": 956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001914,
+      "loss": 1.9728,
+      "step": 957
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001916,
+      "loss": 1.3973,
+      "step": 958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001918,
+      "loss": 1.7675,
+      "step": 959
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000192,
+      "loss": 1.3825,
+      "step": 960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001922,
+      "loss": 1.6717,
+      "step": 961
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019240000000000001,
+      "loss": 1.1409,
+      "step": 962
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001926,
+      "loss": 1.8961,
+      "step": 963
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001928,
+      "loss": 1.4622,
+      "step": 964
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000193,
+      "loss": 1.8188,
+      "step": 965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001932,
+      "loss": 1.3865,
+      "step": 966
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001934,
+      "loss": 1.5006,
+      "step": 967
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019360000000000002,
+      "loss": 1.8283,
+      "step": 968
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001938,
+      "loss": 1.5776,
+      "step": 969
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000194,
+      "loss": 1.8864,
+      "step": 970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001942,
+      "loss": 1.4982,
+      "step": 971
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001944,
+      "loss": 1.1805,
+      "step": 972
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019460000000000001,
+      "loss": 1.4492,
+      "step": 973
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001948,
+      "loss": 1.4913,
+      "step": 974
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000195,
+      "loss": 1.3173,
+      "step": 975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001952,
+      "loss": 1.6315,
+      "step": 976
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001954,
+      "loss": 1.1087,
+      "step": 977
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001956,
+      "loss": 1.7119,
+      "step": 978
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019580000000000002,
+      "loss": 1.6867,
+      "step": 979
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000196,
+      "loss": 1.4873,
+      "step": 980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001962,
+      "loss": 1.7583,
+      "step": 981
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001964,
+      "loss": 1.6797,
+      "step": 982
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001966,
+      "loss": 1.6804,
+      "step": 983
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001968,
+      "loss": 1.734,
+      "step": 984
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019700000000000002,
+      "loss": 1.6371,
+      "step": 985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001972,
+      "loss": 1.8009,
+      "step": 986
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001974,
+      "loss": 1.6523,
+      "step": 987
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001976,
+      "loss": 1.0529,
+      "step": 988
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001978,
+      "loss": 1.6992,
+      "step": 989
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019800000000000002,
+      "loss": 1.5898,
+      "step": 990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019820000000000002,
+      "loss": 1.5265,
+      "step": 991
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001984,
+      "loss": 1.6513,
+      "step": 992
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986,
+      "loss": 1.7308,
+      "step": 993
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988,
+      "loss": 1.5411,
+      "step": 994
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000199,
+      "loss": 1.1584,
+      "step": 995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019920000000000002,
+      "loss": 2.0123,
+      "step": 996
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019940000000000002,
+      "loss": 1.8727,
+      "step": 997
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001996,
+      "loss": 1.2945,
+      "step": 998
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998,
+      "loss": 1.2009,
+      "step": 999
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 1.1232,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999999616592375,
+      "loss": 1.7888,
+      "step": 1001
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999998466369527,
+      "loss": 1.5906,
+      "step": 1002
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999996549331544,
+      "loss": 1.3714,
+      "step": 1003
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999993865478576,
+      "loss": 1.8174,
+      "step": 1004
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999990414810826,
+      "loss": 1.8005,
+      "step": 1005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999998619732856,
+      "loss": 1.5855,
+      "step": 1006
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000199999812130321,
+      "loss": 1.3855,
+      "step": 1007
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999975461921832,
+      "loss": 2.1379,
+      "step": 1008
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999996894399819,
+      "loss": 1.676,
+      "step": 1009
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999996165926168,
+      "loss": 1.8469,
+      "step": 1010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999953607712856,
+      "loss": 1.3865,
+      "step": 1011
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999944789352342,
+      "loss": 1.4885,
+      "step": 1012
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999935204180805,
+      "loss": 1.6213,
+      "step": 1013
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999992485219899,
+      "loss": 1.2801,
+      "step": 1014
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999913733407686,
+      "loss": 1.5892,
+      "step": 1015
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999901847807743,
+      "loss": 1.7003,
+      "step": 1016
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999889195400074,
+      "loss": 2.0935,
+      "step": 1017
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999987577618565,
+      "loss": 1.2431,
+      "step": 1018
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999861590165504,
+      "loss": 1.4906,
+      "step": 1019
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999846637340718,
+      "loss": 2.198,
+      "step": 1020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999983091771244,
+      "loss": 1.3697,
+      "step": 1021
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999814431281875,
+      "loss": 1.7607,
+      "step": 1022
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999979717805029,
+      "loss": 1.4898,
+      "step": 1023
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999779158019006,
+      "loss": 1.5617,
+      "step": 1024
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999760371189405,
+      "loss": 1.5199,
+      "step": 1025
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999740817562927,
+      "loss": 1.7232,
+      "step": 1026
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999972049714107,
+      "loss": 2.0021,
+      "step": 1027
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999699409925395,
+      "loss": 1.544,
+      "step": 1028
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999677555917523,
+      "loss": 1.8454,
+      "step": 1029
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999965493511912,
+      "loss": 1.8829,
+      "step": 1030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999631547531926,
+      "loss": 1.7424,
+      "step": 1031
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999607393157736,
+      "loss": 1.6949,
+      "step": 1032
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999582471998397,
+      "loss": 1.3812,
+      "step": 1033
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999556784055825,
+      "loss": 1.5353,
+      "step": 1034
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999530329331991,
+      "loss": 1.4454,
+      "step": 1035
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999503107828915,
+      "loss": 2.1288,
+      "step": 1036
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999475119548694,
+      "loss": 1.1999,
+      "step": 1037
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999446364493467,
+      "loss": 1.2898,
+      "step": 1038
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999416842665446,
+      "loss": 1.659,
+      "step": 1039
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999386554066888,
+      "loss": 1.6921,
+      "step": 1040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999935549870012,
+      "loss": 1.5766,
+      "step": 1041
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999323676567522,
+      "loss": 1.8223,
+      "step": 1042
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999929108767153,
+      "loss": 1.3857,
+      "step": 1043
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999925773201465,
+      "loss": 1.3386,
+      "step": 1044
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999223609599436,
+      "loss": 1.241,
+      "step": 1045
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999188720428506,
+      "loss": 1.8402,
+      "step": 1046
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999153064504535,
+      "loss": 1.9254,
+      "step": 1047
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999116641830256,
+      "loss": 1.547,
+      "step": 1048
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019999079452408461,
+      "loss": 1.3689,
+      "step": 1049
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019999041496242007,
+      "loss": 1.5347,
+      "step": 1050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019999002773333797,
+      "loss": 1.8277,
+      "step": 1051
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019998963283686809,
+      "loss": 1.5991,
+      "step": 1052
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019998923027304062,
+      "loss": 1.6625,
+      "step": 1053
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999888200418865,
+      "loss": 1.4909,
+      "step": 1054
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019998840214343716,
+      "loss": 1.9095,
+      "step": 1055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019998797657772463,
+      "loss": 1.4709,
+      "step": 1056
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019998754334478157,
+      "loss": 1.6411,
+      "step": 1057
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019998710244464117,
+      "loss": 1.287,
+      "step": 1058
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999866538773373,
+      "loss": 1.7036,
+      "step": 1059
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999861976429043,
+      "loss": 1.4609,
+      "step": 1060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019998573374137716,
+      "loss": 1.8269,
+      "step": 1061
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019998526217279147,
+      "loss": 1.603,
+      "step": 1062
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999847829371834,
+      "loss": 1.5824,
+      "step": 1063
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019998429603458964,
+      "loss": 1.3179,
+      "step": 1064
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999838014650476,
+      "loss": 2.3337,
+      "step": 1065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019998329922859518,
+      "loss": 1.7483,
+      "step": 1066
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019998278932527086,
+      "loss": 1.5988,
+      "step": 1067
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019998227175511377,
+      "loss": 1.9124,
+      "step": 1068
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019998174651816362,
+      "loss": 1.8659,
+      "step": 1069
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019998121361446063,
+      "loss": 1.3368,
+      "step": 1070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999806730440457,
+      "loss": 1.6167,
+      "step": 1071
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019998012480696028,
+      "loss": 0.6805,
+      "step": 1072
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019997956890324637,
+      "loss": 2.0251,
+      "step": 1073
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019997900533294668,
+      "loss": 1.6272,
+      "step": 1074
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019997843409610437,
+      "loss": 1.6154,
+      "step": 1075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019997785519276323,
+      "loss": 1.941,
+      "step": 1076
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019997726862296767,
+      "loss": 1.4109,
+      "step": 1077
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019997667438676273,
+      "loss": 1.722,
+      "step": 1078
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019997607248419387,
+      "loss": 1.3754,
+      "step": 1079
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999754629153073,
+      "loss": 1.4275,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019997484568014975,
+      "loss": 1.4554,
+      "step": 1081
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019997422077876854,
+      "loss": 1.5501,
+      "step": 1082
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019997358821121164,
+      "loss": 1.9821,
+      "step": 1083
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999729479775275,
+      "loss": 1.7819,
+      "step": 1084
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019997230007776524,
+      "loss": 1.8759,
+      "step": 1085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019997164451197454,
+      "loss": 1.5151,
+      "step": 1086
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019997098128020566,
+      "loss": 1.2533,
+      "step": 1087
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019997031038250946,
+      "loss": 1.892,
+      "step": 1088
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019996963181893743,
+      "loss": 1.5802,
+      "step": 1089
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999689455895415,
+      "loss": 1.8327,
+      "step": 1090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019996825169437437,
+      "loss": 1.8246,
+      "step": 1091
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019996755013348928,
+      "loss": 1.5607,
+      "step": 1092
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019996684090693996,
+      "loss": 1.4499,
+      "step": 1093
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999661240147808,
+      "loss": 1.5973,
+      "step": 1094
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999653994570668,
+      "loss": 1.7155,
+      "step": 1095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999646672338535,
+      "loss": 0.9375,
+      "step": 1096
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019996392734519702,
+      "loss": 0.9851,
+      "step": 1097
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019996317979115418,
+      "loss": 1.4553,
+      "step": 1098
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019996242457178225,
+      "loss": 1.6781,
+      "step": 1099
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019996166168713913,
+      "loss": 1.43,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019996089113728333,
+      "loss": 1.2494,
+      "step": 1101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019996011292227394,
+      "loss": 1.5066,
+      "step": 1102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019995932704217065,
+      "loss": 1.6946,
+      "step": 1103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999585334970337,
+      "loss": 1.6747,
+      "step": 1104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019995773228692395,
+      "loss": 1.6018,
+      "step": 1105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019995692341190287,
+      "loss": 1.917,
+      "step": 1106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999561068720324,
+      "loss": 1.6396,
+      "step": 1107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019995528266737524,
+      "loss": 1.4863,
+      "step": 1108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019995445079799455,
+      "loss": 1.4267,
+      "step": 1109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019995361126395414,
+      "loss": 1.6205,
+      "step": 1110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019995276406531835,
+      "loss": 1.0737,
+      "step": 1111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019995190920215218,
+      "loss": 1.4777,
+      "step": 1112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019995104667452118,
+      "loss": 1.125,
+      "step": 1113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019995017648249147,
+      "loss": 1.9411,
+      "step": 1114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019994929862612976,
+      "loss": 1.6535,
+      "step": 1115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019994841310550342,
+      "loss": 1.766,
+      "step": 1116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019994751992068034,
+      "loss": 1.8591,
+      "step": 1117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000199946619071729,
+      "loss": 1.5824,
+      "step": 1118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019994571055871846,
+      "loss": 1.3342,
+      "step": 1119
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019994479438171841,
+      "loss": 1.582,
+      "step": 1120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999438705407991,
+      "loss": 0.964,
+      "step": 1121
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019994293903603132,
+      "loss": 1.1853,
+      "step": 1122
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999419998674866,
+      "loss": 1.8169,
+      "step": 1123
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999410530352369,
+      "loss": 1.428,
+      "step": 1124
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019994009853935478,
+      "loss": 1.4599,
+      "step": 1125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019993913637991352,
+      "loss": 1.5689,
+      "step": 1126
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019993816655698685,
+      "loss": 1.7341,
+      "step": 1127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019993718907064916,
+      "loss": 1.2249,
+      "step": 1128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019993620392097538,
+      "loss": 1.5978,
+      "step": 1129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019993521110804109,
+      "loss": 1.6899,
+      "step": 1130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019993421063192236,
+      "loss": 1.654,
+      "step": 1131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019993320249269594,
+      "loss": 1.5317,
+      "step": 1132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019993218669043917,
+      "loss": 1.2291,
+      "step": 1133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999311632252299,
+      "loss": 1.7205,
+      "step": 1134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019993013209714658,
+      "loss": 1.5196,
+      "step": 1135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999290933062684,
+      "loss": 1.6713,
+      "step": 1136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999280468526749,
+      "loss": 1.8654,
+      "step": 1137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019992699273644636,
+      "loss": 1.4986,
+      "step": 1138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019992593095766363,
+      "loss": 1.373,
+      "step": 1139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019992486151640808,
+      "loss": 1.9369,
+      "step": 1140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019992378441276176,
+      "loss": 1.5975,
+      "step": 1141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999226996468073,
+      "loss": 1.8011,
+      "step": 1142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019992160721862777,
+      "loss": 1.1717,
+      "step": 1143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000199920507128307,
+      "loss": 1.9716,
+      "step": 1144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019991939937592942,
+      "loss": 2.345,
+      "step": 1145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019991828396157983,
+      "loss": 1.5409,
+      "step": 1146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019991716088534387,
+      "loss": 1.639,
+      "step": 1147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019991603014730763,
+      "loss": 1.5572,
+      "step": 1148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019991489174755778,
+      "loss": 1.4659,
+      "step": 1149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019991374568618166,
+      "loss": 1.4672,
+      "step": 1150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019991259196326715,
+      "loss": 1.7143,
+      "step": 1151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019991143057890268,
+      "loss": 1.4404,
+      "step": 1152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999102615331774,
+      "loss": 1.3292,
+      "step": 1153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019990908482618082,
+      "loss": 1.1631,
+      "step": 1154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019990790045800325,
+      "loss": 1.7489,
+      "step": 1155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019990670842873548,
+      "loss": 1.7194,
+      "step": 1156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019990550873846897,
+      "loss": 1.6635,
+      "step": 1157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019990430138729564,
+      "loss": 1.5782,
+      "step": 1158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019990308637530816,
+      "loss": 1.6522,
+      "step": 1159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019990186370259964,
+      "loss": 1.6093,
+      "step": 1160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999006333692638,
+      "loss": 1.4321,
+      "step": 1161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001998993953753951,
+      "loss": 1.8367,
+      "step": 1162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019989814972108834,
+      "loss": 1.4176,
+      "step": 1163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019989689640643915,
+      "loss": 0.9492,
+      "step": 1164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019989563543154355,
+      "loss": 1.9153,
+      "step": 1165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019989436679649828,
+      "loss": 1.7441,
+      "step": 1166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001998930905014006,
+      "loss": 2.1394,
+      "step": 1167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019989180654634837,
+      "loss": 1.5836,
+      "step": 1168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001998905149314401,
+      "loss": 1.5163,
+      "step": 1169
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001998892156567748,
+      "loss": 1.5021,
+      "step": 1170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019988790872245207,
+      "loss": 1.8304,
+      "step": 1171
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001998865941285722,
+      "loss": 1.4324,
+      "step": 1172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019988527187523588,
+      "loss": 1.6711,
+      "step": 1173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998839419625446,
+      "loss": 1.9164,
+      "step": 1174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019988260439060033,
+      "loss": 1.4164,
+      "step": 1175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019988125915950563,
+      "loss": 1.9128,
+      "step": 1176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019987990626936362,
+      "loss": 1.9676,
+      "step": 1177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019987854572027803,
+      "loss": 1.8576,
+      "step": 1178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019987717751235328,
+      "loss": 1.3179,
+      "step": 1179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998758016456942,
+      "loss": 1.7843,
+      "step": 1180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019987441812040636,
+      "loss": 1.6821,
+      "step": 1181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019987302693659576,
+      "loss": 1.4045,
+      "step": 1182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998716280943692,
+      "loss": 1.43,
+      "step": 1183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019987022159383383,
+      "loss": 1.8158,
+      "step": 1184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019986880743509756,
+      "loss": 1.033,
+      "step": 1185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019986738561826882,
+      "loss": 1.9315,
+      "step": 1186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019986595614345664,
+      "loss": 1.5121,
+      "step": 1187
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019986451901077062,
+      "loss": 1.4966,
+      "step": 1188
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000199863074220321,
+      "loss": 1.7536,
+      "step": 1189
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019986162177221855,
+      "loss": 1.2061,
+      "step": 1190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998601616665746,
+      "loss": 1.7144,
+      "step": 1191
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019985869390350117,
+      "loss": 1.2743,
+      "step": 1192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998572184831108,
+      "loss": 1.7466,
+      "step": 1193
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998557354055166,
+      "loss": 1.224,
+      "step": 1194
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019985424467083233,
+      "loss": 1.33,
+      "step": 1195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019985274627917226,
+      "loss": 1.5989,
+      "step": 1196
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019985124023065134,
+      "loss": 1.3786,
+      "step": 1197
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019984972652538504,
+      "loss": 1.7289,
+      "step": 1198
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998482051634894,
+      "loss": 1.6777,
+      "step": 1199
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998466761450811,
+      "loss": 1.7657,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998451394702774,
+      "loss": 1.3408,
+      "step": 1201
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019984359513919615,
+      "loss": 1.5036,
+      "step": 1202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998420431519557,
+      "loss": 1.5569,
+      "step": 1203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019984048350867514,
+      "loss": 1.0635,
+      "step": 1204
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019983891620947405,
+      "loss": 1.5939,
+      "step": 1205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019983734125447256,
+      "loss": 1.4499,
+      "step": 1206
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019983575864379146,
+      "loss": 1.3887,
+      "step": 1207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998341683775522,
+      "loss": 1.6756,
+      "step": 1208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019983257045587657,
+      "loss": 1.9104,
+      "step": 1209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998309648788872,
+      "loss": 1.0815,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998293516467072,
+      "loss": 1.4782,
+      "step": 1211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019982773075946024,
+      "loss": 1.566,
+      "step": 1212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019982610221727065,
+      "loss": 1.2626,
+      "step": 1213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998244660202633,
+      "loss": 1.399,
+      "step": 1214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019982282216856362,
+      "loss": 1.6034,
+      "step": 1215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998211706622977,
+      "loss": 1.757,
+      "step": 1216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019981951150159217,
+      "loss": 1.0758,
+      "step": 1217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019981784468657424,
+      "loss": 1.6212,
+      "step": 1218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998161702173718,
+      "loss": 1.7444,
+      "step": 1219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019981448809411315,
+      "loss": 1.6132,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019981279831692732,
+      "loss": 1.6791,
+      "step": 1221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019981110088594387,
+      "loss": 1.5254,
+      "step": 1222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019980939580129304,
+      "loss": 1.4117,
+      "step": 1223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019980768306310543,
+      "loss": 1.4271,
+      "step": 1224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019980596267151252,
+      "loss": 1.5898,
+      "step": 1225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019980423462664617,
+      "loss": 1.3551,
+      "step": 1226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998024989286389,
+      "loss": 1.5364,
+      "step": 1227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019980075557762377,
+      "loss": 1.799,
+      "step": 1228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019979900457373447,
+      "loss": 2.1997,
+      "step": 1229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019979724591710535,
+      "loss": 1.2061,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019979547960787117,
+      "loss": 1.4895,
+      "step": 1231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019979370564616742,
+      "loss": 1.8482,
+      "step": 1232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997919240321301,
+      "loss": 1.6174,
+      "step": 1233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019979013476589588,
+      "loss": 1.8139,
+      "step": 1234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019978833784760192,
+      "loss": 1.1238,
+      "step": 1235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000199786533277386,
+      "loss": 1.6926,
+      "step": 1236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019978472105538654,
+      "loss": 1.4497,
+      "step": 1237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019978290118174248,
+      "loss": 2.0624,
+      "step": 1238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997810736565934,
+      "loss": 1.8396,
+      "step": 1239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019977923848007935,
+      "loss": 1.6892,
+      "step": 1240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019977739565234115,
+      "loss": 2.2233,
+      "step": 1241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019977554517352006,
+      "loss": 1.1664,
+      "step": 1242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000199773687043758,
+      "loss": 1.6688,
+      "step": 1243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019977182126319746,
+      "loss": 1.9036,
+      "step": 1244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997699478319815,
+      "loss": 1.8421,
+      "step": 1245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019976806675025378,
+      "loss": 1.8864,
+      "step": 1246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019976617801815852,
+      "loss": 1.9347,
+      "step": 1247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019976428163584058,
+      "loss": 1.3916,
+      "step": 1248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019976237760344537,
+      "loss": 1.1155,
+      "step": 1249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019976046592111888,
+      "loss": 1.3935,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019975854658900772,
+      "loss": 2.0206,
+      "step": 1251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019975661960725907,
+      "loss": 1.8721,
+      "step": 1252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019975468497602066,
+      "loss": 1.6133,
+      "step": 1253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019975274269544085,
+      "loss": 1.2146,
+      "step": 1254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019975079276566863,
+      "loss": 1.4609,
+      "step": 1255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997488351868535,
+      "loss": 1.946,
+      "step": 1256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019974686995914554,
+      "loss": 1.4066,
+      "step": 1257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019974489708269544,
+      "loss": 1.5387,
+      "step": 1258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019974291655765455,
+      "loss": 1.7119,
+      "step": 1259
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019974092838417464,
+      "loss": 1.8502,
+      "step": 1260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019973893256240824,
+      "loss": 1.7773,
+      "step": 1261
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997369290925084,
+      "loss": 1.5886,
+      "step": 1262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997349179746287,
+      "loss": 1.746,
+      "step": 1263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019973289920892338,
+      "loss": 1.6751,
+      "step": 1264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019973087279554722,
+      "loss": 0.8801,
+      "step": 1265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019972883873465569,
+      "loss": 1.6432,
+      "step": 1266
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019972679702640465,
+      "loss": 1.4117,
+      "step": 1267
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019972474767095073,
+      "loss": 1.8821,
+      "step": 1268
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019972269066845106,
+      "loss": 1.6506,
+      "step": 1269
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019972062601906338,
+      "loss": 1.1777,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000199718553722946,
+      "loss": 2.1435,
+      "step": 1271
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019971647378025787,
+      "loss": 1.617,
+      "step": 1272
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019971438619115843,
+      "loss": 1.9702,
+      "step": 1273
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019971229095580773,
+      "loss": 1.8505,
+      "step": 1274
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019971018807436654,
+      "loss": 1.2985,
+      "step": 1275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019970807754699603,
+      "loss": 1.7277,
+      "step": 1276
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019970595937385808,
+      "loss": 1.7242,
+      "step": 1277
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997038335551151,
+      "loss": 1.3299,
+      "step": 1278
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019970170009093007,
+      "loss": 1.4397,
+      "step": 1279
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019969955898146665,
+      "loss": 1.627,
+      "step": 1280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019969741022688898,
+      "loss": 1.7539,
+      "step": 1281
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019969525382736183,
+      "loss": 1.4401,
+      "step": 1282
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019969308978305057,
+      "loss": 1.7672,
+      "step": 1283
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019969091809412116,
+      "loss": 1.7444,
+      "step": 1284
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019968873876074006,
+      "loss": 2.0024,
+      "step": 1285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019968655178307447,
+      "loss": 1.3041,
+      "step": 1286
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019968435716129205,
+      "loss": 1.6253,
+      "step": 1287
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019968215489556105,
+      "loss": 1.5687,
+      "step": 1288
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019967994498605043,
+      "loss": 1.4559,
+      "step": 1289
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019967772743292956,
+      "loss": 1.3908,
+      "step": 1290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019967550223636856,
+      "loss": 1.4958,
+      "step": 1291
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019967326939653802,
+      "loss": 1.5777,
+      "step": 1292
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019967102891360914,
+      "loss": 1.7781,
+      "step": 1293
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001996687807877538,
+      "loss": 1.4241,
+      "step": 1294
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001996665250191443,
+      "loss": 1.3477,
+      "step": 1295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019966426160795365,
+      "loss": 1.0591,
+      "step": 1296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019966199055435543,
+      "loss": 1.0136,
+      "step": 1297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019965971185852377,
+      "loss": 1.7963,
+      "step": 1298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019965742552063343,
+      "loss": 1.3586,
+      "step": 1299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001996551315408597,
+      "loss": 1.6656,
+      "step": 1300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019965282991937845,
+      "loss": 1.5273,
+      "step": 1301
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019965052065636627,
+      "loss": 1.617,
+      "step": 1302
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019964820375200019,
+      "loss": 1.4233,
+      "step": 1303
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019964587920645783,
+      "loss": 1.46,
+      "step": 1304
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019964354701991748,
+      "loss": 1.7434,
+      "step": 1305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000199641207192558,
+      "loss": 1.4286,
+      "step": 1306
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019963885972455877,
+      "loss": 1.8445,
+      "step": 1307
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001996365046160998,
+      "loss": 1.7634,
+      "step": 1308
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019963414186736168,
+      "loss": 1.5406,
+      "step": 1309
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019963177147852565,
+      "loss": 1.5303,
+      "step": 1310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001996293934497734,
+      "loss": 2.3938,
+      "step": 1311
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019962700778128732,
+      "loss": 1.8328,
+      "step": 1312
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019962461447325033,
+      "loss": 1.2343,
+      "step": 1313
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019962221352584598,
+      "loss": 1.2969,
+      "step": 1314
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019961980493925834,
+      "loss": 1.5793,
+      "step": 1315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019961738871367211,
+      "loss": 1.1704,
+      "step": 1316
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001996149648492726,
+      "loss": 2.2522,
+      "step": 1317
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019961253334624567,
+      "loss": 1.9089,
+      "step": 1318
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019961009420477774,
+      "loss": 1.425,
+      "step": 1319
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019960764742505584,
+      "loss": 1.6595,
+      "step": 1320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001996051930072677,
+      "loss": 1.5761,
+      "step": 1321
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019960273095160136,
+      "loss": 1.6294,
+      "step": 1322
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019960026125824572,
+      "loss": 1.5933,
+      "step": 1323
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019959778392739016,
+      "loss": 1.1796,
+      "step": 1324
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019959529895922465,
+      "loss": 1.5412,
+      "step": 1325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019959280635393966,
+      "loss": 1.6835,
+      "step": 1326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001995903061117264,
+      "loss": 1.5057,
+      "step": 1327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001995877982327766,
+      "loss": 1.4727,
+      "step": 1328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019958528271728254,
+      "loss": 1.2259,
+      "step": 1329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001995827595654371,
+      "loss": 1.4792,
+      "step": 1330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001995802287774338,
+      "loss": 1.659,
+      "step": 1331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019957769035346668,
+      "loss": 2.0023,
+      "step": 1332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019957514429373038,
+      "loss": 1.2688,
+      "step": 1333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019957259059842018,
+      "loss": 1.2248,
+      "step": 1334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019957002926773184,
+      "loss": 1.6277,
+      "step": 1335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001995674603018618,
+      "loss": 2.0301,
+      "step": 1336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019956488370100704,
+      "loss": 2.1038,
+      "step": 1337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019956229946536516,
+      "loss": 1.5272,
+      "step": 1338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019955970759513427,
+      "loss": 1.8783,
+      "step": 1339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001995571080905132,
+      "loss": 2.0565,
+      "step": 1340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019955450095170124,
+      "loss": 1.7713,
+      "step": 1341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019955188617889832,
+      "loss": 1.3607,
+      "step": 1342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019954926377230493,
+      "loss": 1.5728,
+      "step": 1343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019954663373212214,
+      "loss": 1.0985,
+      "step": 1344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019954399605855166,
+      "loss": 1.9218,
+      "step": 1345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019954135075179574,
+      "loss": 1.3834,
+      "step": 1346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019953869781205724,
+      "loss": 1.5837,
+      "step": 1347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019953603723953957,
+      "loss": 1.7272,
+      "step": 1348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019953336903444677,
+      "loss": 1.7699,
+      "step": 1349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019953069319698344,
+      "loss": 1.5583,
+      "step": 1350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019952800972735472,
+      "loss": 1.7621,
+      "step": 1351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019952531862576643,
+      "loss": 1.8665,
+      "step": 1352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019952261989242493,
+      "loss": 1.7089,
+      "step": 1353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019951991352753714,
+      "loss": 1.8749,
+      "step": 1354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019951719953131063,
+      "loss": 1.4643,
+      "step": 1355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019951447790395345,
+      "loss": 1.8035,
+      "step": 1356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019951174864567432,
+      "loss": 1.8336,
+      "step": 1357
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019950901175668255,
+      "loss": 1.2814,
+      "step": 1358
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019950626723718798,
+      "loss": 1.3214,
+      "step": 1359
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019950351508740108,
+      "loss": 0.7732,
+      "step": 1360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001995007553075329,
+      "loss": 1.6707,
+      "step": 1361
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019949798789779504,
+      "loss": 1.6796,
+      "step": 1362
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001994952128583997,
+      "loss": 1.6437,
+      "step": 1363
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001994924301895597,
+      "loss": 1.6176,
+      "step": 1364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019948963989148844,
+      "loss": 1.8725,
+      "step": 1365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001994868419643998,
+      "loss": 1.4018,
+      "step": 1366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019948403640850844,
+      "loss": 1.6709,
+      "step": 1367
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019948122322402943,
+      "loss": 1.5499,
+      "step": 1368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019947840241117847,
+      "loss": 1.572,
+      "step": 1369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019947557397017194,
+      "loss": 2.5425,
+      "step": 1370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019947273790122666,
+      "loss": 1.7939,
+      "step": 1371
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001994698942045601,
+      "loss": 1.9019,
+      "step": 1372
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019946704288039037,
+      "loss": 1.6677,
+      "step": 1373
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001994641839289361,
+      "loss": 1.251,
+      "step": 1374
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001994613173504165,
+      "loss": 1.3164,
+      "step": 1375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019945844314505138,
+      "loss": 1.7033,
+      "step": 1376
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019945556131306116,
+      "loss": 0.6639,
+      "step": 1377
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019945267185466682,
+      "loss": 1.6279,
+      "step": 1378
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019944977477008992,
+      "loss": 1.4283,
+      "step": 1379
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001994468700595526,
+      "loss": 1.8649,
+      "step": 1380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019944395772327763,
+      "loss": 1.0382,
+      "step": 1381
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001994410377614883,
+      "loss": 1.3025,
+      "step": 1382
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019943811017440852,
+      "loss": 1.579,
+      "step": 1383
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001994351749622628,
+      "loss": 1.8438,
+      "step": 1384
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019943223212527623,
+      "loss": 2.1278,
+      "step": 1385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019942928166367442,
+      "loss": 1.1415,
+      "step": 1386
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001994263235776837,
+      "loss": 1.6444,
+      "step": 1387
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019942335786753078,
+      "loss": 1.6326,
+      "step": 1388
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001994203845334432,
+      "loss": 1.4917,
+      "step": 1389
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019941740357564886,
+      "loss": 1.7738,
+      "step": 1390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001994144149943764,
+      "loss": 1.6398,
+      "step": 1391
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019941141878985501,
+      "loss": 1.8267,
+      "step": 1392
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019940841496231436,
+      "loss": 1.6322,
+      "step": 1393
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019940540351198486,
+      "loss": 1.7445,
+      "step": 1394
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019940238443909742,
+      "loss": 1.3975,
+      "step": 1395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019939935774388351,
+      "loss": 2.2161,
+      "step": 1396
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001993963234265753,
+      "loss": 1.0875,
+      "step": 1397
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019939328148740537,
+      "loss": 1.2952,
+      "step": 1398
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019939023192660704,
+      "loss": 1.7818,
+      "step": 1399
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019938717474441412,
+      "loss": 1.2388,
+      "step": 1400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001993841099410611,
+      "loss": 1.3778,
+      "step": 1401
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019938103751678291,
+      "loss": 1.9278,
+      "step": 1402
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019937795747181522,
+      "loss": 1.4843,
+      "step": 1403
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001993748698063942,
+      "loss": 1.6311,
+      "step": 1404
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019937177452075658,
+      "loss": 1.6259,
+      "step": 1405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019936867161513973,
+      "loss": 1.6534,
+      "step": 1406
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019936556108978158,
+      "loss": 1.6014,
+      "step": 1407
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019936244294492067,
+      "loss": 1.5944,
+      "step": 1408
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019935931718079609,
+      "loss": 1.3875,
+      "step": 1409
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019935618379764752,
+      "loss": 1.5147,
+      "step": 1410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019935304279571528,
+      "loss": 1.7531,
+      "step": 1411
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019934989417524012,
+      "loss": 1.5171,
+      "step": 1412
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001993467379364636,
+      "loss": 1.5102,
+      "step": 1413
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001993435740796277,
+      "loss": 1.2143,
+      "step": 1414
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019934040260497502,
+      "loss": 1.4768,
+      "step": 1415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019933722351274876,
+      "loss": 1.4289,
+      "step": 1416
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019933403680319267,
+      "loss": 1.4513,
+      "step": 1417
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019933084247655117,
+      "loss": 1.2343,
+      "step": 1418
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019932764053306917,
+      "loss": 1.2792,
+      "step": 1419
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001993244309729922,
+      "loss": 1.2887,
+      "step": 1420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019932121379656636,
+      "loss": 1.6972,
+      "step": 1421
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019931798900403838,
+      "loss": 1.0689,
+      "step": 1422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019931475659565553,
+      "loss": 1.6821,
+      "step": 1423
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019931151657166568,
+      "loss": 1.7134,
+      "step": 1424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019930826893231726,
+      "loss": 1.8119,
+      "step": 1425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019930501367785932,
+      "loss": 2.0958,
+      "step": 1426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019930175080854149,
+      "loss": 1.4323,
+      "step": 1427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019929848032461395,
+      "loss": 1.5479,
+      "step": 1428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019929520222632748,
+      "loss": 1.1042,
+      "step": 1429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019929191651393347,
+      "loss": 1.9748,
+      "step": 1430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019928862318768388,
+      "loss": 1.408,
+      "step": 1431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992853222478312,
+      "loss": 1.3941,
+      "step": 1432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019928201369462864,
+      "loss": 1.5302,
+      "step": 1433
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992786975283298,
+      "loss": 2.0303,
+      "step": 1434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019927537374918902,
+      "loss": 1.3624,
+      "step": 1435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019927204235746122,
+      "loss": 1.3195,
+      "step": 1436
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019926870335340176,
+      "loss": 1.8568,
+      "step": 1437
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019926535673726674,
+      "loss": 1.8448,
+      "step": 1438
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019926200250931276,
+      "loss": 1.4233,
+      "step": 1439
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019925864066979706,
+      "loss": 1.2177,
+      "step": 1440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019925527121897741,
+      "loss": 1.2802,
+      "step": 1441
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019925189415711216,
+      "loss": 1.6954,
+      "step": 1442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019924850948446033,
+      "loss": 1.996,
+      "step": 1443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992451172012814,
+      "loss": 1.6295,
+      "step": 1444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019924171730783548,
+      "loss": 1.3455,
+      "step": 1445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019923830980438337,
+      "loss": 1.9775,
+      "step": 1446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992348946911863,
+      "loss": 1.8241,
+      "step": 1447
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019923147196850612,
+      "loss": 1.9538,
+      "step": 1448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019922804163660537,
+      "loss": 1.7473,
+      "step": 1449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019922460369574703,
+      "loss": 1.3144,
+      "step": 1450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019922115814619478,
+      "loss": 1.5808,
+      "step": 1451
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019921770498821276,
+      "loss": 1.5098,
+      "step": 1452
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992142442220658,
+      "loss": 1.6141,
+      "step": 1453
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019921077584801928,
+      "loss": 2.0328,
+      "step": 1454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019920729986633914,
+      "loss": 0.9811,
+      "step": 1455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019920381627729197,
+      "loss": 1.3345,
+      "step": 1456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019920032508114485,
+      "loss": 1.311,
+      "step": 1457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019919682627816553,
+      "loss": 1.3019,
+      "step": 1458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019919331986862227,
+      "loss": 1.98,
+      "step": 1459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019918980585278395,
+      "loss": 1.7383,
+      "step": 1460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019918628423092006,
+      "loss": 1.4734,
+      "step": 1461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019918275500330058,
+      "loss": 2.059,
+      "step": 1462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019917921817019617,
+      "loss": 1.6211,
+      "step": 1463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001991756737318781,
+      "loss": 1.4745,
+      "step": 1464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019917212168861807,
+      "loss": 1.5716,
+      "step": 1465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019916856204068851,
+      "loss": 1.5245,
+      "step": 1466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019916499478836238,
+      "loss": 1.427,
+      "step": 1467
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019916141993191316,
+      "loss": 1.3728,
+      "step": 1468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019915783747161508,
+      "loss": 1.717,
+      "step": 1469
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019915424740774278,
+      "loss": 1.5211,
+      "step": 1470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019915064974057156,
+      "loss": 1.5558,
+      "step": 1471
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019914704447037727,
+      "loss": 1.9065,
+      "step": 1472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019914343159743643,
+      "loss": 1.3734,
+      "step": 1473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019913981112202602,
+      "loss": 1.8626,
+      "step": 1474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001991361830444237,
+      "loss": 1.3503,
+      "step": 1475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019913254736490766,
+      "loss": 1.7776,
+      "step": 1476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019912890408375672,
+      "loss": 1.3826,
+      "step": 1477
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001991252532012502,
+      "loss": 1.7031,
+      "step": 1478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019912159471766812,
+      "loss": 1.4063,
+      "step": 1479
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019911792863329094,
+      "loss": 1.0815,
+      "step": 1480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019911425494839984,
+      "loss": 1.6197,
+      "step": 1481
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001991105736632765,
+      "loss": 1.8958,
+      "step": 1482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001991068847782032,
+      "loss": 1.6812,
+      "step": 1483
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019910318829346286,
+      "loss": 1.7266,
+      "step": 1484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019909948420933886,
+      "loss": 1.4034,
+      "step": 1485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019909577252611528,
+      "loss": 1.5799,
+      "step": 1486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019909205324407676,
+      "loss": 1.7536,
+      "step": 1487
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019908832636350842,
+      "loss": 1.7164,
+      "step": 1488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001990845918846961,
+      "loss": 1.2271,
+      "step": 1489
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019908084980792618,
+      "loss": 1.7138,
+      "step": 1490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019907710013348555,
+      "loss": 1.2916,
+      "step": 1491
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001990733428616618,
+      "loss": 1.5942,
+      "step": 1492
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019906957799274298,
+      "loss": 1.8726,
+      "step": 1493
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019906580552701787,
+      "loss": 1.8333,
+      "step": 1494
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019906202546477565,
+      "loss": 1.5677,
+      "step": 1495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019905823780630628,
+      "loss": 1.8167,
+      "step": 1496
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019905444255190012,
+      "loss": 1.005,
+      "step": 1497
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019905063970184826,
+      "loss": 1.6079,
+      "step": 1498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001990468292564423,
+      "loss": 1.825,
+      "step": 1499
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019904301121597437,
+      "loss": 1.4976,
+      "step": 1500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019903918558073728,
+      "loss": 1.9777,
+      "step": 1501
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019903535235102438,
+      "loss": 1.7396,
+      "step": 1502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019903151152712967,
+      "loss": 1.9393,
+      "step": 1503
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001990276631093476,
+      "loss": 1.3465,
+      "step": 1504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019902380709797325,
+      "loss": 1.8387,
+      "step": 1505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001990199434933024,
+      "loss": 1.3958,
+      "step": 1506
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019901607229563123,
+      "loss": 1.3382,
+      "step": 1507
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019901219350525664,
+      "loss": 1.9568,
+      "step": 1508
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019900830712247603,
+      "loss": 1.9641,
+      "step": 1509
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019900441314758745,
+      "loss": 1.4606,
+      "step": 1510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019900051158088943,
+      "loss": 1.4122,
+      "step": 1511
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019899660242268121,
+      "loss": 1.8886,
+      "step": 1512
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019899268567326255,
+      "loss": 1.5668,
+      "step": 1513
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019898876133293375,
+      "loss": 1.7061,
+      "step": 1514
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019898482940199576,
+      "loss": 1.3365,
+      "step": 1515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001989808898807501,
+      "loss": 1.7079,
+      "step": 1516
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019897694276949882,
+      "loss": 1.6279,
+      "step": 1517
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019897298806854466,
+      "loss": 1.8062,
+      "step": 1518
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001989690257781908,
+      "loss": 0.9167,
+      "step": 1519
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019896505589874107,
+      "loss": 1.664,
+      "step": 1520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019896107843049994,
+      "loss": 1.7497,
+      "step": 1521
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019895709337377238,
+      "loss": 1.8846,
+      "step": 1522
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000198953100728864,
+      "loss": 1.799,
+      "step": 1523
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019894910049608088,
+      "loss": 1.6051,
+      "step": 1524
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019894509267572987,
+      "loss": 1.6329,
+      "step": 1525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019894107726811826,
+      "loss": 1.7622,
+      "step": 1526
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001989370542735539,
+      "loss": 1.1252,
+      "step": 1527
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019893302369234535,
+      "loss": 2.0822,
+      "step": 1528
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019892898552480165,
+      "loss": 1.6121,
+      "step": 1529
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019892493977123247,
+      "loss": 1.5094,
+      "step": 1530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019892088643194803,
+      "loss": 1.6254,
+      "step": 1531
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019891682550725914,
+      "loss": 1.6999,
+      "step": 1532
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001989127569974772,
+      "loss": 1.6781,
+      "step": 1533
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019890868090291418,
+      "loss": 1.5586,
+      "step": 1534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001989045972238827,
+      "loss": 1.9336,
+      "step": 1535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019890050596069582,
+      "loss": 1.6956,
+      "step": 1536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019889640711366735,
+      "loss": 1.6847,
+      "step": 1537
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001988923006831115,
+      "loss": 1.6616,
+      "step": 1538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019888818666934327,
+      "loss": 1.8854,
+      "step": 1539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019888406507267802,
+      "loss": 1.5648,
+      "step": 1540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019887993589343185,
+      "loss": 1.7745,
+      "step": 1541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001988757991319214,
+      "loss": 1.4555,
+      "step": 1542
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001988716547884639,
+      "loss": 1.7239,
+      "step": 1543
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019886750286337706,
+      "loss": 1.787,
+      "step": 1544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019886334335697932,
+      "loss": 1.6121,
+      "step": 1545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019885917626958965,
+      "loss": 1.3997,
+      "step": 1546
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019885500160152756,
+      "loss": 2.272,
+      "step": 1547
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001988508193531132,
+      "loss": 1.4404,
+      "step": 1548
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019884662952466723,
+      "loss": 1.5828,
+      "step": 1549
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019884243211651094,
+      "loss": 1.6946,
+      "step": 1550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019883822712896622,
+      "loss": 1.9118,
+      "step": 1551
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001988340145623555,
+      "loss": 1.5252,
+      "step": 1552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001988297944170018,
+      "loss": 1.6766,
+      "step": 1553
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019882556669322874,
+      "loss": 1.6392,
+      "step": 1554
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001988213313913605,
+      "loss": 1.6279,
+      "step": 1555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019881708851172185,
+      "loss": 1.9981,
+      "step": 1556
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019881283805463814,
+      "loss": 1.6679,
+      "step": 1557
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019880858002043533,
+      "loss": 1.8964,
+      "step": 1558
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019880431440943988,
+      "loss": 1.564,
+      "step": 1559
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001988000412219789,
+      "loss": 1.6576,
+      "step": 1560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001987957604583801,
+      "loss": 1.4925,
+      "step": 1561
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019879147211897169,
+      "loss": 1.7825,
+      "step": 1562
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019878717620408254,
+      "loss": 1.6132,
+      "step": 1563
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019878287271404203,
+      "loss": 1.8057,
+      "step": 1564
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001987785616491802,
+      "loss": 1.8167,
+      "step": 1565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001987742430098276,
+      "loss": 1.7433,
+      "step": 1566
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001987699167963154,
+      "loss": 2.1354,
+      "step": 1567
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019876558300897538,
+      "loss": 1.4817,
+      "step": 1568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019876124164813977,
+      "loss": 2.0497,
+      "step": 1569
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019875689271414153,
+      "loss": 1.3135,
+      "step": 1570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019875253620731416,
+      "loss": 1.1902,
+      "step": 1571
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019874817212799169,
+      "loss": 1.8308,
+      "step": 1572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019874380047650876,
+      "loss": 1.5427,
+      "step": 1573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019873942125320064,
+      "loss": 1.8618,
+      "step": 1574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019873503445840308,
+      "loss": 1.0873,
+      "step": 1575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001987306400924525,
+      "loss": 1.6183,
+      "step": 1576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019872623815568588,
+      "loss": 1.5537,
+      "step": 1577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019872182864844073,
+      "loss": 1.6859,
+      "step": 1578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019871741157105518,
+      "loss": 1.5013,
+      "step": 1579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019871298692386796,
+      "loss": 1.6614,
+      "step": 1580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019870855470721836,
+      "loss": 1.6294,
+      "step": 1581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019870411492144623,
+      "loss": 1.7148,
+      "step": 1582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019869966756689203,
+      "loss": 1.2225,
+      "step": 1583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001986952126438968,
+      "loss": 1.6724,
+      "step": 1584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001986907501528021,
+      "loss": 1.6493,
+      "step": 1585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001986862800939502,
+      "loss": 1.5294,
+      "step": 1586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019868180246768383,
+      "loss": 1.702,
+      "step": 1587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001986773172743463,
+      "loss": 1.2998,
+      "step": 1588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001986728245142816,
+      "loss": 1.7484,
+      "step": 1589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019866832418783428,
+      "loss": 1.5024,
+      "step": 1590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019866381629534931,
+      "loss": 1.4937,
+      "step": 1591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019865930083717248,
+      "loss": 1.8946,
+      "step": 1592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019865477781364995,
+      "loss": 1.7346,
+      "step": 1593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001986502472251286,
+      "loss": 1.6607,
+      "step": 1594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001986457090719559,
+      "loss": 1.4158,
+      "step": 1595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019864116335447968,
+      "loss": 1.3133,
+      "step": 1596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019863661007304867,
+      "loss": 1.5621,
+      "step": 1597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019863204922801196,
+      "loss": 1.4285,
+      "step": 1598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019862748081971927,
+      "loss": 2.1741,
+      "step": 1599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001986229048485209,
+      "loss": 1.6359,
+      "step": 1600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019861832131476783,
+      "loss": 1.8731,
+      "step": 1601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019861373021881143,
+      "loss": 1.5141,
+      "step": 1602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019860913156100383,
+      "loss": 1.3765,
+      "step": 1603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019860452534169762,
+      "loss": 1.1116,
+      "step": 1604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019859991156124602,
+      "loss": 1.0032,
+      "step": 1605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001985952902200028,
+      "loss": 1.5728,
+      "step": 1606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019859066131832238,
+      "loss": 1.5409,
+      "step": 1607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019858602485655965,
+      "loss": 1.8583,
+      "step": 1608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001985813808350702,
+      "loss": 1.6484,
+      "step": 1609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001985767292542101,
+      "loss": 1.5577,
+      "step": 1610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001985720701143361,
+      "loss": 1.8338,
+      "step": 1611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019856740341580537,
+      "loss": 1.4185,
+      "step": 1612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019856272915897585,
+      "loss": 1.7001,
+      "step": 1613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001985580473442059,
+      "loss": 1.5326,
+      "step": 1614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001985533579718546,
+      "loss": 1.665,
+      "step": 1615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019854866104228148,
+      "loss": 1.9196,
+      "step": 1616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019854395655584677,
+      "loss": 1.969,
+      "step": 1617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019853924451291113,
+      "loss": 1.685,
+      "step": 1618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019853452491383594,
+      "loss": 1.4493,
+      "step": 1619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001985297977589831,
+      "loss": 1.7276,
+      "step": 1620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001985250630487151,
+      "loss": 1.9013,
+      "step": 1621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000198520320783395,
+      "loss": 1.4744,
+      "step": 1622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019851557096338645,
+      "loss": 1.5524,
+      "step": 1623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019851081358905368,
+      "loss": 1.5209,
+      "step": 1624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019850604866076147,
+      "loss": 1.8781,
+      "step": 1625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001985012761788752,
+      "loss": 1.1325,
+      "step": 1626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019849649614376086,
+      "loss": 2.0907,
+      "step": 1627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019849170855578495,
+      "loss": 2.2304,
+      "step": 1628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019848691341531463,
+      "loss": 1.5723,
+      "step": 1629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019848211072271758,
+      "loss": 1.706,
+      "step": 1630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019847730047836207,
+      "loss": 1.7547,
+      "step": 1631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019847248268261699,
+      "loss": 1.2735,
+      "step": 1632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001984676573358517,
+      "loss": 1.5775,
+      "step": 1633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001984628244384363,
+      "loss": 1.4095,
+      "step": 1634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019845798399074135,
+      "loss": 1.667,
+      "step": 1635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019845313599313805,
+      "loss": 1.6426,
+      "step": 1636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019844828044599807,
+      "loss": 1.1883,
+      "step": 1637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019844341734969384,
+      "loss": 1.5695,
+      "step": 1638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019843854670459823,
+      "loss": 1.716,
+      "step": 1639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001984336685110847,
+      "loss": 1.6099,
+      "step": 1640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019842878276952737,
+      "loss": 1.784,
+      "step": 1641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001984238894803008,
+      "loss": 1.1405,
+      "step": 1642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019841898864378035,
+      "loss": 1.7758,
+      "step": 1643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019841408026034172,
+      "loss": 1.6016,
+      "step": 1644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001984091643303613,
+      "loss": 1.6177,
+      "step": 1645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019840424085421608,
+      "loss": 2.2188,
+      "step": 1646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001983993098322836,
+      "loss": 1.2416,
+      "step": 1647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019839437126494192,
+      "loss": 1.7221,
+      "step": 1648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019838942515256985,
+      "loss": 2.0985,
+      "step": 1649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019838447149554656,
+      "loss": 1.5393,
+      "step": 1650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019837951029425197,
+      "loss": 1.537,
+      "step": 1651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019837454154906646,
+      "loss": 1.6845,
+      "step": 1652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019836956526037106,
+      "loss": 1.5273,
+      "step": 1653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019836458142854743,
+      "loss": 2.1019,
+      "step": 1654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001983595900539776,
+      "loss": 1.718,
+      "step": 1655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019835459113704443,
+      "loss": 1.8613,
+      "step": 1656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019834958467813122,
+      "loss": 1.4445,
+      "step": 1657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019834457067762181,
+      "loss": 1.8627,
+      "step": 1658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019833954913590077,
+      "loss": 1.7775,
+      "step": 1659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001983345200533531,
+      "loss": 1.8206,
+      "step": 1660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019832948343036447,
+      "loss": 1.4839,
+      "step": 1661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001983244392673211,
+      "loss": 1.4074,
+      "step": 1662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019831938756460974,
+      "loss": 1.5359,
+      "step": 1663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001983143283226178,
+      "loss": 1.4209,
+      "step": 1664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001983092615417332,
+      "loss": 2.3472,
+      "step": 1665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001983041872223445,
+      "loss": 1.8014,
+      "step": 1666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019829910536484083,
+      "loss": 1.3254,
+      "step": 1667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001982940159696118,
+      "loss": 1.3401,
+      "step": 1668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019828891903704769,
+      "loss": 1.5109,
+      "step": 1669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019828381456753942,
+      "loss": 1.3983,
+      "step": 1670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019827870256147832,
+      "loss": 2.2853,
+      "step": 1671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001982735830192564,
+      "loss": 1.6241,
+      "step": 1672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019826845594126628,
+      "loss": 1.6241,
+      "step": 1673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001982633213279011,
+      "loss": 1.4865,
+      "step": 1674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019825817917955453,
+      "loss": 0.763,
+      "step": 1675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019825302949662095,
+      "loss": 1.8661,
+      "step": 1676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001982478722794952,
+      "loss": 1.5548,
+      "step": 1677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019824270752857276,
+      "loss": 1.8334,
+      "step": 1678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019823753524424967,
+      "loss": 1.3032,
+      "step": 1679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019823235542692258,
+      "loss": 1.2807,
+      "step": 1680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019822716807698862,
+      "loss": 1.1518,
+      "step": 1681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019822197319484565,
+      "loss": 1.4208,
+      "step": 1682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001982167707808919,
+      "loss": 1.6015,
+      "step": 1683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019821156083552643,
+      "loss": 1.773,
+      "step": 1684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019820634335914868,
+      "loss": 1.7888,
+      "step": 1685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019820111835215873,
+      "loss": 1.4091,
+      "step": 1686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019819588581495724,
+      "loss": 2.1795,
+      "step": 1687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019819064574794548,
+      "loss": 1.4805,
+      "step": 1688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019818539815152523,
+      "loss": 1.7979,
+      "step": 1689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019818014302609892,
+      "loss": 1.4398,
+      "step": 1690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001981748803720695,
+      "loss": 1.5326,
+      "step": 1691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019816961018984052,
+      "loss": 1.3781,
+      "step": 1692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001981643324798161,
+      "loss": 2.3842,
+      "step": 1693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019815904724240096,
+      "loss": 2.4454,
+      "step": 1694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019815375447800038,
+      "loss": 1.903,
+      "step": 1695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019814845418702017,
+      "loss": 2.0372,
+      "step": 1696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019814314636986682,
+      "loss": 1.6979,
+      "step": 1697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019813783102694735,
+      "loss": 1.7907,
+      "step": 1698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019813250815866928,
+      "loss": 1.8774,
+      "step": 1699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019812717776544086,
+      "loss": 1.8545,
+      "step": 1700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019812183984767073,
+      "loss": 1.36,
+      "step": 1701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001981164944057683,
+      "loss": 1.3304,
+      "step": 1702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019811114144014348,
+      "loss": 1.387,
+      "step": 1703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019810578095120663,
+      "loss": 1.8198,
+      "step": 1704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019810041293936891,
+      "loss": 1.4018,
+      "step": 1705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019809503740504192,
+      "loss": 1.4006,
+      "step": 1706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001980896543486378,
+      "loss": 1.3799,
+      "step": 1707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019808426377056942,
+      "loss": 1.6757,
+      "step": 1708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001980788656712501,
+      "loss": 1.7669,
+      "step": 1709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019807346005109376,
+      "loss": 1.8213,
+      "step": 1710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019806804691051494,
+      "loss": 1.7408,
+      "step": 1711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019806262624992872,
+      "loss": 2.1271,
+      "step": 1712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019805719806975075,
+      "loss": 1.2982,
+      "step": 1713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019805176237039729,
+      "loss": 1.3378,
+      "step": 1714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019804631915228512,
+      "loss": 1.5828,
+      "step": 1715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019804086841583167,
+      "loss": 1.4387,
+      "step": 1716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001980354101614549,
+      "loss": 1.7554,
+      "step": 1717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001980299443895734,
+      "loss": 1.5877,
+      "step": 1718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001980244711006062,
+      "loss": 1.2423,
+      "step": 1719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001980189902949731,
+      "loss": 1.9791,
+      "step": 1720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001980135019730943,
+      "loss": 1.5506,
+      "step": 1721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019800800613539068,
+      "loss": 1.674,
+      "step": 1722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019800250278228368,
+      "loss": 1.2964,
+      "step": 1723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001979969919141953,
+      "loss": 1.4136,
+      "step": 1724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019799147353154813,
+      "loss": 1.6591,
+      "step": 1725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019798594763476528,
+      "loss": 1.6527,
+      "step": 1726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019798041422427055,
+      "loss": 1.7984,
+      "step": 1727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019797487330048822,
+      "loss": 1.5454,
+      "step": 1728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001979693248638432,
+      "loss": 1.4277,
+      "step": 1729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001979637689147609,
+      "loss": 1.6528,
+      "step": 1730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001979582054536674,
+      "loss": 1.9291,
+      "step": 1731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001979526344809893,
+      "loss": 1.2944,
+      "step": 1732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019794705599715382,
+      "loss": 1.7818,
+      "step": 1733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019794147000258867,
+      "loss": 1.5221,
+      "step": 1734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019793587649772224,
+      "loss": 1.9702,
+      "step": 1735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019793027548298343,
+      "loss": 1.4438,
+      "step": 1736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019792466695880173,
+      "loss": 1.4629,
+      "step": 1737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019791905092560722,
+      "loss": 1.3288,
+      "step": 1738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019791342738383056,
+      "loss": 1.3713,
+      "step": 1739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019790779633390294,
+      "loss": 1.5747,
+      "step": 1740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001979021577762562,
+      "loss": 1.9945,
+      "step": 1741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019789651171132265,
+      "loss": 1.3085,
+      "step": 1742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019789085813953525,
+      "loss": 2.0554,
+      "step": 1743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001978851970613276,
+      "loss": 1.8751,
+      "step": 1744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019787952847713373,
+      "loss": 1.8022,
+      "step": 1745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019787385238738834,
+      "loss": 1.4621,
+      "step": 1746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001978681687925267,
+      "loss": 2.1184,
+      "step": 1747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019786247769298457,
+      "loss": 1.7887,
+      "step": 1748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019785677908919843,
+      "loss": 1.6106,
+      "step": 1749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001978510729816052,
+      "loss": 1.8694,
+      "step": 1750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019784535937064247,
+      "loss": 1.5669,
+      "step": 1751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019783963825674832,
+      "loss": 1.179,
+      "step": 1752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019783390964036153,
+      "loss": 1.3184,
+      "step": 1753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019782817352192132,
+      "loss": 1.7786,
+      "step": 1754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019782242990186755,
+      "loss": 1.7135,
+      "step": 1755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019781667878064068,
+      "loss": 1.637,
+      "step": 1756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001978109201586817,
+      "loss": 1.5029,
+      "step": 1757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019780515403643216,
+      "loss": 1.6933,
+      "step": 1758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019779938041433428,
+      "loss": 1.75,
+      "step": 1759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019779359929283073,
+      "loss": 2.0842,
+      "step": 1760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019778781067236482,
+      "loss": 1.6766,
+      "step": 1761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019778201455338044,
+      "loss": 2.219,
+      "step": 1762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019777621093632208,
+      "loss": 1.3477,
+      "step": 1763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019777039982163476,
+      "loss": 1.6945,
+      "step": 1764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019776458120976403,
+      "loss": 1.3282,
+      "step": 1765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001977587551011561,
+      "loss": 1.053,
+      "step": 1766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019775292149625776,
+      "loss": 1.5487,
+      "step": 1767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019774708039551627,
+      "loss": 1.6721,
+      "step": 1768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019774123179937963,
+      "loss": 1.8296,
+      "step": 1769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019773537570829622,
+      "loss": 1.5298,
+      "step": 1770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019772951212271517,
+      "loss": 1.1138,
+      "step": 1771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019772364104308606,
+      "loss": 1.4403,
+      "step": 1772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019771776246985915,
+      "loss": 1.7502,
+      "step": 1773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019771187640348515,
+      "loss": 1.8835,
+      "step": 1774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019770598284441542,
+      "loss": 1.3865,
+      "step": 1775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019770008179310192,
+      "loss": 0.9773,
+      "step": 1776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019769417324999716,
+      "loss": 1.7437,
+      "step": 1777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001976882572155542,
+      "loss": 1.6239,
+      "step": 1778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019768233369022665,
+      "loss": 1.4523,
+      "step": 1779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001976764026744688,
+      "loss": 1.6029,
+      "step": 1780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019767046416873543,
+      "loss": 1.2047,
+      "step": 1781
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001976645181734819,
+      "loss": 1.6607,
+      "step": 1782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019765856468916413,
+      "loss": 0.9571,
+      "step": 1783
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019765260371623871,
+      "loss": 1.6763,
+      "step": 1784
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019764663525516268,
+      "loss": 0.9215,
+      "step": 1785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019764065930639375,
+      "loss": 1.7415,
+      "step": 1786
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019763467587039014,
+      "loss": 1.3168,
+      "step": 1787
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001976286849476107,
+      "loss": 1.6845,
+      "step": 1788
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019762268653851476,
+      "loss": 1.3877,
+      "step": 1789
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019761668064356233,
+      "loss": 1.4111,
+      "step": 1790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019761066726321395,
+      "loss": 1.4097,
+      "step": 1791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019760464639793077,
+      "loss": 1.2232,
+      "step": 1792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001975986180481744,
+      "loss": 1.2966,
+      "step": 1793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019759258221440718,
+      "loss": 1.428,
+      "step": 1794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019758653889709188,
+      "loss": 1.5208,
+      "step": 1795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019758048809669196,
+      "loss": 1.6043,
+      "step": 1796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001975744298136714,
+      "loss": 1.4989,
+      "step": 1797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001975683640484947,
+      "loss": 1.8667,
+      "step": 1798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019756229080162709,
+      "loss": 1.2962,
+      "step": 1799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001975562100735342,
+      "loss": 1.0854,
+      "step": 1800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019755012186468233,
+      "loss": 1.458,
+      "step": 1801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019754402617553834,
+      "loss": 1.9143,
+      "step": 1802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019753792300656966,
+      "loss": 1.7645,
+      "step": 1803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019753181235824426,
+      "loss": 1.6605,
+      "step": 1804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019752569423103078,
+      "loss": 0.6468,
+      "step": 1805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001975195686253983,
+      "loss": 1.7029,
+      "step": 1806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019751343554181658,
+      "loss": 1.6239,
+      "step": 1807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001975072949807559,
+      "loss": 1.5664,
+      "step": 1808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001975011469426871,
+      "loss": 2.1031,
+      "step": 1809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019749499142808166,
+      "loss": 1.5345,
+      "step": 1810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019748882843741158,
+      "loss": 1.5946,
+      "step": 1811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019748265797114945,
+      "loss": 1.2555,
+      "step": 1812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001974764800297684,
+      "loss": 1.7503,
+      "step": 1813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019747029461374227,
+      "loss": 1.8522,
+      "step": 1814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019746410172354523,
+      "loss": 1.884,
+      "step": 1815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019745790135965226,
+      "loss": 2.0313,
+      "step": 1816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019745169352253876,
+      "loss": 1.5377,
+      "step": 1817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019744547821268075,
+      "loss": 1.6035,
+      "step": 1818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001974392554305549,
+      "loss": 1.6803,
+      "step": 1819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001974330251766383,
+      "loss": 1.7753,
+      "step": 1820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019742678745140872,
+      "loss": 1.2186,
+      "step": 1821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001974205422553445,
+      "loss": 1.2491,
+      "step": 1822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019741428958892454,
+      "loss": 1.7642,
+      "step": 1823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019740802945262826,
+      "loss": 0.7309,
+      "step": 1824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019740176184693573,
+      "loss": 1.3849,
+      "step": 1825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019739548677232755,
+      "loss": 1.7196,
+      "step": 1826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001973892042292849,
+      "loss": 1.618,
+      "step": 1827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019738291421828952,
+      "loss": 1.5363,
+      "step": 1828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019737661673982374,
+      "loss": 1.5882,
+      "step": 1829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001973703117943705,
+      "loss": 1.9267,
+      "step": 1830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019736399938241326,
+      "loss": 1.3972,
+      "step": 1831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000197357679504436,
+      "loss": 1.7938,
+      "step": 1832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019735135216092346,
+      "loss": 1.7985,
+      "step": 1833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019734501735236073,
+      "loss": 1.9166,
+      "step": 1834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019733867507923362,
+      "loss": 1.4424,
+      "step": 1835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019733232534202842,
+      "loss": 1.514,
+      "step": 1836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019732596814123204,
+      "loss": 1.001,
+      "step": 1837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019731960347733204,
+      "loss": 2.0273,
+      "step": 1838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019731323135081642,
+      "loss": 1.2652,
+      "step": 1839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019730685176217377,
+      "loss": 1.9648,
+      "step": 1840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019730046471189331,
+      "loss": 1.4204,
+      "step": 1841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019729407020046487,
+      "loss": 1.4792,
+      "step": 1842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001972876682283787,
+      "loss": 1.4998,
+      "step": 1843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019728125879612575,
+      "loss": 1.5384,
+      "step": 1844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001972748419041975,
+      "loss": 1.3097,
+      "step": 1845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019726841755308606,
+      "loss": 1.2573,
+      "step": 1846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000197261985743284,
+      "loss": 1.9107,
+      "step": 1847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001972555464752845,
+      "loss": 2.1236,
+      "step": 1848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019724909974958138,
+      "loss": 2.0249,
+      "step": 1849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019724264556666895,
+      "loss": 1.9263,
+      "step": 1850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019723618392704218,
+      "loss": 1.9351,
+      "step": 1851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019722971483119653,
+      "loss": 1.7432,
+      "step": 1852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019722323827962805,
+      "loss": 1.4716,
+      "step": 1853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019721675427283338,
+      "loss": 1.5685,
+      "step": 1854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001972102628113097,
+      "loss": 1.3638,
+      "step": 1855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019720376389555482,
+      "loss": 1.6562,
+      "step": 1856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019719725752606707,
+      "loss": 1.4161,
+      "step": 1857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019719074370334538,
+      "loss": 1.7038,
+      "step": 1858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019718422242788922,
+      "loss": 1.5867,
+      "step": 1859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001971776937001987,
+      "loss": 2.2932,
+      "step": 1860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019717115752077436,
+      "loss": 1.3529,
+      "step": 1861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019716461389011748,
+      "loss": 1.7994,
+      "step": 1862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019715806280872985,
+      "loss": 1.2575,
+      "step": 1863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019715150427711376,
+      "loss": 1.6392,
+      "step": 1864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019714493829577214,
+      "loss": 1.7277,
+      "step": 1865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019713836486520852,
+      "loss": 2.0725,
+      "step": 1866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001971317839859269,
+      "loss": 2.0316,
+      "step": 1867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019712519565843196,
+      "loss": 1.5045,
+      "step": 1868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019711859988322887,
+      "loss": 1.7322,
+      "step": 1869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019711199666082343,
+      "loss": 1.63,
+      "step": 1870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019710538599172197,
+      "loss": 1.4958,
+      "step": 1871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019709876787643143,
+      "loss": 1.5847,
+      "step": 1872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019709214231545928,
+      "loss": 1.7967,
+      "step": 1873
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019708550930931356,
+      "loss": 1.0983,
+      "step": 1874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001970788688585029,
+      "loss": 1.5889,
+      "step": 1875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019707222096353654,
+      "loss": 1.5664,
+      "step": 1876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019706556562492424,
+      "loss": 1.404,
+      "step": 1877
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001970589028431763,
+      "loss": 1.6107,
+      "step": 1878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019705223261880366,
+      "loss": 1.8161,
+      "step": 1879
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019704555495231782,
+      "loss": 1.1279,
+      "step": 1880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019703886984423082,
+      "loss": 1.1523,
+      "step": 1881
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019703217729505524,
+      "loss": 1.352,
+      "step": 1882
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019702547730530436,
+      "loss": 1.8711,
+      "step": 1883
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001970187698754919,
+      "loss": 1.6453,
+      "step": 1884
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019701205500613218,
+      "loss": 2.399,
+      "step": 1885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019700533269774014,
+      "loss": 1.567,
+      "step": 1886
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019699860295083123,
+      "loss": 1.6423,
+      "step": 1887
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001969918657659215,
+      "loss": 1.8901,
+      "step": 1888
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019698512114352758,
+      "loss": 1.9407,
+      "step": 1889
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019697836908416668,
+      "loss": 1.531,
+      "step": 1890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019697160958835649,
+      "loss": 1.894,
+      "step": 1891
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019696484265661538,
+      "loss": 2.0667,
+      "step": 1892
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019695806828946228,
+      "loss": 1.7176,
+      "step": 1893
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001969512864874166,
+      "loss": 1.2177,
+      "step": 1894
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019694449725099844,
+      "loss": 1.7886,
+      "step": 1895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001969377005807284,
+      "loss": 1.8977,
+      "step": 1896
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019693089647712756,
+      "loss": 1.7262,
+      "step": 1897
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019692408494071778,
+      "loss": 1.831,
+      "step": 1898
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019691726597202133,
+      "loss": 1.4923,
+      "step": 1899
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019691043957156113,
+      "loss": 1.4864,
+      "step": 1900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001969036057398606,
+      "loss": 1.3602,
+      "step": 1901
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001968967644774438,
+      "loss": 1.7918,
+      "step": 1902
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001968899157848353,
+      "loss": 1.5538,
+      "step": 1903
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019688305966256034,
+      "loss": 1.4892,
+      "step": 1904
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019687619611114454,
+      "loss": 1.9118,
+      "step": 1905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019686932513111432,
+      "loss": 1.6268,
+      "step": 1906
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019686244672299647,
+      "loss": 1.4005,
+      "step": 1907
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019685556088731848,
+      "loss": 1.9286,
+      "step": 1908
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019684866762460838,
+      "loss": 1.3809,
+      "step": 1909
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019684176693539473,
+      "loss": 1.5349,
+      "step": 1910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001968348588202067,
+      "loss": 1.6003,
+      "step": 1911
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000196827943279574,
+      "loss": 1.9105,
+      "step": 1912
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019682102031402693,
+      "loss": 1.4869,
+      "step": 1913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019681408992409634,
+      "loss": 1.9283,
+      "step": 1914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019680715211031373,
+      "loss": 1.6149,
+      "step": 1915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000196800206873211,
+      "loss": 2.021,
+      "step": 1916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001967932542133208,
+      "loss": 1.8421,
+      "step": 1917
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001967862941311762,
+      "loss": 1.7332,
+      "step": 1918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000196779326627311,
+      "loss": 1.6536,
+      "step": 1919
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019677235170225942,
+      "loss": 1.4315,
+      "step": 1920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019676536935655633,
+      "loss": 1.4637,
+      "step": 1921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001967583795907371,
+      "loss": 1.1913,
+      "step": 1922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019675138240533776,
+      "loss": 1.7188,
+      "step": 1923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019674437780089487,
+      "loss": 1.8349,
+      "step": 1924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019673736577794553,
+      "loss": 1.861,
+      "step": 1925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019673034633702744,
+      "loss": 1.7283,
+      "step": 1926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019672331947867888,
+      "loss": 1.6143,
+      "step": 1927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019671628520343868,
+      "loss": 1.349,
+      "step": 1928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019670924351184618,
+      "loss": 1.3587,
+      "step": 1929
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019670219440444143,
+      "loss": 1.5671,
+      "step": 1930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001966951378817649,
+      "loss": 1.4828,
+      "step": 1931
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019668807394435773,
+      "loss": 2.1173,
+      "step": 1932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001966810025927616,
+      "loss": 1.5934,
+      "step": 1933
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019667392382751874,
+      "loss": 1.856,
+      "step": 1934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001966668376491719,
+      "loss": 2.2196,
+      "step": 1935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019665974405826456,
+      "loss": 1.3045,
+      "step": 1936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019665264305534063,
+      "loss": 1.6472,
+      "step": 1937
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019664553464094465,
+      "loss": 1.7159,
+      "step": 1938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019663841881562163,
+      "loss": 1.9235,
+      "step": 1939
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019663129557991727,
+      "loss": 1.7741,
+      "step": 1940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019662416493437782,
+      "loss": 1.9075,
+      "step": 1941
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019661702687955,
+      "loss": 1.9563,
+      "step": 1942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001966098814159812,
+      "loss": 1.7159,
+      "step": 1943
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019660272854421935,
+      "loss": 1.8404,
+      "step": 1944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019659556826481298,
+      "loss": 1.7123,
+      "step": 1945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019658840057831106,
+      "loss": 1.5112,
+      "step": 1946
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001965812254852633,
+      "loss": 1.4216,
+      "step": 1947
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019657404298621987,
+      "loss": 1.6597,
+      "step": 1948
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019656685308173155,
+      "loss": 1.5805,
+      "step": 1949
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019655965577234963,
+      "loss": 1.5466,
+      "step": 1950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019655245105862604,
+      "loss": 1.6441,
+      "step": 1951
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019654523894111324,
+      "loss": 1.072,
+      "step": 1952
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019653801942036428,
+      "loss": 1.495,
+      "step": 1953
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019653079249693276,
+      "loss": 1.3816,
+      "step": 1954
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019652355817137285,
+      "loss": 1.4486,
+      "step": 1955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019651631644423926,
+      "loss": 1.7867,
+      "step": 1956
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019650906731608737,
+      "loss": 1.5814,
+      "step": 1957
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019650181078747296,
+      "loss": 1.9163,
+      "step": 1958
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019649454685895254,
+      "loss": 2.0607,
+      "step": 1959
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001964872755310831,
+      "loss": 1.3368,
+      "step": 1960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001964799968044222,
+      "loss": 2.0411,
+      "step": 1961
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019647271067952803,
+      "loss": 1.2898,
+      "step": 1962
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019646541715695923,
+      "loss": 1.4794,
+      "step": 1963
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019645811623727512,
+      "loss": 1.4432,
+      "step": 1964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019645080792103555,
+      "loss": 1.248,
+      "step": 1965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019644349220880096,
+      "loss": 1.0779,
+      "step": 1966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019643616910113227,
+      "loss": 2.0849,
+      "step": 1967
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019642883859859108,
+      "loss": 1.9316,
+      "step": 1968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001964215007017394,
+      "loss": 2.1628,
+      "step": 1969
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019641415541114005,
+      "loss": 1.6859,
+      "step": 1970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019640680272735623,
+      "loss": 1.4459,
+      "step": 1971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019639944265095172,
+      "loss": 1.543,
+      "step": 1972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019639207518249092,
+      "loss": 1.5044,
+      "step": 1973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001963847003225388,
+      "loss": 1.3582,
+      "step": 1974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019637731807166083,
+      "loss": 1.6538,
+      "step": 1975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019636992843042315,
+      "loss": 1.4783,
+      "step": 1976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019636253139939235,
+      "loss": 1.882,
+      "step": 1977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019635512697913567,
+      "loss": 1.3977,
+      "step": 1978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019634771517022093,
+      "loss": 1.8381,
+      "step": 1979
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019634029597321643,
+      "loss": 1.0639,
+      "step": 1980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001963328693886911,
+      "loss": 1.4192,
+      "step": 1981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019632543541721442,
+      "loss": 1.6645,
+      "step": 1982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019631799405935642,
+      "loss": 1.7848,
+      "step": 1983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019631054531568778,
+      "loss": 1.5819,
+      "step": 1984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019630308918677962,
+      "loss": 1.9735,
+      "step": 1985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001962956256732037,
+      "loss": 1.7593,
+      "step": 1986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019628815477553232,
+      "loss": 1.6955,
+      "step": 1987
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019628067649433842,
+      "loss": 1.8354,
+      "step": 1988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019627319083019538,
+      "loss": 1.6571,
+      "step": 1989
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019626569778367724,
+      "loss": 2.5172,
+      "step": 1990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019625819735535855,
+      "loss": 1.8034,
+      "step": 1991
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001962506895458145,
+      "loss": 1.6538,
+      "step": 1992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001962431743556208,
+      "loss": 1.6813,
+      "step": 1993
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019623565178535367,
+      "loss": 1.0554,
+      "step": 1994
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019622812183559002,
+      "loss": 1.5512,
+      "step": 1995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019622058450690716,
+      "loss": 1.4466,
+      "step": 1996
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019621303979988318,
+      "loss": 1.5688,
+      "step": 1997
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019620548771509658,
+      "loss": 1.9477,
+      "step": 1998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019619792825312642,
+      "loss": 1.5495,
+      "step": 1999
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019619036141455242,
+      "loss": 1.7295,
+      "step": 2000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001961827871999548,
+      "loss": 1.4158,
+      "step": 2001
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001961752056099144,
+      "loss": 1.6282,
+      "step": 2002
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001961676166450125,
+      "loss": 1.6624,
+      "step": 2003
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001961600203058311,
+      "loss": 1.9212,
+      "step": 2004
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019615241659295268,
+      "loss": 1.8345,
+      "step": 2005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001961448055069603,
+      "loss": 1.5379,
+      "step": 2006
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019613718704843762,
+      "loss": 1.554,
+      "step": 2007
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019612956121796882,
+      "loss": 1.7051,
+      "step": 2008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019612192801613865,
+      "loss": 1.7942,
+      "step": 2009
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001961142874435324,
+      "loss": 1.4561,
+      "step": 2010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019610663950073604,
+      "loss": 1.5149,
+      "step": 2011
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019609898418833596,
+      "loss": 1.6678,
+      "step": 2012
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019609132150691925,
+      "loss": 2.2842,
+      "step": 2013
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019608365145707343,
+      "loss": 1.559,
+      "step": 2014
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019607597403938665,
+      "loss": 1.8857,
+      "step": 2015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019606828925444766,
+      "loss": 1.6018,
+      "step": 2016
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019606059710284575,
+      "loss": 1.9765,
+      "step": 2017
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001960528975851707,
+      "loss": 1.7737,
+      "step": 2018
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019604519070201302,
+      "loss": 1.6763,
+      "step": 2019
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001960374764539636,
+      "loss": 1.7541,
+      "step": 2020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019602975484161406,
+      "loss": 1.6099,
+      "step": 2021
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019602202586555643,
+      "loss": 1.5908,
+      "step": 2022
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019601428952638342,
+      "loss": 1.5177,
+      "step": 2023
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019600654582468822,
+      "loss": 1.7144,
+      "step": 2024
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019599879476106467,
+      "loss": 1.8721,
+      "step": 2025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019599103633610712,
+      "loss": 1.2876,
+      "step": 2026
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019598327055041055,
+      "loss": 1.5372,
+      "step": 2027
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019597549740457037,
+      "loss": 1.6314,
+      "step": 2028
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001959677168991827,
+      "loss": 2.045,
+      "step": 2029
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001959599290348441,
+      "loss": 1.5722,
+      "step": 2030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019595213381215179,
+      "loss": 1.4748,
+      "step": 2031
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019594433123170353,
+      "loss": 1.4335,
+      "step": 2032
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001959365212940976,
+      "loss": 1.5209,
+      "step": 2033
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001959287039999329,
+      "loss": 1.9165,
+      "step": 2034
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001959208793498089,
+      "loss": 1.7564,
+      "step": 2035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019591304734432557,
+      "loss": 1.5594,
+      "step": 2036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019590520798408348,
+      "loss": 1.3209,
+      "step": 2037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019589736126968376,
+      "loss": 2.3969,
+      "step": 2038
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019588950720172813,
+      "loss": 2.2395,
+      "step": 2039
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019588164578081884,
+      "loss": 2.358,
+      "step": 2040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019587377700755874,
+      "loss": 1.6477,
+      "step": 2041
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019586590088255116,
+      "loss": 1.0859,
+      "step": 2042
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019585801740640011,
+      "loss": 2.035,
+      "step": 2043
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019585012657971008,
+      "loss": 1.5369,
+      "step": 2044
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019584222840308616,
+      "loss": 1.2849,
+      "step": 2045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000195834322877134,
+      "loss": 1.8995,
+      "step": 2046
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001958264100024598,
+      "loss": 1.6701,
+      "step": 2047
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019581848977967035,
+      "loss": 2.0281,
+      "step": 2048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019581056220937293,
+      "loss": 1.697,
+      "step": 2049
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001958026272921755,
+      "loss": 1.5506,
+      "step": 2050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019579468502868645,
+      "loss": 1.5627,
+      "step": 2051
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019578673541951488,
+      "loss": 1.6686,
+      "step": 2052
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019577877846527035,
+      "loss": 1.2579,
+      "step": 2053
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019577081416656303,
+      "loss": 1.6057,
+      "step": 2054
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001957628425240036,
+      "loss": 1.1976,
+      "step": 2055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019575486353820334,
+      "loss": 1.005,
+      "step": 2056
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001957468772097741,
+      "loss": 1.7498,
+      "step": 2057
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001957388835393283,
+      "loss": 1.5046,
+      "step": 2058
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019573088252747888,
+      "loss": 1.3336,
+      "step": 2059
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001957228741748394,
+      "loss": 1.6395,
+      "step": 2060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001957148584820239,
+      "loss": 1.4685,
+      "step": 2061
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019570683544964713,
+      "loss": 2.0739,
+      "step": 2062
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019569880507832422,
+      "loss": 1.6716,
+      "step": 2063
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019569076736867099,
+      "loss": 1.6985,
+      "step": 2064
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019568272232130376,
+      "loss": 1.4286,
+      "step": 2065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019567466993683944,
+      "loss": 1.7128,
+      "step": 2066
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019566661021589555,
+      "loss": 1.7669,
+      "step": 2067
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019565854315909005,
+      "loss": 2.0816,
+      "step": 2068
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019565046876704158,
+      "loss": 1.6196,
+      "step": 2069
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019564238704036929,
+      "loss": 1.4429,
+      "step": 2070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019563429797969287,
+      "loss": 1.5287,
+      "step": 2071
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019562620158563264,
+      "loss": 1.6299,
+      "step": 2072
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019561809785880946,
+      "loss": 1.2751,
+      "step": 2073
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019560998679984465,
+      "loss": 1.2949,
+      "step": 2074
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019560186840936026,
+      "loss": 2.0358,
+      "step": 2075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019559374268797875,
+      "loss": 1.7807,
+      "step": 2076
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019558560963632332,
+      "loss": 1.8591,
+      "step": 2077
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019557746925501752,
+      "loss": 1.7701,
+      "step": 2078
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001955693215446856,
+      "loss": 1.3299,
+      "step": 2079
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019556116650595238,
+      "loss": 1.4005,
+      "step": 2080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019555300413944316,
+      "loss": 1.2543,
+      "step": 2081
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019554483444578384,
+      "loss": 1.6015,
+      "step": 2082
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001955366574256009,
+      "loss": 1.92,
+      "step": 2083
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019552847307952135,
+      "loss": 1.5289,
+      "step": 2084
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019552028140817278,
+      "loss": 2.0756,
+      "step": 2085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019551208241218337,
+      "loss": 1.7316,
+      "step": 2086
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019550387609218182,
+      "loss": 1.614,
+      "step": 2087
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019549566244879735,
+      "loss": 1.5324,
+      "step": 2088
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019548744148265989,
+      "loss": 1.1603,
+      "step": 2089
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019547921319439975,
+      "loss": 1.5576,
+      "step": 2090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019547097758464794,
+      "loss": 1.8769,
+      "step": 2091
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019546273465403596,
+      "loss": 1.9048,
+      "step": 2092
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019545448440319585,
+      "loss": 1.4192,
+      "step": 2093
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019544622683276034,
+      "loss": 1.0893,
+      "step": 2094
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019543796194336258,
+      "loss": 1.477,
+      "step": 2095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001954296897356363,
+      "loss": 1.7956,
+      "step": 2096
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001954214102102159,
+      "loss": 1.5579,
+      "step": 2097
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019541312336773624,
+      "loss": 1.4804,
+      "step": 2098
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019540482920883277,
+      "loss": 1.6016,
+      "step": 2099
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019539652773414146,
+      "loss": 1.5193,
+      "step": 2100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001953882189442989,
+      "loss": 1.6186,
+      "step": 2101
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019537990283994224,
+      "loss": 1.2664,
+      "step": 2102
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019537157942170922,
+      "loss": 1.3651,
+      "step": 2103
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019536324869023796,
+      "loss": 1.8141,
+      "step": 2104
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019535491064616737,
+      "loss": 1.6484,
+      "step": 2105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019534656529013683,
+      "loss": 2.0468,
+      "step": 2106
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001953382126227862,
+      "loss": 1.4856,
+      "step": 2107
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019532985264475604,
+      "loss": 1.324,
+      "step": 2108
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019532148535668739,
+      "loss": 1.3525,
+      "step": 2109
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019531311075922183,
+      "loss": 1.5677,
+      "step": 2110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019530472885300162,
+      "loss": 1.6711,
+      "step": 2111
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019529633963866942,
+      "loss": 2.0118,
+      "step": 2112
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019528794311686855,
+      "loss": 1.8816,
+      "step": 2113
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019527953928824289,
+      "loss": 1.5739,
+      "step": 2114
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019527112815343682,
+      "loss": 1.7916,
+      "step": 2115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019526270971309537,
+      "loss": 2.1956,
+      "step": 2116
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019525428396786402,
+      "loss": 1.7191,
+      "step": 2117
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019524585091838892,
+      "loss": 1.8557,
+      "step": 2118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001952374105653167,
+      "loss": 1.7166,
+      "step": 2119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019522896290929457,
+      "loss": 1.9221,
+      "step": 2120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019522050795097037,
+      "loss": 1.361,
+      "step": 2121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019521204569099236,
+      "loss": 1.526,
+      "step": 2122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019520357613000947,
+      "loss": 1.9852,
+      "step": 2123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019519509926867117,
+      "loss": 1.3621,
+      "step": 2124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019518661510762746,
+      "loss": 1.667,
+      "step": 2125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019517812364752894,
+      "loss": 1.5182,
+      "step": 2126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019516962488902676,
+      "loss": 1.6916,
+      "step": 2127
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019516111883277257,
+      "loss": 1.9004,
+      "step": 2128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019515260547941864,
+      "loss": 1.6228,
+      "step": 2129
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019514408482961782,
+      "loss": 1.6073,
+      "step": 2130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019513555688402344,
+      "loss": 1.4014,
+      "step": 2131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001951270216432895,
+      "loss": 1.6478,
+      "step": 2132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019511847910807045,
+      "loss": 1.423,
+      "step": 2133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019510992927902133,
+      "loss": 1.825,
+      "step": 2134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001951013721567978,
+      "loss": 1.7227,
+      "step": 2135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019509280774205598,
+      "loss": 1.4421,
+      "step": 2136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019508423603545265,
+      "loss": 1.6993,
+      "step": 2137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001950756570376451,
+      "loss": 1.6645,
+      "step": 2138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001950670707492911,
+      "loss": 1.4614,
+      "step": 2139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001950584771710492,
+      "loss": 1.892,
+      "step": 2140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019504987630357824,
+      "loss": 1.8624,
+      "step": 2141
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019504126814753783,
+      "loss": 1.5937,
+      "step": 2142
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019503265270358802,
+      "loss": 1.785,
+      "step": 2143
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019502402997238945,
+      "loss": 1.8364,
+      "step": 2144
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019501539995460333,
+      "loss": 1.0942,
+      "step": 2145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019500676265089145,
+      "loss": 1.2624,
+      "step": 2146
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019499811806191608,
+      "loss": 2.2839,
+      "step": 2147
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019498946618834014,
+      "loss": 1.808,
+      "step": 2148
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019498080703082707,
+      "loss": 1.3153,
+      "step": 2149
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019497214059004087,
+      "loss": 1.6356,
+      "step": 2150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019496346686664605,
+      "loss": 2.0141,
+      "step": 2151
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001949547858613078,
+      "loss": 1.5536,
+      "step": 2152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019494609757469172,
+      "loss": 1.9646,
+      "step": 2153
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019493740200746408,
+      "loss": 2.1307,
+      "step": 2154
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019492869916029165,
+      "loss": 1.7233,
+      "step": 2155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001949199890338418,
+      "loss": 1.6259,
+      "step": 2156
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001949112716287824,
+      "loss": 0.8872,
+      "step": 2157
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019490254694578196,
+      "loss": 1.554,
+      "step": 2158
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019489381498550946,
+      "loss": 1.5639,
+      "step": 2159
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001948850757486345,
+      "loss": 1.5196,
+      "step": 2160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019487632923582724,
+      "loss": 1.4447,
+      "step": 2161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019486757544775834,
+      "loss": 1.4685,
+      "step": 2162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019485881438509908,
+      "loss": 1.3722,
+      "step": 2163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019485004604852124,
+      "loss": 1.7109,
+      "step": 2164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019484127043869722,
+      "loss": 1.2406,
+      "step": 2165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019483248755629992,
+      "loss": 2.1524,
+      "step": 2166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019482369740200285,
+      "loss": 1.7317,
+      "step": 2167
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019481489997648005,
+      "loss": 1.6452,
+      "step": 2168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001948060952804061,
+      "loss": 0.5222,
+      "step": 2169
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001947972833144562,
+      "loss": 1.515,
+      "step": 2170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019478846407930598,
+      "loss": 1.5285,
+      "step": 2171
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001947796375756318,
+      "loss": 1.5075,
+      "step": 2172
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019477080380411046,
+      "loss": 1.6316,
+      "step": 2173
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019476196276541934,
+      "loss": 1.759,
+      "step": 2174
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019475311446023638,
+      "loss": 1.4368,
+      "step": 2175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001947442588892401,
+      "loss": 1.7943,
+      "step": 2176
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019473539605310955,
+      "loss": 1.8598,
+      "step": 2177
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019472652595252434,
+      "loss": 1.5524,
+      "step": 2178
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019471764858816466,
+      "loss": 1.5472,
+      "step": 2179
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001947087639607112,
+      "loss": 1.4984,
+      "step": 2180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001946998720708453,
+      "loss": 1.9134,
+      "step": 2181
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019469097291924877,
+      "loss": 1.633,
+      "step": 2182
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019468206650660402,
+      "loss": 1.563,
+      "step": 2183
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000194673152833594,
+      "loss": 1.8615,
+      "step": 2184
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019466423190090223,
+      "loss": 1.4878,
+      "step": 2185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001946553037092128,
+      "loss": 1.5551,
+      "step": 2186
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001946463682592103,
+      "loss": 1.8377,
+      "step": 2187
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019463742555157992,
+      "loss": 1.3601,
+      "step": 2188
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001946284755870074,
+      "loss": 1.4778,
+      "step": 2189
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001946195183661791,
+      "loss": 1.6737,
+      "step": 2190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001946105538897818,
+      "loss": 1.7211,
+      "step": 2191
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001946015821585029,
+      "loss": 1.7799,
+      "step": 2192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019459260317303043,
+      "loss": 1.2559,
+      "step": 2193
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019458361693405288,
+      "loss": 1.7218,
+      "step": 2194
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001945746234422593,
+      "loss": 1.651,
+      "step": 2195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001945656226983394,
+      "loss": 1.056,
+      "step": 2196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001945566147029833,
+      "loss": 1.6413,
+      "step": 2197
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019454759945688178,
+      "loss": 1.0271,
+      "step": 2198
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019453857696072612,
+      "loss": 1.8076,
+      "step": 2199
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001945295472152082,
+      "loss": 1.5011,
+      "step": 2200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019452051022102044,
+      "loss": 1.5145,
+      "step": 2201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019451146597885578,
+      "loss": 1.1116,
+      "step": 2202
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001945024144894078,
+      "loss": 1.869,
+      "step": 2203
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019449335575337048,
+      "loss": 1.397,
+      "step": 2204
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019448428977143856,
+      "loss": 1.5953,
+      "step": 2205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001944752165443072,
+      "loss": 1.4058,
+      "step": 2206
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019446613607267214,
+      "loss": 1.7525,
+      "step": 2207
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019445704835722972,
+      "loss": 1.4754,
+      "step": 2208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019444795339867673,
+      "loss": 1.1795,
+      "step": 2209
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019443885119771066,
+      "loss": 1.6787,
+      "step": 2210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019442974175502942,
+      "loss": 1.1549,
+      "step": 2211
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019442062507133155,
+      "loss": 1.5038,
+      "step": 2212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019441150114731616,
+      "loss": 1.6936,
+      "step": 2213
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001944023699836829,
+      "loss": 2.0193,
+      "step": 2214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001943932315811319,
+      "loss": 1.3189,
+      "step": 2215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019438408594036393,
+      "loss": 1.6679,
+      "step": 2216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019437493306208032,
+      "loss": 1.5461,
+      "step": 2217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001943657729469829,
+      "loss": 1.7044,
+      "step": 2218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001943566055957741,
+      "loss": 1.0625,
+      "step": 2219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019434743100915685,
+      "loss": 1.8124,
+      "step": 2220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001943382491878347,
+      "loss": 1.797,
+      "step": 2221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019432906013251177,
+      "loss": 1.5339,
+      "step": 2222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019431986384389256,
+      "loss": 1.2317,
+      "step": 2223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019431066032268239,
+      "loss": 1.947,
+      "step": 2224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019430144956958693,
+      "loss": 1.8194,
+      "step": 2225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019429223158531252,
+      "loss": 1.8817,
+      "step": 2226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019428300637056595,
+      "loss": 1.8439,
+      "step": 2227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001942737739260547,
+      "loss": 1.3916,
+      "step": 2228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019426453425248663,
+      "loss": 1.6301,
+      "step": 2229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019425528735057032,
+      "loss": 1.1995,
+      "step": 2230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019424603322101486,
+      "loss": 1.6648,
+      "step": 2231
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019423677186452979,
+      "loss": 1.4955,
+      "step": 2232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019422750328182534,
+      "loss": 1.2873,
+      "step": 2233
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019421822747361223,
+      "loss": 1.2299,
+      "step": 2234
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001942089444406017,
+      "loss": 0.9871,
+      "step": 2235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001941996541835057,
+      "loss": 2.4646,
+      "step": 2236
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001941903567030365,
+      "loss": 2.1204,
+      "step": 2237
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019418105199990712,
+      "loss": 0.9025,
+      "step": 2238
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019417174007483102,
+      "loss": 1.475,
+      "step": 2239
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019416242092852226,
+      "loss": 1.7494,
+      "step": 2240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019415309456169544,
+      "loss": 1.6914,
+      "step": 2241
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019414376097506576,
+      "loss": 1.6947,
+      "step": 2242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019413442016934888,
+      "loss": 1.5869,
+      "step": 2243
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019412507214526114,
+      "loss": 1.7743,
+      "step": 2244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019411571690351926,
+      "loss": 1.67,
+      "step": 2245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019410635444484065,
+      "loss": 1.3331,
+      "step": 2246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001940969847699433,
+      "loss": 1.7788,
+      "step": 2247
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019408760787954562,
+      "loss": 1.7675,
+      "step": 2248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019407822377436672,
+      "loss": 1.3736,
+      "step": 2249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019406883245512608,
+      "loss": 1.7085,
+      "step": 2250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019405943392254393,
+      "loss": 1.8305,
+      "step": 2251
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001940500281773409,
+      "loss": 1.5688,
+      "step": 2252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001940406152202383,
+      "loss": 1.2134,
+      "step": 2253
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019403119505195787,
+      "loss": 1.5514,
+      "step": 2254
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019402176767322203,
+      "loss": 1.5071,
+      "step": 2255
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019401233308475362,
+      "loss": 1.562,
+      "step": 2256
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019400289128727614,
+      "loss": 1.4949,
+      "step": 2257
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019399344228151355,
+      "loss": 1.5415,
+      "step": 2258
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001939839860681905,
+      "loss": 1.6693,
+      "step": 2259
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019397452264803203,
+      "loss": 1.8231,
+      "step": 2260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019396505202176384,
+      "loss": 1.8541,
+      "step": 2261
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019395557419011217,
+      "loss": 1.7049,
+      "step": 2262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019394608915380374,
+      "loss": 1.2788,
+      "step": 2263
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019393659691356593,
+      "loss": 1.6723,
+      "step": 2264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001939270974701266,
+      "loss": 1.6683,
+      "step": 2265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001939175908242142,
+      "loss": 2.1625,
+      "step": 2266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019390807697655768,
+      "loss": 1.5376,
+      "step": 2267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001938985559278866,
+      "loss": 1.9433,
+      "step": 2268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019388902767893105,
+      "loss": 1.3466,
+      "step": 2269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019387949223042164,
+      "loss": 1.2764,
+      "step": 2270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019386994958308957,
+      "loss": 1.1634,
+      "step": 2271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019386039973766662,
+      "loss": 1.8483,
+      "step": 2272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001938508426948851,
+      "loss": 1.8009,
+      "step": 2273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019384127845547777,
+      "loss": 1.72,
+      "step": 2274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019383170702017813,
+      "loss": 1.6083,
+      "step": 2275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019382212838972007,
+      "loss": 1.4681,
+      "step": 2276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001938125425648381,
+      "loss": 1.5737,
+      "step": 2277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001938029495462673,
+      "loss": 2.3404,
+      "step": 2278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019379334933474328,
+      "loss": 1.1623,
+      "step": 2279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019378374193100215,
+      "loss": 1.3319,
+      "step": 2280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019377412733578066,
+      "loss": 1.4872,
+      "step": 2281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019376450554981612,
+      "loss": 2.175,
+      "step": 2282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001937548765738462,
+      "loss": 1.9504,
+      "step": 2283
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001937452404086094,
+      "loss": 1.642,
+      "step": 2284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019373559705484464,
+      "loss": 0.3898,
+      "step": 2285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019372594651329125,
+      "loss": 1.6796,
+      "step": 2286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019371628878468936,
+      "loss": 1.972,
+      "step": 2287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019370662386977955,
+      "loss": 1.8041,
+      "step": 2288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001936969517693029,
+      "loss": 1.4932,
+      "step": 2289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019368727248400105,
+      "loss": 1.8732,
+      "step": 2290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019367758601461626,
+      "loss": 1.721,
+      "step": 2291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019366789236189132,
+      "loss": 1.5823,
+      "step": 2292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019365819152656954,
+      "loss": 1.9663,
+      "step": 2293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001936484835093948,
+      "loss": 2.0636,
+      "step": 2294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019363876831111147,
+      "loss": 1.6215,
+      "step": 2295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001936290459324646,
+      "loss": 1.1877,
+      "step": 2296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019361931637419972,
+      "loss": 1.6883,
+      "step": 2297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019360957963706284,
+      "loss": 1.4124,
+      "step": 2298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019359983572180063,
+      "loss": 1.4949,
+      "step": 2299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019359008462916026,
+      "loss": 1.258,
+      "step": 2300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019358032635988947,
+      "loss": 1.4602,
+      "step": 2301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019357056091473654,
+      "loss": 1.4228,
+      "step": 2302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019356078829445026,
+      "loss": 1.4326,
+      "step": 2303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019355100849978007,
+      "loss": 1.339,
+      "step": 2304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019354122153147584,
+      "loss": 2.0782,
+      "step": 2305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019353142739028808,
+      "loss": 1.5998,
+      "step": 2306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019352162607696785,
+      "loss": 1.8443,
+      "step": 2307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001935118175922667,
+      "loss": 1.3186,
+      "step": 2308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019350200193693671,
+      "loss": 2.1857,
+      "step": 2309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019349217911173063,
+      "loss": 0.7936,
+      "step": 2310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019348234911740166,
+      "loss": 1.4768,
+      "step": 2311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001934725119547036,
+      "loss": 1.3258,
+      "step": 2312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019346266762439073,
+      "loss": 1.2781,
+      "step": 2313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000193452816127218,
+      "loss": 1.7786,
+      "step": 2314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019344295746394078,
+      "loss": 1.4773,
+      "step": 2315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001934330916353151,
+      "loss": 1.3018,
+      "step": 2316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001934232186420974,
+      "loss": 1.9882,
+      "step": 2317
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019341333848504487,
+      "loss": 1.472,
+      "step": 2318
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019340345116491503,
+      "loss": 1.3952,
+      "step": 2319
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001933935566824661,
+      "loss": 1.9662,
+      "step": 2320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019338365503845681,
+      "loss": 2.3686,
+      "step": 2321
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019337374623364646,
+      "loss": 1.4724,
+      "step": 2322
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001933638302687948,
+      "loss": 1.7731,
+      "step": 2323
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019335390714466226,
+      "loss": 1.8344,
+      "step": 2324
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019334397686200972,
+      "loss": 0.7496,
+      "step": 2325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001933340394215987,
+      "loss": 1.5057,
+      "step": 2326
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019332409482419115,
+      "loss": 2.2039,
+      "step": 2327
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019331414307054972,
+      "loss": 1.7641,
+      "step": 2328
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019330418416143745,
+      "loss": 1.3838,
+      "step": 2329
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019329421809761806,
+      "loss": 1.8904,
+      "step": 2330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001932842448798557,
+      "loss": 1.8777,
+      "step": 2331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001932742645089152,
+      "loss": 1.9787,
+      "step": 2332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019326427698556183,
+      "loss": 1.6425,
+      "step": 2333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019325428231056147,
+      "loss": 1.1474,
+      "step": 2334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001932442804846805,
+      "loss": 1.9078,
+      "step": 2335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001932342715086859,
+      "loss": 1.8349,
+      "step": 2336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019322425538334517,
+      "loss": 1.7129,
+      "step": 2337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019321423210942636,
+      "loss": 1.5988,
+      "step": 2338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019320420168769803,
+      "loss": 1.5256,
+      "step": 2339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019319416411892938,
+      "loss": 1.1879,
+      "step": 2340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001931841194038901,
+      "loss": 1.4881,
+      "step": 2341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019317406754335045,
+      "loss": 1.621,
+      "step": 2342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019316400853808114,
+      "loss": 1.7688,
+      "step": 2343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001931539423888536,
+      "loss": 1.7503,
+      "step": 2344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019314386909643968,
+      "loss": 1.0693,
+      "step": 2345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001931337886616118,
+      "loss": 1.4904,
+      "step": 2346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000193123701085143,
+      "loss": 1.3415,
+      "step": 2347
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019311360636780673,
+      "loss": 0.988,
+      "step": 2348
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019310350451037716,
+      "loss": 1.8869,
+      "step": 2349
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019309339551362882,
+      "loss": 1.5737,
+      "step": 2350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019308327937833693,
+      "loss": 1.2678,
+      "step": 2351
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019307315610527726,
+      "loss": 2.1658,
+      "step": 2352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000193063025695226,
+      "loss": 1.1554,
+      "step": 2353
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019305288814895998,
+      "loss": 1.5615,
+      "step": 2354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001930427434672566,
+      "loss": 1.1446,
+      "step": 2355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019303259165089371,
+      "loss": 1.7365,
+      "step": 2356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019302243270064983,
+      "loss": 1.8986,
+      "step": 2357
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019301226661730395,
+      "loss": 1.2228,
+      "step": 2358
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001930020934016356,
+      "loss": 1.6165,
+      "step": 2359
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019299191305442487,
+      "loss": 1.7476,
+      "step": 2360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019298172557645244,
+      "loss": 1.838,
+      "step": 2361
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019297153096849948,
+      "loss": 1.4386,
+      "step": 2362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001929613292313477,
+      "loss": 1.4113,
+      "step": 2363
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019295112036577945,
+      "loss": 1.5931,
+      "step": 2364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001929409043725775,
+      "loss": 1.5334,
+      "step": 2365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019293068125252527,
+      "loss": 1.5086,
+      "step": 2366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019292045100640666,
+      "loss": 1.6026,
+      "step": 2367
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019291021363500614,
+      "loss": 1.1397,
+      "step": 2368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019289996913910876,
+      "loss": 1.9568,
+      "step": 2369
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019288971751950005,
+      "loss": 1.5322,
+      "step": 2370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019287945877696615,
+      "loss": 1.5284,
+      "step": 2371
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019286919291229367,
+      "loss": 0.599,
+      "step": 2372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001928589199262698,
+      "loss": 1.968,
+      "step": 2373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001928486398196824,
+      "loss": 1.8683,
+      "step": 2374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019283835259331964,
+      "loss": 1.0529,
+      "step": 2375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019282805824797042,
+      "loss": 1.6412,
+      "step": 2376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019281775678442413,
+      "loss": 1.9235,
+      "step": 2377
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019280744820347067,
+      "loss": 1.8618,
+      "step": 2378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019279713250590054,
+      "loss": 1.7451,
+      "step": 2379
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019278680969250477,
+      "loss": 2.1008,
+      "step": 2380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019277647976407493,
+      "loss": 2.2865,
+      "step": 2381
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019276614272140309,
+      "loss": 1.8492,
+      "step": 2382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019275579856528198,
+      "loss": 1.9044,
+      "step": 2383
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019274544729650477,
+      "loss": 2.0173,
+      "step": 2384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019273508891586518,
+      "loss": 1.4285,
+      "step": 2385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019272472342415754,
+      "loss": 1.547,
+      "step": 2386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001927143508221767,
+      "loss": 1.6938,
+      "step": 2387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019270397111071803,
+      "loss": 1.4728,
+      "step": 2388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001926935842905775,
+      "loss": 1.4736,
+      "step": 2389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019268319036255149,
+      "loss": 1.4399,
+      "step": 2390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019267278932743714,
+      "loss": 1.4485,
+      "step": 2391
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019266238118603192,
+      "loss": 1.5546,
+      "step": 2392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019265196593913403,
+      "loss": 1.8248,
+      "step": 2393
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019264154358754207,
+      "loss": 1.8946,
+      "step": 2394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019263111413205523,
+      "loss": 1.6225,
+      "step": 2395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001926206775734733,
+      "loss": 1.8611,
+      "step": 2396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019261023391259658,
+      "loss": 1.5352,
+      "step": 2397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019259978315022585,
+      "loss": 1.3238,
+      "step": 2398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001925893252871625,
+      "loss": 1.9289,
+      "step": 2399
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019257886032420852,
+      "loss": 1.7807,
+      "step": 2400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001925683882621663,
+      "loss": 1.8652,
+      "step": 2401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019255790910183889,
+      "loss": 1.645,
+      "step": 2402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019254742284402986,
+      "loss": 1.7194,
+      "step": 2403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001925369294895433,
+      "loss": 1.2985,
+      "step": 2404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019252642903918384,
+      "loss": 0.7506,
+      "step": 2405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001925159214937567,
+      "loss": 1.6547,
+      "step": 2406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019250540685406757,
+      "loss": 1.6457,
+      "step": 2407
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019249488512092277,
+      "loss": 1.7817,
+      "step": 2408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001924843562951291,
+      "loss": 1.2737,
+      "step": 2409
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019247382037749396,
+      "loss": 1.6105,
+      "step": 2410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019246327736882523,
+      "loss": 1.836,
+      "step": 2411
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019245272726993136,
+      "loss": 1.8091,
+      "step": 2412
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019244217008162137,
+      "loss": 1.2218,
+      "step": 2413
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019243160580470478,
+      "loss": 1.654,
+      "step": 2414
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001924210344399917,
+      "loss": 1.5592,
+      "step": 2415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019241045598829276,
+      "loss": 1.7212,
+      "step": 2416
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019239987045041907,
+      "loss": 1.6204,
+      "step": 2417
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019238927782718238,
+      "loss": 1.3097,
+      "step": 2418
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019237867811939502,
+      "loss": 2.2082,
+      "step": 2419
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001923680713278697,
+      "loss": 1.6311,
+      "step": 2420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019235745745341978,
+      "loss": 1.5304,
+      "step": 2421
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019234683649685917,
+      "loss": 1.9842,
+      "step": 2422
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019233620845900232,
+      "loss": 1.3998,
+      "step": 2423
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019232557334066414,
+      "loss": 1.6481,
+      "step": 2424
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001923149311426602,
+      "loss": 1.6044,
+      "step": 2425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019230428186580657,
+      "loss": 2.0892,
+      "step": 2426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001922936255109198,
+      "loss": 2.0086,
+      "step": 2427
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019228296207881706,
+      "loss": 1.3303,
+      "step": 2428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019227229157031605,
+      "loss": 1.7058,
+      "step": 2429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019226161398623498,
+      "loss": 1.6721,
+      "step": 2430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001922509293273927,
+      "loss": 1.3815,
+      "step": 2431
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001922402375946084,
+      "loss": 1.2731,
+      "step": 2432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000192229538788702,
+      "loss": 1.4416,
+      "step": 2433
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001922188329104939,
+      "loss": 1.3437,
+      "step": 2434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019220811996080507,
+      "loss": 1.6891,
+      "step": 2435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000192197399940457,
+      "loss": 1.1461,
+      "step": 2436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001921866728502716,
+      "loss": 1.6005,
+      "step": 2437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019217593869107156,
+      "loss": 1.6003,
+      "step": 2438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019216519746368,
+      "loss": 2.1611,
+      "step": 2439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019215444916892053,
+      "loss": 1.9937,
+      "step": 2440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019214369380761732,
+      "loss": 1.8223,
+      "step": 2441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019213293138059512,
+      "loss": 1.6653,
+      "step": 2442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019212216188867928,
+      "loss": 1.976,
+      "step": 2443
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001921113853326955,
+      "loss": 1.7402,
+      "step": 2444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019210060171347024,
+      "loss": 1.4861,
+      "step": 2445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019208981103183037,
+      "loss": 1.4884,
+      "step": 2446
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019207901328860335,
+      "loss": 1.8463,
+      "step": 2447
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019206820848461712,
+      "loss": 1.4084,
+      "step": 2448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001920573966207003,
+      "loss": 0.9806,
+      "step": 2449
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019204657769768187,
+      "loss": 1.6916,
+      "step": 2450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019203575171639149,
+      "loss": 2.0359,
+      "step": 2451
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001920249186776593,
+      "loss": 1.8773,
+      "step": 2452
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019201407858231598,
+      "loss": 1.4405,
+      "step": 2453
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001920032314311928,
+      "loss": 1.9152,
+      "step": 2454
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001919923772251215,
+      "loss": 1.9665,
+      "step": 2455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001919815159649344,
+      "loss": 1.8071,
+      "step": 2456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001919706476514644,
+      "loss": 1.2337,
+      "step": 2457
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019195977228554485,
+      "loss": 1.3352,
+      "step": 2458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001919488898680097,
+      "loss": 1.641,
+      "step": 2459
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019193800039969347,
+      "loss": 1.6492,
+      "step": 2460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019192710388143112,
+      "loss": 1.0669,
+      "step": 2461
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019191620031405827,
+      "loss": 1.6445,
+      "step": 2462
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019190528969841096,
+      "loss": 1.2617,
+      "step": 2463
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001918943720353259,
+      "loss": 1.2137,
+      "step": 2464
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001918834473256402,
+      "loss": 1.6524,
+      "step": 2465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019187251557019165,
+      "loss": 1.1711,
+      "step": 2466
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019186157676981847,
+      "loss": 1.2994,
+      "step": 2467
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001918506309253595,
+      "loss": 1.313,
+      "step": 2468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019183967803765404,
+      "loss": 1.4263,
+      "step": 2469
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019182871810754202,
+      "loss": 1.4072,
+      "step": 2470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019181775113586382,
+      "loss": 1.709,
+      "step": 2471
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019180677712346046,
+      "loss": 1.4585,
+      "step": 2472
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001917957960711734,
+      "loss": 2.0383,
+      "step": 2473
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019178480797984467,
+      "loss": 0.7558,
+      "step": 2474
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001917738128503169,
+      "loss": 2.0341,
+      "step": 2475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019176281068343317,
+      "loss": 2.2036,
+      "step": 2476
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001917518014800372,
+      "loss": 1.5169,
+      "step": 2477
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019174078524097313,
+      "loss": 1.7745,
+      "step": 2478
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019172976196708575,
+      "loss": 1.73,
+      "step": 2479
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001917187316592203,
+      "loss": 1.9008,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019170769431822265,
+      "loss": 1.9448,
+      "step": 2481
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001916966499449391,
+      "loss": 1.8154,
+      "step": 2482
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019168559854021658,
+      "loss": 1.2381,
+      "step": 2483
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019167454010490256,
+      "loss": 1.6959,
+      "step": 2484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019166347463984496,
+      "loss": 2.0542,
+      "step": 2485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019165240214589233,
+      "loss": 1.8298,
+      "step": 2486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019164132262389374,
+      "loss": 1.5436,
+      "step": 2487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019163023607469872,
+      "loss": 1.3455,
+      "step": 2488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001916191424991575,
+      "loss": 1.1445,
+      "step": 2489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019160804189812065,
+      "loss": 1.4016,
+      "step": 2490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019159693427243946,
+      "loss": 2.1878,
+      "step": 2491
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001915858196229656,
+      "loss": 1.8452,
+      "step": 2492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019157469795055149,
+      "loss": 1.4137,
+      "step": 2493
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019156356925604983,
+      "loss": 1.8887,
+      "step": 2494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019155243354031405,
+      "loss": 1.6169,
+      "step": 2495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019154129080419804,
+      "loss": 1.9306,
+      "step": 2496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001915301410485562,
+      "loss": 1.7363,
+      "step": 2497
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019151898427424358,
+      "loss": 1.6751,
+      "step": 2498
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019150782048211565,
+      "loss": 1.9636,
+      "step": 2499
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019149664967302847,
+      "loss": 1.9584,
+      "step": 2500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019148547184783866,
+      "loss": 1.5145,
+      "step": 2501
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019147428700740335,
+      "loss": 1.6218,
+      "step": 2502
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001914630951525802,
+      "loss": 1.5028,
+      "step": 2503
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019145189628422743,
+      "loss": 1.4802,
+      "step": 2504
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019144069040320373,
+      "loss": 1.5523,
+      "step": 2505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019142947751036844,
+      "loss": 1.4168,
+      "step": 2506
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019141825760658141,
+      "loss": 1.6243,
+      "step": 2507
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019140703069270293,
+      "loss": 1.6107,
+      "step": 2508
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019139579676959393,
+      "loss": 1.6775,
+      "step": 2509
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019138455583811583,
+      "loss": 1.4224,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019137330789913062,
+      "loss": 1.559,
+      "step": 2511
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001913620529535008,
+      "loss": 1.9611,
+      "step": 2512
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019135079100208945,
+      "loss": 1.5585,
+      "step": 2513
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019133952204576006,
+      "loss": 1.4638,
+      "step": 2514
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019132824608537687,
+      "loss": 1.2377,
+      "step": 2515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019131696312180445,
+      "loss": 1.543,
+      "step": 2516
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019130567315590804,
+      "loss": 1.6833,
+      "step": 2517
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019129437618855335,
+      "loss": 1.617,
+      "step": 2518
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019128307222060666,
+      "loss": 1.73,
+      "step": 2519
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019127176125293476,
+      "loss": 1.9117,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019126044328640501,
+      "loss": 1.6171,
+      "step": 2521
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001912491183218853,
+      "loss": 1.6271,
+      "step": 2522
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000191237786360244,
+      "loss": 1.0977,
+      "step": 2523
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019122644740235013,
+      "loss": 1.5447,
+      "step": 2524
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001912151014490731,
+      "loss": 1.2306,
+      "step": 2525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000191203748501283,
+      "loss": 1.537,
+      "step": 2526
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019119238855985033,
+      "loss": 1.772,
+      "step": 2527
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019118102162564627,
+      "loss": 1.5522,
+      "step": 2528
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019116964769954237,
+      "loss": 1.8142,
+      "step": 2529
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019115826678241083,
+      "loss": 1.6852,
+      "step": 2530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001911468788751244,
+      "loss": 1.4172,
+      "step": 2531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019113548397855628,
+      "loss": 1.5824,
+      "step": 2532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001911240820935802,
+      "loss": 1.2954,
+      "step": 2533
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001911126732210706,
+      "loss": 1.2486,
+      "step": 2534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019110125736190221,
+      "loss": 1.58,
+      "step": 2535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019108983451695047,
+      "loss": 1.4898,
+      "step": 2536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001910784046870913,
+      "loss": 1.563,
+      "step": 2537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019106696787320114,
+      "loss": 1.6056,
+      "step": 2538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019105552407615702,
+      "loss": 1.2975,
+      "step": 2539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019104407329683642,
+      "loss": 1.8177,
+      "step": 2540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019103261553611743,
+      "loss": 1.3696,
+      "step": 2541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001910211507948786,
+      "loss": 1.3155,
+      "step": 2542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019100967907399917,
+      "loss": 1.665,
+      "step": 2543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001909982003743587,
+      "loss": 1.2371,
+      "step": 2544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019098671469683747,
+      "loss": 1.74,
+      "step": 2545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019097522204231617,
+      "loss": 1.1322,
+      "step": 2546
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019096372241167607,
+      "loss": 1.766,
+      "step": 2547
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019095221580579905,
+      "loss": 1.5024,
+      "step": 2548
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019094070222556737,
+      "loss": 1.6489,
+      "step": 2549
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019092918167186397,
+      "loss": 1.7003,
+      "step": 2550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019091765414557224,
+      "loss": 1.6021,
+      "step": 2551
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001909061196475761,
+      "loss": 1.6307,
+      "step": 2552
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019089457817876007,
+      "loss": 1.1737,
+      "step": 2553
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019088302974000918,
+      "loss": 1.6194,
+      "step": 2554
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019087147433220894,
+      "loss": 1.5233,
+      "step": 2555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019085991195624545,
+      "loss": 1.7539,
+      "step": 2556
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019084834261300535,
+      "loss": 1.8776,
+      "step": 2557
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019083676630337574,
+      "loss": 1.2559,
+      "step": 2558
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001908251830282444,
+      "loss": 1.3928,
+      "step": 2559
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019081359278849948,
+      "loss": 1.9546,
+      "step": 2560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019080199558502975,
+      "loss": 1.5111,
+      "step": 2561
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019079039141872453,
+      "loss": 1.8164,
+      "step": 2562
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019077878029047365,
+      "loss": 1.9133,
+      "step": 2563
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001907671622011674,
+      "loss": 1.1215,
+      "step": 2564
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019075553715169674,
+      "loss": 1.4118,
+      "step": 2565
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019074390514295305,
+      "loss": 0.2893,
+      "step": 2566
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019073226617582836,
+      "loss": 1.9943,
+      "step": 2567
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001907206202512151,
+      "loss": 1.7051,
+      "step": 2568
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001907089673700063,
+      "loss": 1.3211,
+      "step": 2569
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019069730753309557,
+      "loss": 1.7523,
+      "step": 2570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019068564074137693,
+      "loss": 1.9011,
+      "step": 2571
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001906739669957451,
+      "loss": 1.8173,
+      "step": 2572
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019066228629709516,
+      "loss": 1.7542,
+      "step": 2573
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001906505986463228,
+      "loss": 2.0536,
+      "step": 2574
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019063890404432435,
+      "loss": 1.4976,
+      "step": 2575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019062720249199644,
+      "loss": 2.2601,
+      "step": 2576
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019061549399023643,
+      "loss": 1.4672,
+      "step": 2577
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019060377853994217,
+      "loss": 1.64,
+      "step": 2578
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019059205614201197,
+      "loss": 1.9782,
+      "step": 2579
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019058032679734473,
+      "loss": 1.4795,
+      "step": 2580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001905685905068399,
+      "loss": 1.3898,
+      "step": 2581
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001905568472713974,
+      "loss": 1.1585,
+      "step": 2582
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019054509709191776,
+      "loss": 1.7416,
+      "step": 2583
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019053333996930195,
+      "loss": 1.9598,
+      "step": 2584
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019052157590445157,
+      "loss": 1.3237,
+      "step": 2585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001905098048982687,
+      "loss": 1.2754,
+      "step": 2586
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019049802695165593,
+      "loss": 1.6008,
+      "step": 2587
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019048624206551644,
+      "loss": 1.3081,
+      "step": 2588
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001904744502407539,
+      "loss": 1.3571,
+      "step": 2589
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019046265147827253,
+      "loss": 2.2739,
+      "step": 2590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019045084577897707,
+      "loss": 1.4888,
+      "step": 2591
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019043903314377284,
+      "loss": 1.5152,
+      "step": 2592
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019042721357356556,
+      "loss": 2.1841,
+      "step": 2593
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019041538706926166,
+      "loss": 1.6987,
+      "step": 2594
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019040355363176797,
+      "loss": 1.6127,
+      "step": 2595
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001903917132619919,
+      "loss": 1.689,
+      "step": 2596
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019037986596084143,
+      "loss": 1.6833,
+      "step": 2597
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019036801172922496,
+      "loss": 1.8892,
+      "step": 2598
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019035615056805155,
+      "loss": 1.9108,
+      "step": 2599
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019034428247823067,
+      "loss": 1.3248,
+      "step": 2600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019033240746067247,
+      "loss": 1.6498,
+      "step": 2601
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019032052551628748,
+      "loss": 1.8038,
+      "step": 2602
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001903086366459868,
+      "loss": 1.3563,
+      "step": 2603
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001902967408506822,
+      "loss": 1.8673,
+      "step": 2604
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019028483813128574,
+      "loss": 1.1542,
+      "step": 2605
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019027292848871022,
+      "loss": 1.7244,
+      "step": 2606
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019026101192386885,
+      "loss": 1.4883,
+      "step": 2607
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019024908843767542,
+      "loss": 1.9069,
+      "step": 2608
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019023715803104425,
+      "loss": 1.3015,
+      "step": 2609
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019022522070489018,
+      "loss": 1.731,
+      "step": 2610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019021327646012858,
+      "loss": 1.7558,
+      "step": 2611
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019020132529767536,
+      "loss": 1.7204,
+      "step": 2612
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019018936721844693,
+      "loss": 1.6526,
+      "step": 2613
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019017740222336026,
+      "loss": 1.6246,
+      "step": 2614
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019016543031333288,
+      "loss": 1.575,
+      "step": 2615
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019015345148928276,
+      "loss": 0.8769,
+      "step": 2616
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001901414657521285,
+      "loss": 1.5131,
+      "step": 2617
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019012947310278916,
+      "loss": 1.9858,
+      "step": 2618
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019011747354218437,
+      "loss": 1.4725,
+      "step": 2619
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019010546707123424,
+      "loss": 2.0267,
+      "step": 2620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019009345369085952,
+      "loss": 1.1826,
+      "step": 2621
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019008143340198133,
+      "loss": 1.4859,
+      "step": 2622
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019006940620552146,
+      "loss": 1.4286,
+      "step": 2623
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001900573721024021,
+      "loss": 1.6783,
+      "step": 2624
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019004533109354618,
+      "loss": 1.6176,
+      "step": 2625
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001900332831798769,
+      "loss": 1.7043,
+      "step": 2626
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001900212283623181,
+      "loss": 1.7857,
+      "step": 2627
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019000916664179424,
+      "loss": 1.5641,
+      "step": 2628
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018999709801923023,
+      "loss": 1.7603,
+      "step": 2629
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018998502249555147,
+      "loss": 1.5563,
+      "step": 2630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018997294007168397,
+      "loss": 1.7611,
+      "step": 2631
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018996085074855415,
+      "loss": 1.2974,
+      "step": 2632
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018994875452708914,
+      "loss": 1.6947,
+      "step": 2633
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001899366514082164,
+      "loss": 1.7031,
+      "step": 2634
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018992454139286414,
+      "loss": 2.0341,
+      "step": 2635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018991242448196082,
+      "loss": 1.5099,
+      "step": 2636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001899003006764357,
+      "loss": 1.6413,
+      "step": 2637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001898881699772184,
+      "loss": 1.8008,
+      "step": 2638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001898760323852391,
+      "loss": 1.9195,
+      "step": 2639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018986388790142865,
+      "loss": 1.3391,
+      "step": 2640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018985173652671814,
+      "loss": 1.9212,
+      "step": 2641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018983957826203946,
+      "loss": 1.1619,
+      "step": 2642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018982741310832492,
+      "loss": 1.8522,
+      "step": 2643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018981524106650733,
+      "loss": 1.4735,
+      "step": 2644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018980306213752004,
+      "loss": 1.484,
+      "step": 2645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000189790876322297,
+      "loss": 1.5203,
+      "step": 2646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018977868362177262,
+      "loss": 2.1187,
+      "step": 2647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018976648403688188,
+      "loss": 1.5664,
+      "step": 2648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018975427756856022,
+      "loss": 1.5246,
+      "step": 2649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018974206421774366,
+      "loss": 1.8517,
+      "step": 2650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018972984398536876,
+      "loss": 1.8922,
+      "step": 2651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018971761687237252,
+      "loss": 1.1422,
+      "step": 2652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018970538287969263,
+      "loss": 1.3539,
+      "step": 2653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018969314200826718,
+      "loss": 1.1837,
+      "step": 2654
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018968089425903476,
+      "loss": 1.2098,
+      "step": 2655
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018966863963293463,
+      "loss": 0.8963,
+      "step": 2656
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018965637813090646,
+      "loss": 2.0355,
+      "step": 2657
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018964410975389045,
+      "loss": 1.8091,
+      "step": 2658
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001896318345028274,
+      "loss": 1.6363,
+      "step": 2659
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018961955237865855,
+      "loss": 1.8091,
+      "step": 2660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018960726338232577,
+      "loss": 1.7989,
+      "step": 2661
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018959496751477136,
+      "loss": 1.5832,
+      "step": 2662
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018958266477693818,
+      "loss": 2.3646,
+      "step": 2663
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018957035516976967,
+      "loss": 1.2325,
+      "step": 2664
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001895580386942097,
+      "loss": 1.7292,
+      "step": 2665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001895457153512027,
+      "loss": 1.9445,
+      "step": 2666
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018953338514169374,
+      "loss": 1.4352,
+      "step": 2667
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001895210480666282,
+      "loss": 1.3504,
+      "step": 2668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018950870412695218,
+      "loss": 1.5782,
+      "step": 2669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018949635332361222,
+      "loss": 1.422,
+      "step": 2670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001894839956575554,
+      "loss": 1.7209,
+      "step": 2671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018947163112972926,
+      "loss": 1.1693,
+      "step": 2672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018945925974108208,
+      "loss": 2.0294,
+      "step": 2673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018944688149256236,
+      "loss": 1.6939,
+      "step": 2674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018943449638511933,
+      "loss": 1.3696,
+      "step": 2675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018942210441970277,
+      "loss": 1.0032,
+      "step": 2676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018940970559726283,
+      "loss": 1.432,
+      "step": 2677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018939729991875032,
+      "loss": 1.7074,
+      "step": 2678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001893848873851165,
+      "loss": 1.9988,
+      "step": 2679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018937246799731317,
+      "loss": 1.5891,
+      "step": 2680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018936004175629272,
+      "loss": 1.3625,
+      "step": 2681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018934760866300796,
+      "loss": 1.6394,
+      "step": 2682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001893351687184123,
+      "loss": 0.9994,
+      "step": 2683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018932272192345967,
+      "loss": 1.4702,
+      "step": 2684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018931026827910446,
+      "loss": 1.4774,
+      "step": 2685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001892978077863017,
+      "loss": 1.5203,
+      "step": 2686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001892853404460068,
+      "loss": 1.4555,
+      "step": 2687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018927286625917587,
+      "loss": 1.3147,
+      "step": 2688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018926038522676538,
+      "loss": 1.7938,
+      "step": 2689
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018924789734973242,
+      "loss": 1.8303,
+      "step": 2690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018923540262903455,
+      "loss": 2.1293,
+      "step": 2691
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018922290106562995,
+      "loss": 1.6429,
+      "step": 2692
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018921039266047717,
+      "loss": 1.6874,
+      "step": 2693
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018919787741453545,
+      "loss": 1.3553,
+      "step": 2694
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018918535532876446,
+      "loss": 1.147,
+      "step": 2695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001891728264041244,
+      "loss": 1.5965,
+      "step": 2696
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018916029064157598,
+      "loss": 1.7562,
+      "step": 2697
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001891477480420805,
+      "loss": 1.7297,
+      "step": 2698
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018913519860659975,
+      "loss": 1.6414,
+      "step": 2699
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018912264233609603,
+      "loss": 1.4391,
+      "step": 2700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018911007923153218,
+      "loss": 1.7342,
+      "step": 2701
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001890975092938715,
+      "loss": 1.5231,
+      "step": 2702
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018908493252407796,
+      "loss": 1.4047,
+      "step": 2703
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001890723489231159,
+      "loss": 1.4594,
+      "step": 2704
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001890597584919503,
+      "loss": 1.0312,
+      "step": 2705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018904716123154658,
+      "loss": 1.551,
+      "step": 2706
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018903455714287074,
+      "loss": 1.2647,
+      "step": 2707
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018902194622688923,
+      "loss": 2.0637,
+      "step": 2708
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018900932848456913,
+      "loss": 2.0576,
+      "step": 2709
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018899670391687797,
+      "loss": 1.6261,
+      "step": 2710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018898407252478382,
+      "loss": 1.5555,
+      "step": 2711
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001889714343092553,
+      "loss": 1.5791,
+      "step": 2712
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018895878927126147,
+      "loss": 1.5798,
+      "step": 2713
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018894613741177204,
+      "loss": 1.7114,
+      "step": 2714
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001889334787317571,
+      "loss": 1.3672,
+      "step": 2715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018892081323218737,
+      "loss": 1.0889,
+      "step": 2716
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018890814091403408,
+      "loss": 1.5039,
+      "step": 2717
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018889546177826896,
+      "loss": 1.2917,
+      "step": 2718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018888277582586422,
+      "loss": 1.852,
+      "step": 2719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018887008305779268,
+      "loss": 1.6612,
+      "step": 2720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018885738347502762,
+      "loss": 1.7464,
+      "step": 2721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001888446770785429,
+      "loss": 1.7482,
+      "step": 2722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018883196386931282,
+      "loss": 1.4033,
+      "step": 2723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001888192438483123,
+      "loss": 1.5658,
+      "step": 2724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018880651701651667,
+      "loss": 1.4862,
+      "step": 2725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018879378337490187,
+      "loss": 1.9038,
+      "step": 2726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018878104292444433,
+      "loss": 1.5109,
+      "step": 2727
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018876829566612103,
+      "loss": 1.351,
+      "step": 2728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018875554160090942,
+      "loss": 1.8432,
+      "step": 2729
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018874278072978756,
+      "loss": 2.2455,
+      "step": 2730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018873001305373387,
+      "loss": 1.9589,
+      "step": 2731
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018871723857372748,
+      "loss": 1.4599,
+      "step": 2732
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018870445729074793,
+      "loss": 2.2941,
+      "step": 2733
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018869166920577527,
+      "loss": 1.8055,
+      "step": 2734
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001886788743197902,
+      "loss": 1.6662,
+      "step": 2735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018866607263377376,
+      "loss": 1.6144,
+      "step": 2736
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018865326414870765,
+      "loss": 1.7468,
+      "step": 2737
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018864044886557406,
+      "loss": 1.5721,
+      "step": 2738
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018862762678535563,
+      "loss": 1.7978,
+      "step": 2739
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018861479790903562,
+      "loss": 1.6315,
+      "step": 2740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018860196223759775,
+      "loss": 1.2803,
+      "step": 2741
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001885891197720263,
+      "loss": 1.0721,
+      "step": 2742
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018857627051330602,
+      "loss": 1.6515,
+      "step": 2743
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018856341446242222,
+      "loss": 1.883,
+      "step": 2744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018855055162036072,
+      "loss": 0.8799,
+      "step": 2745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001885376819881079,
+      "loss": 1.6254,
+      "step": 2746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018852480556665055,
+      "loss": 1.527,
+      "step": 2747
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018851192235697615,
+      "loss": 1.7622,
+      "step": 2748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001884990323600725,
+      "loss": 0.842,
+      "step": 2749
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001884861355769281,
+      "loss": 1.8189,
+      "step": 2750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018847323200853184,
+      "loss": 1.5029,
+      "step": 2751
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018846032165587326,
+      "loss": 1.211,
+      "step": 2752
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018844740451994226,
+      "loss": 1.0306,
+      "step": 2753
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018843448060172939,
+      "loss": 1.3589,
+      "step": 2754
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001884215499022257,
+      "loss": 1.5713,
+      "step": 2755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001884086124224227,
+      "loss": 1.7063,
+      "step": 2756
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018839566816331244,
+      "loss": 1.722,
+      "step": 2757
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018838271712588752,
+      "loss": 1.3651,
+      "step": 2758
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001883697593111411,
+      "loss": 1.648,
+      "step": 2759
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001883567947200667,
+      "loss": 1.8065,
+      "step": 2760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018834382335365858,
+      "loss": 1.6843,
+      "step": 2761
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018833084521291133,
+      "loss": 1.1939,
+      "step": 2762
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018831786029882013,
+      "loss": 1.6438,
+      "step": 2763
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018830486861238073,
+      "loss": 1.7182,
+      "step": 2764
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018829187015458932,
+      "loss": 1.5772,
+      "step": 2765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018827886492644262,
+      "loss": 1.1898,
+      "step": 2766
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018826585292893794,
+      "loss": 1.5991,
+      "step": 2767
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018825283416307303,
+      "loss": 1.6269,
+      "step": 2768
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001882398086298462,
+      "loss": 1.6223,
+      "step": 2769
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001882267763302563,
+      "loss": 1.442,
+      "step": 2770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001882137372653026,
+      "loss": 1.4384,
+      "step": 2771
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018820069143598499,
+      "loss": 1.8034,
+      "step": 2772
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018818763884330384,
+      "loss": 1.3056,
+      "step": 2773
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018817457948826002,
+      "loss": 1.6124,
+      "step": 2774
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000188161513371855,
+      "loss": 1.6053,
+      "step": 2775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018814844049509067,
+      "loss": 1.7558,
+      "step": 2776
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018813536085896947,
+      "loss": 1.8071,
+      "step": 2777
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018812227446449438,
+      "loss": 2.0119,
+      "step": 2778
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001881091813126689,
+      "loss": 1.7043,
+      "step": 2779
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000188096081404497,
+      "loss": 1.3891,
+      "step": 2780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018808297474098325,
+      "loss": 1.6763,
+      "step": 2781
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001880698613231326,
+      "loss": 1.6932,
+      "step": 2782
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018805674115195071,
+      "loss": 1.1663,
+      "step": 2783
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001880436142284436,
+      "loss": 1.6125,
+      "step": 2784
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018803048055361784,
+      "loss": 1.7247,
+      "step": 2785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001880173401284806,
+      "loss": 1.6559,
+      "step": 2786
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001880041929540395,
+      "loss": 1.4527,
+      "step": 2787
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018799103903130264,
+      "loss": 1.838,
+      "step": 2788
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001879778783612787,
+      "loss": 1.3826,
+      "step": 2789
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018796471094497687,
+      "loss": 1.8184,
+      "step": 2790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018795153678340687,
+      "loss": 1.6702,
+      "step": 2791
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018793835587757886,
+      "loss": 1.3019,
+      "step": 2792
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001879251682285036,
+      "loss": 1.2574,
+      "step": 2793
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018791197383719233,
+      "loss": 1.3667,
+      "step": 2794
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001878987727046569,
+      "loss": 1.2103,
+      "step": 2795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018788556483190946,
+      "loss": 1.6445,
+      "step": 2796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018787235021996285,
+      "loss": 1.6164,
+      "step": 2797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018785912886983046,
+      "loss": 1.4067,
+      "step": 2798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018784590078252603,
+      "loss": 1.7375,
+      "step": 2799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018783266595906396,
+      "loss": 1.363,
+      "step": 2800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018781942440045912,
+      "loss": 0.8963,
+      "step": 2801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018780617610772687,
+      "loss": 1.376,
+      "step": 2802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018779292108188312,
+      "loss": 1.5551,
+      "step": 2803
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018777965932394427,
+      "loss": 1.6153,
+      "step": 2804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001877663908349273,
+      "loss": 1.1697,
+      "step": 2805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018775311561584958,
+      "loss": 2.1471,
+      "step": 2806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018773983366772914,
+      "loss": 1.5931,
+      "step": 2807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018772654499158442,
+      "loss": 1.5603,
+      "step": 2808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018771324958843447,
+      "loss": 1.3529,
+      "step": 2809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018769994745929873,
+      "loss": 1.4971,
+      "step": 2810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001876866386051973,
+      "loss": 1.2379,
+      "step": 2811
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018767332302715067,
+      "loss": 2.0775,
+      "step": 2812
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001876600007261799,
+      "loss": 1.5916,
+      "step": 2813
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018764667170330655,
+      "loss": 1.8326,
+      "step": 2814
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001876333359595528,
+      "loss": 1.8408,
+      "step": 2815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018761999349594118,
+      "loss": 1.7107,
+      "step": 2816
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018760664431349483,
+      "loss": 1.808,
+      "step": 2817
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018759328841323733,
+      "loss": 1.4847,
+      "step": 2818
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018757992579619294,
+      "loss": 1.829,
+      "step": 2819
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018756655646338623,
+      "loss": 1.1967,
+      "step": 2820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018755318041584245,
+      "loss": 1.2928,
+      "step": 2821
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018753979765458723,
+      "loss": 1.3305,
+      "step": 2822
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018752640818064684,
+      "loss": 1.8855,
+      "step": 2823
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018751301199504796,
+      "loss": 1.621,
+      "step": 2824
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018749960909881787,
+      "loss": 1.4953,
+      "step": 2825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018748619949298428,
+      "loss": 0.9967,
+      "step": 2826
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001874727831785755,
+      "loss": 1.5892,
+      "step": 2827
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001874593601566203,
+      "loss": 1.8013,
+      "step": 2828
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018744593042814794,
+      "loss": 1.5697,
+      "step": 2829
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018743249399418832,
+      "loss": 1.1345,
+      "step": 2830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001874190508557717,
+      "loss": 1.0526,
+      "step": 2831
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001874056010139289,
+      "loss": 1.6357,
+      "step": 2832
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018739214446969132,
+      "loss": 1.9843,
+      "step": 2833
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018737868122409084,
+      "loss": 1.6259,
+      "step": 2834
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018736521127815978,
+      "loss": 1.6787,
+      "step": 2835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001873517346329311,
+      "loss": 1.5605,
+      "step": 2836
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018733825128943818,
+      "loss": 1.3438,
+      "step": 2837
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018732476124871494,
+      "loss": 1.858,
+      "step": 2838
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018731126451179583,
+      "loss": 1.5588,
+      "step": 2839
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001872977610797158,
+      "loss": 1.5414,
+      "step": 2840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001872842509535103,
+      "loss": 1.6114,
+      "step": 2841
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001872707341342153,
+      "loss": 1.7307,
+      "step": 2842
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018725721062286735,
+      "loss": 1.6424,
+      "step": 2843
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018724368042050336,
+      "loss": 1.4499,
+      "step": 2844
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001872301435281609,
+      "loss": 1.6068,
+      "step": 2845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018721659994687802,
+      "loss": 1.996,
+      "step": 2846
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001872030496776932,
+      "loss": 1.3927,
+      "step": 2847
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001871894927216456,
+      "loss": 1.7419,
+      "step": 2848
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018717592907977469,
+      "loss": 1.8427,
+      "step": 2849
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018716235875312055,
+      "loss": 1.3739,
+      "step": 2850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018714878174272382,
+      "loss": 1.3659,
+      "step": 2851
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001871351980496256,
+      "loss": 1.3919,
+      "step": 2852
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001871216076748675,
+      "loss": 1.4671,
+      "step": 2853
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018710801061949163,
+      "loss": 1.1134,
+      "step": 2854
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001870944068845407,
+      "loss": 1.2353,
+      "step": 2855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018708079647105776,
+      "loss": 1.4934,
+      "step": 2856
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018706717938008657,
+      "loss": 1.2103,
+      "step": 2857
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018705355561267125,
+      "loss": 2.1223,
+      "step": 2858
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018703992516985656,
+      "loss": 1.6195,
+      "step": 2859
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001870262880526876,
+      "loss": 1.7065,
+      "step": 2860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018701264426221017,
+      "loss": 1.3607,
+      "step": 2861
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001869989937994705,
+      "loss": 1.5067,
+      "step": 2862
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018698533666551526,
+      "loss": 1.0519,
+      "step": 2863
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018697167286139174,
+      "loss": 1.6306,
+      "step": 2864
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018695800238814774,
+      "loss": 1.1908,
+      "step": 2865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018694432524683143,
+      "loss": 1.7963,
+      "step": 2866
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001869306414384917,
+      "loss": 1.0767,
+      "step": 2867
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018691695096417782,
+      "loss": 1.3535,
+      "step": 2868
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001869032538249395,
+      "loss": 1.2566,
+      "step": 2869
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001868895500218272,
+      "loss": 1.5933,
+      "step": 2870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001868758395558917,
+      "loss": 1.5693,
+      "step": 2871
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018686212242818425,
+      "loss": 1.6728,
+      "step": 2872
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018684839863975685,
+      "loss": 1.2496,
+      "step": 2873
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018683466819166176,
+      "loss": 1.4394,
+      "step": 2874
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018682093108495185,
+      "loss": 1.8494,
+      "step": 2875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018680718732068058,
+      "loss": 1.714,
+      "step": 2876
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018679343689990174,
+      "loss": 1.8161,
+      "step": 2877
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001867796798236698,
+      "loss": 1.1775,
+      "step": 2878
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018676591609303968,
+      "loss": 1.7892,
+      "step": 2879
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018675214570906676,
+      "loss": 1.7734,
+      "step": 2880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000186738368672807,
+      "loss": 1.3872,
+      "step": 2881
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018672458498531684,
+      "loss": 1.5834,
+      "step": 2882
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018671079464765327,
+      "loss": 1.1375,
+      "step": 2883
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018669699766087368,
+      "loss": 1.5349,
+      "step": 2884
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018668319402603606,
+      "loss": 1.4113,
+      "step": 2885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018666938374419897,
+      "loss": 1.7975,
+      "step": 2886
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018665556681642134,
+      "loss": 1.6529,
+      "step": 2887
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018664174324376265,
+      "loss": 1.6673,
+      "step": 2888
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018662791302728298,
+      "loss": 0.759,
+      "step": 2889
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018661407616804283,
+      "loss": 1.7738,
+      "step": 2890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018660023266710318,
+      "loss": 0.9907,
+      "step": 2891
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018658638252552564,
+      "loss": 1.7542,
+      "step": 2892
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001865725257443722,
+      "loss": 1.8791,
+      "step": 2893
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001865586623247055,
+      "loss": 1.6155,
+      "step": 2894
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018654479226758852,
+      "loss": 1.3394,
+      "step": 2895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018653091557408485,
+      "loss": 1.8145,
+      "step": 2896
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018651703224525866,
+      "loss": 1.4847,
+      "step": 2897
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018650314228217446,
+      "loss": 1.789,
+      "step": 2898
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018648924568589735,
+      "loss": 1.4905,
+      "step": 2899
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018647534245749302,
+      "loss": 1.6838,
+      "step": 2900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001864614325980275,
+      "loss": 1.7227,
+      "step": 2901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001864475161085675,
+      "loss": 1.9698,
+      "step": 2902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018643359299018008,
+      "loss": 1.1353,
+      "step": 2903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018641966324393295,
+      "loss": 1.7767,
+      "step": 2904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018640572687089423,
+      "loss": 1.9325,
+      "step": 2905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018639178387213262,
+      "loss": 1.7646,
+      "step": 2906
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018637783424871722,
+      "loss": 1.4312,
+      "step": 2907
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018636387800171776,
+      "loss": 1.6053,
+      "step": 2908
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018634991513220443,
+      "loss": 1.1037,
+      "step": 2909
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001863359456412479,
+      "loss": 1.6096,
+      "step": 2910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001863219695299194,
+      "loss": 1.2911,
+      "step": 2911
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018630798679929061,
+      "loss": 1.6486,
+      "step": 2912
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018629399745043377,
+      "loss": 2.1544,
+      "step": 2913
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018628000148442157,
+      "loss": 1.1531,
+      "step": 2914
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001862659989023273,
+      "loss": 1.3262,
+      "step": 2915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018625198970522464,
+      "loss": 1.604,
+      "step": 2916
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018623797389418786,
+      "loss": 1.5996,
+      "step": 2917
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018622395147029172,
+      "loss": 1.0764,
+      "step": 2918
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018620992243461147,
+      "loss": 2.016,
+      "step": 2919
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001861958867882229,
+      "loss": 1.6949,
+      "step": 2920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018618184453220225,
+      "loss": 1.5287,
+      "step": 2921
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018616779566762634,
+      "loss": 1.6801,
+      "step": 2922
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018615374019557247,
+      "loss": 1.913,
+      "step": 2923
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018613967811711835,
+      "loss": 1.4409,
+      "step": 2924
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018612560943334234,
+      "loss": 1.493,
+      "step": 2925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018611153414532328,
+      "loss": 1.8882,
+      "step": 2926
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018609745225414042,
+      "loss": 1.4655,
+      "step": 2927
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001860833637608736,
+      "loss": 1.498,
+      "step": 2928
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018606926866660318,
+      "loss": 1.5995,
+      "step": 2929
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018605516697240993,
+      "loss": 1.577,
+      "step": 2930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018604105867937526,
+      "loss": 1.6002,
+      "step": 2931
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000186026943788581,
+      "loss": 1.3351,
+      "step": 2932
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018601282230110947,
+      "loss": 1.6582,
+      "step": 2933
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018599869421804356,
+      "loss": 1.7977,
+      "step": 2934
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018598455954046657,
+      "loss": 1.4236,
+      "step": 2935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018597041826946248,
+      "loss": 2.2267,
+      "step": 2936
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018595627040611553,
+      "loss": 1.3204,
+      "step": 2937
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001859421159515107,
+      "loss": 1.4116,
+      "step": 2938
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018592795490673337,
+      "loss": 1.7926,
+      "step": 2939
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018591378727286935,
+      "loss": 1.8443,
+      "step": 2940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018589961305100517,
+      "loss": 1.4625,
+      "step": 2941
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018588543224222755,
+      "loss": 1.8748,
+      "step": 2942
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018587124484762408,
+      "loss": 1.6614,
+      "step": 2943
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018585705086828256,
+      "loss": 1.6412,
+      "step": 2944
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018584285030529143,
+      "loss": 1.611,
+      "step": 2945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018582864315973963,
+      "loss": 1.7591,
+      "step": 2946
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001858144294327165,
+      "loss": 1.2645,
+      "step": 2947
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018580020912531216,
+      "loss": 1.4924,
+      "step": 2948
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018578598223861686,
+      "loss": 1.9972,
+      "step": 2949
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001857717487737216,
+      "loss": 1.0615,
+      "step": 2950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018575750873171782,
+      "loss": 1.1918,
+      "step": 2951
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018574326211369751,
+      "loss": 1.4977,
+      "step": 2952
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018572900892075306,
+      "loss": 1.5727,
+      "step": 2953
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018571474915397745,
+      "loss": 1.5263,
+      "step": 2954
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001857004828144642,
+      "loss": 1.3888,
+      "step": 2955
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018568620990330718,
+      "loss": 1.7579,
+      "step": 2956
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018567193042160092,
+      "loss": 1.8524,
+      "step": 2957
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018565764437044035,
+      "loss": 1.4958,
+      "step": 2958
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000185643351750921,
+      "loss": 1.4959,
+      "step": 2959
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018562905256413879,
+      "loss": 1.1869,
+      "step": 2960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018561474681119022,
+      "loss": 0.9522,
+      "step": 2961
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018560043449317233,
+      "loss": 1.3157,
+      "step": 2962
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018558611561118253,
+      "loss": 1.5698,
+      "step": 2963
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001855717901663189,
+      "loss": 1.1092,
+      "step": 2964
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018555745815967983,
+      "loss": 1.4784,
+      "step": 2965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018554311959236442,
+      "loss": 1.941,
+      "step": 2966
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018552877446547213,
+      "loss": 1.9091,
+      "step": 2967
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018551442278010296,
+      "loss": 1.4573,
+      "step": 2968
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018550006453735743,
+      "loss": 1.7051,
+      "step": 2969
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018548569973833657,
+      "loss": 1.6088,
+      "step": 2970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018547132838414186,
+      "loss": 1.2705,
+      "step": 2971
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018545695047587533,
+      "loss": 1.5127,
+      "step": 2972
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001854425660146395,
+      "loss": 1.6473,
+      "step": 2973
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018542817500153736,
+      "loss": 1.716,
+      "step": 2974
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018541377743767253,
+      "loss": 1.7396,
+      "step": 2975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018539937332414896,
+      "loss": 1.753,
+      "step": 2976
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001853849626620712,
+      "loss": 1.4755,
+      "step": 2977
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018537054545254427,
+      "loss": 1.5929,
+      "step": 2978
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001853561216966737,
+      "loss": 1.49,
+      "step": 2979
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018534169139556556,
+      "loss": 1.6521,
+      "step": 2980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018532725455032637,
+      "loss": 1.5616,
+      "step": 2981
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018531281116206314,
+      "loss": 1.8725,
+      "step": 2982
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001852983612318834,
+      "loss": 1.857,
+      "step": 2983
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018528390476089528,
+      "loss": 1.768,
+      "step": 2984
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018526944175020727,
+      "loss": 1.6321,
+      "step": 2985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018525497220092842,
+      "loss": 1.8723,
+      "step": 2986
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018524049611416826,
+      "loss": 1.5782,
+      "step": 2987
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018522601349103685,
+      "loss": 1.6875,
+      "step": 2988
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018521152433264475,
+      "loss": 1.8945,
+      "step": 2989
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018519702864010302,
+      "loss": 1.7945,
+      "step": 2990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018518252641452315,
+      "loss": 1.463,
+      "step": 2991
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018516801765701727,
+      "loss": 2.0543,
+      "step": 2992
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001851535023686979,
+      "loss": 1.1995,
+      "step": 2993
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018513898055067812,
+      "loss": 1.2648,
+      "step": 2994
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018512445220407141,
+      "loss": 1.5708,
+      "step": 2995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001851099173299919,
+      "loss": 1.522,
+      "step": 2996
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018509537592955412,
+      "loss": 1.7042,
+      "step": 2997
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018508082800387316,
+      "loss": 1.0941,
+      "step": 2998
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018506627355406453,
+      "loss": 1.6866,
+      "step": 2999
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018505171258124427,
+      "loss": 1.3466,
+      "step": 3000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000185037145086529,
+      "loss": 1.7442,
+      "step": 3001
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018502257107103577,
+      "loss": 1.9741,
+      "step": 3002
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001850079905358821,
+      "loss": 1.4212,
+      "step": 3003
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018499340348218605,
+      "loss": 1.2746,
+      "step": 3004
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018497880991106622,
+      "loss": 1.6641,
+      "step": 3005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018496420982364162,
+      "loss": 1.8519,
+      "step": 3006
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018494960322103183,
+      "loss": 1.5941,
+      "step": 3007
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018493499010435694,
+      "loss": 1.5497,
+      "step": 3008
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001849203704747374,
+      "loss": 1.7898,
+      "step": 3009
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001849057443332944,
+      "loss": 1.1279,
+      "step": 3010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001848911116811494,
+      "loss": 1.1854,
+      "step": 3011
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001848764725194245,
+      "loss": 1.7184,
+      "step": 3012
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018486182684924224,
+      "loss": 1.706,
+      "step": 3013
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018484717467172562,
+      "loss": 1.3792,
+      "step": 3014
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018483251598799827,
+      "loss": 1.4808,
+      "step": 3015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018481785079918424,
+      "loss": 1.3157,
+      "step": 3016
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018480317910640802,
+      "loss": 1.3206,
+      "step": 3017
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001847885009107947,
+      "loss": 1.0848,
+      "step": 3018
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001847738162134698,
+      "loss": 1.5014,
+      "step": 3019
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001847591250155594,
+      "loss": 1.5209,
+      "step": 3020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018474442731819,
+      "loss": 1.5247,
+      "step": 3021
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001847297231224887,
+      "loss": 1.317,
+      "step": 3022
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018471501242958298,
+      "loss": 2.2593,
+      "step": 3023
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018470029524060092,
+      "loss": 1.6251,
+      "step": 3024
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018468557155667102,
+      "loss": 1.8755,
+      "step": 3025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018467084137892235,
+      "loss": 1.2266,
+      "step": 3026
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001846561047084844,
+      "loss": 1.4377,
+      "step": 3027
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018464136154648726,
+      "loss": 1.6608,
+      "step": 3028
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001846266118940614,
+      "loss": 1.629,
+      "step": 3029
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018461185575233789,
+      "loss": 1.3593,
+      "step": 3030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018459709312244825,
+      "loss": 1.7308,
+      "step": 3031
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018458232400552445,
+      "loss": 1.5814,
+      "step": 3032
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018456754840269905,
+      "loss": 1.3481,
+      "step": 3033
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018455276631510509,
+      "loss": 1.8932,
+      "step": 3034
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018453797774387601,
+      "loss": 1.5413,
+      "step": 3035
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001845231826901459,
+      "loss": 1.3937,
+      "step": 3036
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001845083811550492,
+      "loss": 1.3276,
+      "step": 3037
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018449357313972097,
+      "loss": 1.7111,
+      "step": 3038
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001844787586452967,
+      "loss": 1.6808,
+      "step": 3039
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018446393767291232,
+      "loss": 1.2878,
+      "step": 3040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018444911022370443,
+      "loss": 1.7013,
+      "step": 3041
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018443427629880993,
+      "loss": 1.8378,
+      "step": 3042
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018441943589936636,
+      "loss": 1.3675,
+      "step": 3043
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018440458902651172,
+      "loss": 1.4153,
+      "step": 3044
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018438973568138444,
+      "loss": 1.7638,
+      "step": 3045
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018437487586512352,
+      "loss": 1.793,
+      "step": 3046
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018436000957886846,
+      "loss": 1.5785,
+      "step": 3047
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018434513682375915,
+      "loss": 1.4004,
+      "step": 3048
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018433025760093615,
+      "loss": 2.2292,
+      "step": 3049
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018431537191154038,
+      "loss": 1.5007,
+      "step": 3050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018430047975671329,
+      "loss": 1.2973,
+      "step": 3051
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018428558113759685,
+      "loss": 1.8696,
+      "step": 3052
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018427067605533348,
+      "loss": 1.6895,
+      "step": 3053
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018425576451106617,
+      "loss": 1.4376,
+      "step": 3054
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001842408465059383,
+      "loss": 1.1971,
+      "step": 3055
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018422592204109386,
+      "loss": 2.2975,
+      "step": 3056
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018421099111767727,
+      "loss": 1.5194,
+      "step": 3057
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018419605373683342,
+      "loss": 1.2287,
+      "step": 3058
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018418110989970777,
+      "loss": 2.2499,
+      "step": 3059
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018416615960744622,
+      "loss": 1.6172,
+      "step": 3060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018415120286119523,
+      "loss": 1.5786,
+      "step": 3061
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001841362396621016,
+      "loss": 1.4687,
+      "step": 3062
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001841212700113128,
+      "loss": 1.6602,
+      "step": 3063
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018410629390997675,
+      "loss": 1.7848,
+      "step": 3064
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018409131135924177,
+      "loss": 1.503,
+      "step": 3065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018407632236025682,
+      "loss": 1.2226,
+      "step": 3066
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018406132691417123,
+      "loss": 1.8058,
+      "step": 3067
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018404632502213488,
+      "loss": 1.5883,
+      "step": 3068
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018403131668529813,
+      "loss": 2.4926,
+      "step": 3069
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001840163019048119,
+      "loss": 1.1691,
+      "step": 3070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018400128068182746,
+      "loss": 1.8364,
+      "step": 3071
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018398625301749673,
+      "loss": 1.2764,
+      "step": 3072
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000183971218912972,
+      "loss": 1.9105,
+      "step": 3073
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018395617836940614,
+      "loss": 1.2544,
+      "step": 3074
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018394113138795249,
+      "loss": 1.5155,
+      "step": 3075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018392607796976487,
+      "loss": 2.086,
+      "step": 3076
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001839110181159976,
+      "loss": 1.647,
+      "step": 3077
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018389595182780543,
+      "loss": 1.4542,
+      "step": 3078
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018388087910634375,
+      "loss": 1.922,
+      "step": 3079
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018386579995276835,
+      "loss": 2.3265,
+      "step": 3080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018385071436823543,
+      "loss": 1.5043,
+      "step": 3081
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018383562235390192,
+      "loss": 1.0051,
+      "step": 3082
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018382052391092502,
+      "loss": 1.8994,
+      "step": 3083
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018380541904046248,
+      "loss": 1.289,
+      "step": 3084
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018379030774367261,
+      "loss": 1.3523,
+      "step": 3085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018377519002171414,
+      "loss": 1.57,
+      "step": 3086
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018376006587574633,
+      "loss": 1.9388,
+      "step": 3087
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 1.0713248252868652,
+      "eval_runtime": 20.5406,
+      "eval_samples_per_second": 138.896,
+      "eval_steps_per_second": 138.896,
+      "step": 3087
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018374493530692895,
+      "loss": 1.8295,
+      "step": 3088
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018372979831642215,
+      "loss": 1.8445,
+      "step": 3089
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018371465490538675,
+      "loss": 1.5606,
+      "step": 3090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018369950507498393,
+      "loss": 1.9061,
+      "step": 3091
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001836843488263754,
+      "loss": 1.4473,
+      "step": 3092
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018366918616072339,
+      "loss": 1.4569,
+      "step": 3093
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018365401707919055,
+      "loss": 1.5115,
+      "step": 3094
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001836388415829401,
+      "loss": 1.7681,
+      "step": 3095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018362365967313572,
+      "loss": 1.9774,
+      "step": 3096
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018360847135094158,
+      "loss": 1.926,
+      "step": 3097
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018359327661752236,
+      "loss": 1.8721,
+      "step": 3098
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018357807547404316,
+      "loss": 1.4281,
+      "step": 3099
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001835628679216697,
+      "loss": 1.5982,
+      "step": 3100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018354765396156805,
+      "loss": 1.6903,
+      "step": 3101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018353243359490486,
+      "loss": 1.6337,
+      "step": 3102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001835172068228473,
+      "loss": 1.3144,
+      "step": 3103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018350197364656295,
+      "loss": 1.4996,
+      "step": 3104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018348673406721986,
+      "loss": 1.5175,
+      "step": 3105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018347148808598669,
+      "loss": 1.1543,
+      "step": 3106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018345623570403254,
+      "loss": 1.4885,
+      "step": 3107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018344097692252694,
+      "loss": 1.3107,
+      "step": 3108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018342571174263995,
+      "loss": 1.8753,
+      "step": 3109
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018341044016554215,
+      "loss": 1.6151,
+      "step": 3110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018339516219240457,
+      "loss": 1.8814,
+      "step": 3111
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001833798778243988,
+      "loss": 0.9895,
+      "step": 3112
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001833645870626968,
+      "loss": 1.4015,
+      "step": 3113
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018334928990847114,
+      "loss": 1.4799,
+      "step": 3114
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001833339863628948,
+      "loss": 1.8945,
+      "step": 3115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001833186764271413,
+      "loss": 1.764,
+      "step": 3116
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018330336010238457,
+      "loss": 1.5043,
+      "step": 3117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001832880373897992,
+      "loss": 2.1298,
+      "step": 3118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018327270829056008,
+      "loss": 2.1869,
+      "step": 3119
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018325737280584266,
+      "loss": 1.9191,
+      "step": 3120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018324203093682292,
+      "loss": 1.1009,
+      "step": 3121
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001832266826846773,
+      "loss": 1.8564,
+      "step": 3122
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018321132805058273,
+      "loss": 1.4993,
+      "step": 3123
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001831959670357166,
+      "loss": 1.4736,
+      "step": 3124
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018318059964125683,
+      "loss": 1.499,
+      "step": 3125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001831652258683818,
+      "loss": 1.6164,
+      "step": 3126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018314984571827044,
+      "loss": 2.1832,
+      "step": 3127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001831344591921021,
+      "loss": 0.9482,
+      "step": 3128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001831190662910566,
+      "loss": 1.6347,
+      "step": 3129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018310366701631436,
+      "loss": 1.3403,
+      "step": 3130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018308826136905618,
+      "loss": 1.7869,
+      "step": 3131
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018307284935046339,
+      "loss": 1.5977,
+      "step": 3132
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018305743096171782,
+      "loss": 1.3846,
+      "step": 3133
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018304200620400176,
+      "loss": 2.0407,
+      "step": 3134
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018302657507849804,
+      "loss": 1.7665,
+      "step": 3135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018301113758638989,
+      "loss": 1.9428,
+      "step": 3136
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001829956937288611,
+      "loss": 1.5469,
+      "step": 3137
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018298024350709593,
+      "loss": 1.61,
+      "step": 3138
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001829647869222791,
+      "loss": 1.6737,
+      "step": 3139
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018294932397559593,
+      "loss": 1.2643,
+      "step": 3140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001829338546682321,
+      "loss": 2.3608,
+      "step": 3141
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018291837900137375,
+      "loss": 2.5318,
+      "step": 3142
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018290289697620767,
+      "loss": 1.0969,
+      "step": 3143
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018288740859392103,
+      "loss": 1.5805,
+      "step": 3144
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018287191385570143,
+      "loss": 1.6689,
+      "step": 3145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018285641276273712,
+      "loss": 1.5752,
+      "step": 3146
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001828409053162167,
+      "loss": 1.6219,
+      "step": 3147
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001828253915173293,
+      "loss": 1.6058,
+      "step": 3148
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018280987136726456,
+      "loss": 1.6174,
+      "step": 3149
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018279434486721263,
+      "loss": 1.7734,
+      "step": 3150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018277881201836404,
+      "loss": 1.2438,
+      "step": 3151
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018276327282190988,
+      "loss": 1.8637,
+      "step": 3152
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018274772727904174,
+      "loss": 1.1383,
+      "step": 3153
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018273217539095167,
+      "loss": 1.5597,
+      "step": 3154
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018271661715883218,
+      "loss": 1.8205,
+      "step": 3155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018270105258387637,
+      "loss": 2.1964,
+      "step": 3156
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001826854816672777,
+      "loss": 1.7196,
+      "step": 3157
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001826699044102302,
+      "loss": 1.7378,
+      "step": 3158
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018265432081392832,
+      "loss": 1.629,
+      "step": 3159
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001826387308795671,
+      "loss": 2.0238,
+      "step": 3160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018262313460834192,
+      "loss": 1.1034,
+      "step": 3161
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001826075320014488,
+      "loss": 1.8229,
+      "step": 3162
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001825919230600841,
+      "loss": 1.4007,
+      "step": 3163
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001825763077854448,
+      "loss": 1.7519,
+      "step": 3164
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018256068617872827,
+      "loss": 1.2837,
+      "step": 3165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018254505824113242,
+      "loss": 2.091,
+      "step": 3166
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001825294239738556,
+      "loss": 1.6742,
+      "step": 3167
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001825137833780967,
+      "loss": 1.3722,
+      "step": 3168
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000182498136455055,
+      "loss": 1.5159,
+      "step": 3169
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018248248320593038,
+      "loss": 1.6504,
+      "step": 3170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018246682363192319,
+      "loss": 1.1253,
+      "step": 3171
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018245115773423413,
+      "loss": 1.4122,
+      "step": 3172
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001824354855140646,
+      "loss": 1.6083,
+      "step": 3173
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001824198069726163,
+      "loss": 1.6782,
+      "step": 3174
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018240412211109147,
+      "loss": 1.1687,
+      "step": 3175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018238843093069292,
+      "loss": 1.5781,
+      "step": 3176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001823727334326238,
+      "loss": 1.1063,
+      "step": 3177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018235702961808786,
+      "loss": 1.1042,
+      "step": 3178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001823413194882893,
+      "loss": 1.5474,
+      "step": 3179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018232560304443272,
+      "loss": 1.4402,
+      "step": 3180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001823098802877234,
+      "loss": 1.2868,
+      "step": 3181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018229415121936692,
+      "loss": 2.1049,
+      "step": 3182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018227841584056938,
+      "loss": 2.1321,
+      "step": 3183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018226267415253746,
+      "loss": 1.5037,
+      "step": 3184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018224692615647817,
+      "loss": 1.6128,
+      "step": 3185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001822311718535992,
+      "loss": 1.4938,
+      "step": 3186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018221541124510853,
+      "loss": 1.2663,
+      "step": 3187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018219964433221475,
+      "loss": 1.9044,
+      "step": 3188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018218387111612686,
+      "loss": 1.7801,
+      "step": 3189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001821680915980544,
+      "loss": 1.7571,
+      "step": 3190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018215230577920733,
+      "loss": 1.3476,
+      "step": 3191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001821365136607962,
+      "loss": 1.298,
+      "step": 3192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018212071524403191,
+      "loss": 1.7834,
+      "step": 3193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018210491053012592,
+      "loss": 1.4133,
+      "step": 3194
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018208909952029017,
+      "loss": 1.655,
+      "step": 3195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018207328221573712,
+      "loss": 1.5339,
+      "step": 3196
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018205745861767957,
+      "loss": 1.5857,
+      "step": 3197
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018204162872733094,
+      "loss": 1.3706,
+      "step": 3198
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018202579254590512,
+      "loss": 1.3385,
+      "step": 3199
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018200995007461638,
+      "loss": 1.8474,
+      "step": 3200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018199410131467962,
+      "loss": 0.8503,
+      "step": 3201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018197824626731013,
+      "loss": 1.265,
+      "step": 3202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018196238493372365,
+      "loss": 1.7064,
+      "step": 3203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018194651731513652,
+      "loss": 1.2593,
+      "step": 3204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018193064341276544,
+      "loss": 1.5983,
+      "step": 3205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001819147632278277,
+      "loss": 1.9365,
+      "step": 3206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018189887676154097,
+      "loss": 1.4544,
+      "step": 3207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018188298401512342,
+      "loss": 1.7032,
+      "step": 3208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018186708498979382,
+      "loss": 1.4378,
+      "step": 3209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018185117968677126,
+      "loss": 1.941,
+      "step": 3210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018183526810727543,
+      "loss": 1.3287,
+      "step": 3211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001818193502525264,
+      "loss": 0.9542,
+      "step": 3212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018180342612374482,
+      "loss": 1.5628,
+      "step": 3213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018178749572215174,
+      "loss": 1.5434,
+      "step": 3214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001817715590489688,
+      "loss": 1.8259,
+      "step": 3215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018175561610541796,
+      "loss": 1.1648,
+      "step": 3216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018173966689272178,
+      "loss": 0.9135,
+      "step": 3217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018172371141210328,
+      "loss": 2.325,
+      "step": 3218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018170774966478598,
+      "loss": 1.4852,
+      "step": 3219
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018169178165199377,
+      "loss": 1.6793,
+      "step": 3220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001816758073749512,
+      "loss": 1.4246,
+      "step": 3221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018165982683488314,
+      "loss": 1.6777,
+      "step": 3222
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000181643840033015,
+      "loss": 1.6822,
+      "step": 3223
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018162784697057268,
+      "loss": 1.9102,
+      "step": 3224
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018161184764878258,
+      "loss": 1.8076,
+      "step": 3225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018159584206887153,
+      "loss": 1.3796,
+      "step": 3226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018157983023206687,
+      "loss": 1.555,
+      "step": 3227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001815638121395964,
+      "loss": 1.8478,
+      "step": 3228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001815477877926884,
+      "loss": 1.9827,
+      "step": 3229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018153175719257166,
+      "loss": 1.4678,
+      "step": 3230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018151572034047546,
+      "loss": 1.5974,
+      "step": 3231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018149967723762945,
+      "loss": 1.8972,
+      "step": 3232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018148362788526393,
+      "loss": 1.3924,
+      "step": 3233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018146757228460953,
+      "loss": 0.7557,
+      "step": 3234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018145151043689746,
+      "loss": 1.0989,
+      "step": 3235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018143544234335934,
+      "loss": 2.0561,
+      "step": 3236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018141936800522728,
+      "loss": 1.6342,
+      "step": 3237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001814032874237339,
+      "loss": 1.2582,
+      "step": 3238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018138720060011232,
+      "loss": 1.6139,
+      "step": 3239
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018137110753559604,
+      "loss": 1.5196,
+      "step": 3240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018135500823141914,
+      "loss": 1.5884,
+      "step": 3241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018133890268881616,
+      "loss": 2.5685,
+      "step": 3242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018132279090902201,
+      "loss": 1.3506,
+      "step": 3243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018130667289327227,
+      "loss": 1.8035,
+      "step": 3244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018129054864280284,
+      "loss": 1.6617,
+      "step": 3245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018127441815885014,
+      "loss": 1.4844,
+      "step": 3246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018125828144265112,
+      "loss": 2.3066,
+      "step": 3247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001812421384954431,
+      "loss": 1.7835,
+      "step": 3248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018122598931846407,
+      "loss": 1.1692,
+      "step": 3249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018120983391295225,
+      "loss": 1.702,
+      "step": 3250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018119367228014652,
+      "loss": 1.1465,
+      "step": 3251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018117750442128613,
+      "loss": 1.8145,
+      "step": 3252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018116133033761097,
+      "loss": 1.7754,
+      "step": 3253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018114515003036114,
+      "loss": 1.4537,
+      "step": 3254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018112896350077747,
+      "loss": 1.4967,
+      "step": 3255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018111277075010115,
+      "loss": 1.5847,
+      "step": 3256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018109657177957389,
+      "loss": 1.7131,
+      "step": 3257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001810803665904378,
+      "loss": 1.9627,
+      "step": 3258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001810641551839355,
+      "loss": 1.256,
+      "step": 3259
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001810479375613102,
+      "loss": 1.2793,
+      "step": 3260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001810317137238054,
+      "loss": 1.2946,
+      "step": 3261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018101548367266523,
+      "loss": 1.0411,
+      "step": 3262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018099924740913422,
+      "loss": 1.0666,
+      "step": 3263
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018098300493445737,
+      "loss": 1.488,
+      "step": 3264
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018096675624988014,
+      "loss": 1.091,
+      "step": 3265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018095050135664864,
+      "loss": 1.9521,
+      "step": 3266
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018093424025600918,
+      "loss": 1.6074,
+      "step": 3267
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018091797294920875,
+      "loss": 1.8986,
+      "step": 3268
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018090169943749476,
+      "loss": 2.0847,
+      "step": 3269
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018088541972211504,
+      "loss": 1.7079,
+      "step": 3270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018086913380431797,
+      "loss": 1.8025,
+      "step": 3271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018085284168535242,
+      "loss": 2.1479,
+      "step": 3272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018083654336646765,
+      "loss": 2.2778,
+      "step": 3273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018082023884891343,
+      "loss": 1.8628,
+      "step": 3274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018080392813394003,
+      "loss": 1.936,
+      "step": 3275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001807876112227982,
+      "loss": 1.0248,
+      "step": 3276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001807712881167391,
+      "loss": 1.7456,
+      "step": 3277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018075495881701446,
+      "loss": 1.6716,
+      "step": 3278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018073862332487642,
+      "loss": 1.3443,
+      "step": 3279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001807222816415776,
+      "loss": 1.4388,
+      "step": 3280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018070593376837108,
+      "loss": 1.3249,
+      "step": 3281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018068957970651051,
+      "loss": 1.2074,
+      "step": 3282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018067321945724985,
+      "loss": 1.4591,
+      "step": 3283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018065685302184374,
+      "loss": 1.1781,
+      "step": 3284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001806404804015471,
+      "loss": 1.7831,
+      "step": 3285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018062410159761544,
+      "loss": 1.8253,
+      "step": 3286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018060771661130472,
+      "loss": 1.6631,
+      "step": 3287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001805913254438713,
+      "loss": 1.4776,
+      "step": 3288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001805749280965722,
+      "loss": 1.9491,
+      "step": 3289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001805585245706647,
+      "loss": 1.5618,
+      "step": 3290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018054211486740664,
+      "loss": 1.6528,
+      "step": 3291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001805256989880564,
+      "loss": 1.4316,
+      "step": 3292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018050927693387275,
+      "loss": 1.5931,
+      "step": 3293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018049284870611497,
+      "loss": 1.9242,
+      "step": 3294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018047641430604273,
+      "loss": 1.6796,
+      "step": 3295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018045997373491635,
+      "loss": 1.4885,
+      "step": 3296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018044352699399645,
+      "loss": 1.7674,
+      "step": 3297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018042707408454423,
+      "loss": 1.7464,
+      "step": 3298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001804106150078213,
+      "loss": 0.8352,
+      "step": 3299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018039414976508974,
+      "loss": 2.0116,
+      "step": 3300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001803776783576122,
+      "loss": 1.3471,
+      "step": 3301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001803612007866517,
+      "loss": 1.8734,
+      "step": 3302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018034471705347173,
+      "loss": 1.521,
+      "step": 3303
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018032822715933634,
+      "loss": 1.2978,
+      "step": 3304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018031173110550998,
+      "loss": 1.4209,
+      "step": 3305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018029522889325759,
+      "loss": 1.5869,
+      "step": 3306
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018027872052384458,
+      "loss": 2.3493,
+      "step": 3307
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018026220599853682,
+      "loss": 1.3859,
+      "step": 3308
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018024568531860074,
+      "loss": 1.8969,
+      "step": 3309
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018022915848530307,
+      "loss": 1.402,
+      "step": 3310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001802126254999112,
+      "loss": 1.7187,
+      "step": 3311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018019608636369287,
+      "loss": 1.2679,
+      "step": 3312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018017954107791632,
+      "loss": 1.8378,
+      "step": 3313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018016298964385028,
+      "loss": 1.2432,
+      "step": 3314
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018014643206276394,
+      "loss": 1.5424,
+      "step": 3315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018012986833592693,
+      "loss": 1.2713,
+      "step": 3316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001801132984646094,
+      "loss": 1.7543,
+      "step": 3317
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018009672245008197,
+      "loss": 1.7462,
+      "step": 3318
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018008014029361572,
+      "loss": 1.4713,
+      "step": 3319
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018006355199648216,
+      "loss": 1.3188,
+      "step": 3320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018004695755995334,
+      "loss": 1.4816,
+      "step": 3321
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018003035698530172,
+      "loss": 1.7013,
+      "step": 3322
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018001375027380028,
+      "loss": 1.4597,
+      "step": 3323
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017999713742672242,
+      "loss": 1.8684,
+      "step": 3324
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017998051844534204,
+      "loss": 1.8312,
+      "step": 3325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017996389333093356,
+      "loss": 2.2166,
+      "step": 3326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017994726208477176,
+      "loss": 0.8481,
+      "step": 3327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017993062470813195,
+      "loss": 1.1148,
+      "step": 3328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017991398120228996,
+      "loss": 1.8953,
+      "step": 3329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000179897331568522,
+      "loss": 0.7719,
+      "step": 3330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001798806758081048,
+      "loss": 1.6972,
+      "step": 3331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017986401392231557,
+      "loss": 1.4492,
+      "step": 3332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017984734591243196,
+      "loss": 1.1911,
+      "step": 3333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017983067177973204,
+      "loss": 1.1637,
+      "step": 3334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017981399152549446,
+      "loss": 1.7736,
+      "step": 3335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001797973051509983,
+      "loss": 1.713,
+      "step": 3336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001797806126575231,
+      "loss": 1.7877,
+      "step": 3337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017976391404634878,
+      "loss": 0.949,
+      "step": 3338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017974720931875592,
+      "loss": 1.6219,
+      "step": 3339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017973049847602542,
+      "loss": 1.7298,
+      "step": 3340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001797137815194387,
+      "loss": 1.3312,
+      "step": 3341
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001796970584502776,
+      "loss": 1.3639,
+      "step": 3342
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017968032926982456,
+      "loss": 1.6226,
+      "step": 3343
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017966359397936232,
+      "loss": 1.5434,
+      "step": 3344
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017964685258017418,
+      "loss": 1.9343,
+      "step": 3345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017963010507354395,
+      "loss": 1.796,
+      "step": 3346
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017961335146075578,
+      "loss": 1.7892,
+      "step": 3347
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001795965917430944,
+      "loss": 1.6208,
+      "step": 3348
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000179579825921845,
+      "loss": 1.2421,
+      "step": 3349
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017956305399829313,
+      "loss": 1.4381,
+      "step": 3350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017954627597372498,
+      "loss": 1.2453,
+      "step": 3351
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000179529491849427,
+      "loss": 1.5405,
+      "step": 3352
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017951270162668638,
+      "loss": 1.7469,
+      "step": 3353
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017949590530679047,
+      "loss": 1.8707,
+      "step": 3354
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017947910289102734,
+      "loss": 0.8437,
+      "step": 3355
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017946229438068535,
+      "loss": 0.479,
+      "step": 3356
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017944547977705343,
+      "loss": 1.8149,
+      "step": 3357
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017942865908142098,
+      "loss": 1.8709,
+      "step": 3358
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017941183229507778,
+      "loss": 1.5093,
+      "step": 3359
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001793949994193142,
+      "loss": 1.8197,
+      "step": 3360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017937816045542095,
+      "loss": 1.0025,
+      "step": 3361
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001793613154046893,
+      "loss": 1.8569,
+      "step": 3362
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017934446426841094,
+      "loss": 1.5815,
+      "step": 3363
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017932760704787804,
+      "loss": 2.1719,
+      "step": 3364
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001793107437443833,
+      "loss": 1.476,
+      "step": 3365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001792938743592197,
+      "loss": 1.4193,
+      "step": 3366
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001792769988936809,
+      "loss": 0.8716,
+      "step": 3367
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017926011734906093,
+      "loss": 1.4425,
+      "step": 3368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017924322972665425,
+      "loss": 1.4274,
+      "step": 3369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017922633602775588,
+      "loss": 1.1659,
+      "step": 3370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017920943625366123,
+      "loss": 1.4001,
+      "step": 3371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001791925304056662,
+      "loss": 1.7919,
+      "step": 3372
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017917561848506718,
+      "loss": 1.304,
+      "step": 3373
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017915870049316093,
+      "loss": 1.1695,
+      "step": 3374
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001791417764312448,
+      "loss": 1.5309,
+      "step": 3375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001791248463006166,
+      "loss": 1.6666,
+      "step": 3376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017910791010257446,
+      "loss": 1.7982,
+      "step": 3377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017909096783841715,
+      "loss": 1.7018,
+      "step": 3378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001790740195094438,
+      "loss": 1.216,
+      "step": 3379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017905706511695403,
+      "loss": 1.9206,
+      "step": 3380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001790401046622479,
+      "loss": 1.3347,
+      "step": 3381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017902313814662603,
+      "loss": 1.9651,
+      "step": 3382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017900616557138942,
+      "loss": 1.4267,
+      "step": 3383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001789891869378395,
+      "loss": 1.8992,
+      "step": 3384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017897220224727826,
+      "loss": 1.4035,
+      "step": 3385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017895521150100812,
+      "loss": 1.4602,
+      "step": 3386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017893821470033196,
+      "loss": 0.9868,
+      "step": 3387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001789212118465531,
+      "loss": 1.6293,
+      "step": 3388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017890420294097532,
+      "loss": 1.0149,
+      "step": 3389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017888718798490292,
+      "loss": 1.3656,
+      "step": 3390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017887016697964065,
+      "loss": 1.382,
+      "step": 3391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017885313992649365,
+      "loss": 2.275,
+      "step": 3392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017883610682676765,
+      "loss": 1.5255,
+      "step": 3393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017881906768176876,
+      "loss": 1.2431,
+      "step": 3394
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017880202249280353,
+      "loss": 1.5863,
+      "step": 3395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017878497126117902,
+      "loss": 1.6656,
+      "step": 3396
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017876791398820275,
+      "loss": 1.1406,
+      "step": 3397
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001787508506751827,
+      "loss": 1.6855,
+      "step": 3398
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017873378132342735,
+      "loss": 1.0348,
+      "step": 3399
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017871670593424554,
+      "loss": 1.9178,
+      "step": 3400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001786996245089467,
+      "loss": 1.0789,
+      "step": 3401
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017868253704884056,
+      "loss": 1.9138,
+      "step": 3402
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001786654435552375,
+      "loss": 1.4811,
+      "step": 3403
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017864834402944827,
+      "loss": 2.0236,
+      "step": 3404
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017863123847278406,
+      "loss": 1.493,
+      "step": 3405
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017861412688655655,
+      "loss": 1.5099,
+      "step": 3406
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001785970092720779,
+      "loss": 1.7137,
+      "step": 3407
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001785798856306607,
+      "loss": 1.2671,
+      "step": 3408
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017856275596361804,
+      "loss": 1.2185,
+      "step": 3409
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001785456202722634,
+      "loss": 1.6382,
+      "step": 3410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017852847855791084,
+      "loss": 2.0042,
+      "step": 3411
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017851133082187474,
+      "loss": 1.1687,
+      "step": 3412
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001784941770654701,
+      "loss": 2.1529,
+      "step": 3413
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017847701729001221,
+      "loss": 0.9344,
+      "step": 3414
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017845985149681694,
+      "loss": 2.3247,
+      "step": 3415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001784426796872006,
+      "loss": 1.7787,
+      "step": 3416
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017842550186247995,
+      "loss": 1.799,
+      "step": 3417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017840831802397222,
+      "loss": 1.0599,
+      "step": 3418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017839112817299505,
+      "loss": 1.3977,
+      "step": 3419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017837393231086662,
+      "loss": 1.5429,
+      "step": 3420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017835673043890555,
+      "loss": 1.8989,
+      "step": 3421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017833952255843087,
+      "loss": 1.6259,
+      "step": 3422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017832230867076215,
+      "loss": 1.507,
+      "step": 3423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001783050887772193,
+      "loss": 1.1382,
+      "step": 3424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017828786287912282,
+      "loss": 1.5274,
+      "step": 3425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017827063097779363,
+      "loss": 1.4393,
+      "step": 3426
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001782533930745531,
+      "loss": 1.9409,
+      "step": 3427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017823614917072302,
+      "loss": 1.4929,
+      "step": 3428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001782188992676257,
+      "loss": 0.9391,
+      "step": 3429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001782016433665839,
+      "loss": 1.6687,
+      "step": 3430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001781843814689208,
+      "loss": 1.6065,
+      "step": 3431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017816711357596007,
+      "loss": 0.7743,
+      "step": 3432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001781498396890259,
+      "loss": 1.9595,
+      "step": 3433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017813255980944281,
+      "loss": 1.1302,
+      "step": 3434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017811527393853585,
+      "loss": 1.7604,
+      "step": 3435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017809798207763056,
+      "loss": 1.7408,
+      "step": 3436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001780806842280529,
+      "loss": 1.8966,
+      "step": 3437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017806338039112927,
+      "loss": 1.107,
+      "step": 3438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001780460705681866,
+      "loss": 1.7332,
+      "step": 3439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017802875476055218,
+      "loss": 1.6402,
+      "step": 3440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017801143296955387,
+      "loss": 1.5164,
+      "step": 3441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001779941051965199,
+      "loss": 1.3842,
+      "step": 3442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017797677144277895,
+      "loss": 1.1269,
+      "step": 3443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017795943170966028,
+      "loss": 1.5624,
+      "step": 3444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017794208599849348,
+      "loss": 1.2388,
+      "step": 3445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017792473431060864,
+      "loss": 1.2979,
+      "step": 3446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017790737664733633,
+      "loss": 1.1983,
+      "step": 3447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017789001301000753,
+      "loss": 1.0357,
+      "step": 3448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017787264339995376,
+      "loss": 1.6067,
+      "step": 3449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017785526781850694,
+      "loss": 1.3608,
+      "step": 3450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017783788626699943,
+      "loss": 0.9668,
+      "step": 3451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017782049874676413,
+      "loss": 1.1133,
+      "step": 3452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017780310525913423,
+      "loss": 1.3103,
+      "step": 3453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017778570580544357,
+      "loss": 1.7207,
+      "step": 3454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001777683003870264,
+      "loss": 1.4019,
+      "step": 3455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001777508890052173,
+      "loss": 2.2864,
+      "step": 3456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017773347166135146,
+      "loss": 0.7379,
+      "step": 3457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017771604835676447,
+      "loss": 1.5182,
+      "step": 3458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017769861909279236,
+      "loss": 1.5374,
+      "step": 3459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017768118387077162,
+      "loss": 1.4729,
+      "step": 3460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017766374269203924,
+      "loss": 1.606,
+      "step": 3461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017764629555793262,
+      "loss": 1.6316,
+      "step": 3462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001776288424697896,
+      "loss": 2.2382,
+      "step": 3463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001776113834289486,
+      "loss": 1.7648,
+      "step": 3464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017759391843674832,
+      "loss": 1.8597,
+      "step": 3465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017757644749452804,
+      "loss": 0.5756,
+      "step": 3466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017755897060362744,
+      "loss": 1.9142,
+      "step": 3467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001775414877653867,
+      "loss": 2.0165,
+      "step": 3468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001775239989811464,
+      "loss": 1.3015,
+      "step": 3469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017750650425224764,
+      "loss": 1.679,
+      "step": 3470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017748900358003194,
+      "loss": 1.6048,
+      "step": 3471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001774714969658412,
+      "loss": 1.97,
+      "step": 3472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017745398441101798,
+      "loss": 1.6833,
+      "step": 3473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017743646591690505,
+      "loss": 1.3573,
+      "step": 3474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017741894148484586,
+      "loss": 1.245,
+      "step": 3475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017740141111618414,
+      "loss": 1.4899,
+      "step": 3476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017738387481226417,
+      "loss": 1.1418,
+      "step": 3477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017736633257443062,
+      "loss": 1.5663,
+      "step": 3478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017734878440402876,
+      "loss": 1.5098,
+      "step": 3479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001773312303024041,
+      "loss": 1.5793,
+      "step": 3480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001773136702709028,
+      "loss": 1.2543,
+      "step": 3481
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001772961043108713,
+      "loss": 1.6608,
+      "step": 3482
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017727853242365663,
+      "loss": 1.8767,
+      "step": 3483
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017726095461060627,
+      "loss": 1.9407,
+      "step": 3484
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017724337087306804,
+      "loss": 1.5638,
+      "step": 3485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017722578121239036,
+      "loss": 1.9934,
+      "step": 3486
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017720818562992196,
+      "loss": 1.314,
+      "step": 3487
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017719058412701214,
+      "loss": 1.919,
+      "step": 3488
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017717297670501064,
+      "loss": 1.4768,
+      "step": 3489
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017715536336526756,
+      "loss": 1.8272,
+      "step": 3490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017713774410913353,
+      "loss": 1.7662,
+      "step": 3491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017712011893795964,
+      "loss": 1.401,
+      "step": 3492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001771024878530974,
+      "loss": 1.212,
+      "step": 3493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017708485085589883,
+      "loss": 1.6764,
+      "step": 3494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001770672079477163,
+      "loss": 1.8133,
+      "step": 3495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017704955912990276,
+      "loss": 1.1878,
+      "step": 3496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017703190440381146,
+      "loss": 1.6872,
+      "step": 3497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001770142437707963,
+      "loss": 1.8447,
+      "step": 3498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017699657723221145,
+      "loss": 1.5563,
+      "step": 3499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001769789047894116,
+      "loss": 1.126,
+      "step": 3500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017696122644375194,
+      "loss": 1.3469,
+      "step": 3501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017694354219658807,
+      "loss": 1.6312,
+      "step": 3502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017692585204927603,
+      "loss": 1.7126,
+      "step": 3503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017690815600317236,
+      "loss": 1.3902,
+      "step": 3504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017689045405963394,
+      "loss": 1.5762,
+      "step": 3505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017687274622001827,
+      "loss": 1.1355,
+      "step": 3506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017685503248568316,
+      "loss": 1.3316,
+      "step": 3507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017683731285798694,
+      "loss": 1.8808,
+      "step": 3508
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017681958733828837,
+      "loss": 1.3154,
+      "step": 3509
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017680185592794671,
+      "loss": 1.4981,
+      "step": 3510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017678411862832158,
+      "loss": 1.6812,
+      "step": 3511
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017676637544077314,
+      "loss": 1.6353,
+      "step": 3512
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017674862636666197,
+      "loss": 1.7972,
+      "step": 3513
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017673087140734905,
+      "loss": 1.5073,
+      "step": 3514
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017671311056419587,
+      "loss": 1.7176,
+      "step": 3515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017669534383856438,
+      "loss": 1.3196,
+      "step": 3516
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017667757123181698,
+      "loss": 2.2277,
+      "step": 3517
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017665979274531646,
+      "loss": 1.4021,
+      "step": 3518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001766420083804261,
+      "loss": 1.4255,
+      "step": 3519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017662421813850968,
+      "loss": 1.6736,
+      "step": 3520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017660642202093133,
+      "loss": 1.5078,
+      "step": 3521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017658862002905567,
+      "loss": 1.1469,
+      "step": 3522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017657081216424785,
+      "loss": 1.9746,
+      "step": 3523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017655299842787338,
+      "loss": 1.735,
+      "step": 3524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017653517882129823,
+      "loss": 1.6094,
+      "step": 3525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017651735334588882,
+      "loss": 1.6993,
+      "step": 3526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017649952200301207,
+      "loss": 1.5781,
+      "step": 3527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017648168479403529,
+      "loss": 1.5459,
+      "step": 3528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017646384172032627,
+      "loss": 1.6746,
+      "step": 3529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017644599278325327,
+      "loss": 1.8621,
+      "step": 3530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017642813798418494,
+      "loss": 1.7317,
+      "step": 3531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001764102773244904,
+      "loss": 1.572,
+      "step": 3532
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017639241080553929,
+      "loss": 1.5718,
+      "step": 3533
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001763745384287016,
+      "loss": 1.7291,
+      "step": 3534
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017635666019534784,
+      "loss": 1.5635,
+      "step": 3535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001763387761068489,
+      "loss": 1.0446,
+      "step": 3536
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017632088616457617,
+      "loss": 1.0485,
+      "step": 3537
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017630299036990152,
+      "loss": 1.7978,
+      "step": 3538
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001762850887241972,
+      "loss": 2.1378,
+      "step": 3539
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017626718122883586,
+      "loss": 1.5581,
+      "step": 3540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001762492678851908,
+      "loss": 1.7109,
+      "step": 3541
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017623134869463558,
+      "loss": 1.6203,
+      "step": 3542
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017621342365854424,
+      "loss": 1.4611,
+      "step": 3543
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001761954927782914,
+      "loss": 2.0452,
+      "step": 3544
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017617755605525194,
+      "loss": 1.0866,
+      "step": 3545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017615961349080127,
+      "loss": 1.4375,
+      "step": 3546
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017614166508631533,
+      "loss": 2.4066,
+      "step": 3547
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017612371084317036,
+      "loss": 2.0046,
+      "step": 3548
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017610575076274315,
+      "loss": 1.9946,
+      "step": 3549
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017608778484641085,
+      "loss": 1.6547,
+      "step": 3550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017606981309555123,
+      "loss": 1.2445,
+      "step": 3551
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001760518355115423,
+      "loss": 1.5498,
+      "step": 3552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017603385209576263,
+      "loss": 0.8153,
+      "step": 3553
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001760158628495912,
+      "loss": 1.054,
+      "step": 3554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001759978677744075,
+      "loss": 1.4157,
+      "step": 3555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017597986687159137,
+      "loss": 1.9128,
+      "step": 3556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017596186014252316,
+      "loss": 0.4085,
+      "step": 3557
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017594384758858368,
+      "loss": 1.7821,
+      "step": 3558
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017592582921115415,
+      "loss": 1.6957,
+      "step": 3559
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001759078050116162,
+      "loss": 1.9381,
+      "step": 3560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017588977499135205,
+      "loss": 1.1523,
+      "step": 3561
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017587173915174415,
+      "loss": 1.4063,
+      "step": 3562
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017585369749417563,
+      "loss": 1.7892,
+      "step": 3563
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017583565002002985,
+      "loss": 1.571,
+      "step": 3564
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017581759673069078,
+      "loss": 1.8829,
+      "step": 3565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017579953762754275,
+      "loss": 1.5252,
+      "step": 3566
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001757814727119706,
+      "loss": 1.8211,
+      "step": 3567
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017576340198535953,
+      "loss": 0.7074,
+      "step": 3568
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017574532544909524,
+      "loss": 1.2567,
+      "step": 3569
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017572724310456388,
+      "loss": 1.6777,
+      "step": 3570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000175709154953152,
+      "loss": 1.7775,
+      "step": 3571
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017569106099624666,
+      "loss": 1.6199,
+      "step": 3572
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017567296123523532,
+      "loss": 1.8545,
+      "step": 3573
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001756548556715059,
+      "loss": 2.0163,
+      "step": 3574
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017563674430644678,
+      "loss": 0.2967,
+      "step": 3575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017561862714144674,
+      "loss": 1.593,
+      "step": 3576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017560050417789503,
+      "loss": 1.5705,
+      "step": 3577
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001755823754171814,
+      "loss": 1.3675,
+      "step": 3578
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001755642408606959,
+      "loss": 1.7984,
+      "step": 3579
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017554610050982917,
+      "loss": 1.582,
+      "step": 3580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017552795436597224,
+      "loss": 2.1391,
+      "step": 3581
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017550980243051658,
+      "loss": 2.0475,
+      "step": 3582
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001754916447048541,
+      "loss": 1.7889,
+      "step": 3583
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017547348119037717,
+      "loss": 2.249,
+      "step": 3584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017545531188847862,
+      "loss": 1.7621,
+      "step": 3585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017543713680055162,
+      "loss": 1.8437,
+      "step": 3586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017541895592798997,
+      "loss": 1.3714,
+      "step": 3587
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001754007692721877,
+      "loss": 1.0791,
+      "step": 3588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001753825768345395,
+      "loss": 1.4719,
+      "step": 3589
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001753643786164403,
+      "loss": 1.5058,
+      "step": 3590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001753461746192856,
+      "loss": 1.2962,
+      "step": 3591
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017532796484447135,
+      "loss": 1.5066,
+      "step": 3592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017530974929339386,
+      "loss": 1.3649,
+      "step": 3593
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017529152796744992,
+      "loss": 1.1966,
+      "step": 3594
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017527330086803682,
+      "loss": 1.6319,
+      "step": 3595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017525506799655216,
+      "loss": 1.6858,
+      "step": 3596
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001752368293543941,
+      "loss": 1.7434,
+      "step": 3597
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017521858494296127,
+      "loss": 2.3103,
+      "step": 3598
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017520033476365262,
+      "loss": 1.0926,
+      "step": 3599
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017518207881786756,
+      "loss": 1.8095,
+      "step": 3600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017516381710700607,
+      "loss": 1.7307,
+      "step": 3601
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017514554963246845,
+      "loss": 2.124,
+      "step": 3602
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017512727639565546,
+      "loss": 1.5198,
+      "step": 3603
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017510899739796835,
+      "loss": 1.559,
+      "step": 3604
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017509071264080873,
+      "loss": 0.6289,
+      "step": 3605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001750724221255788,
+      "loss": 1.2465,
+      "step": 3606
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017505412585368103,
+      "loss": 1.9057,
+      "step": 3607
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001750358238265184,
+      "loss": 1.953,
+      "step": 3608
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017501751604549434,
+      "loss": 1.6777,
+      "step": 3609
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017499920251201276,
+      "loss": 1.2514,
+      "step": 3610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017498088322747795,
+      "loss": 1.0775,
+      "step": 3611
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017496255819329467,
+      "loss": 1.3403,
+      "step": 3612
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017494422741086803,
+      "loss": 1.8484,
+      "step": 3613
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017492589088160382,
+      "loss": 1.4904,
+      "step": 3614
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017490754860690797,
+      "loss": 1.4872,
+      "step": 3615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017488920058818708,
+      "loss": 1.27,
+      "step": 3616
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017487084682684802,
+      "loss": 1.453,
+      "step": 3617
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001748524873242983,
+      "loss": 2.2063,
+      "step": 3618
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017483412208194564,
+      "loss": 1.6254,
+      "step": 3619
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001748157511011984,
+      "loss": 1.3317,
+      "step": 3620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017479737438346523,
+      "loss": 1.5154,
+      "step": 3621
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001747789919301553,
+      "loss": 1.9302,
+      "step": 3622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001747606037426783,
+      "loss": 1.4537,
+      "step": 3623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017474220982244412,
+      "loss": 1.5081,
+      "step": 3624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001747238101708633,
+      "loss": 1.3582,
+      "step": 3625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017470540478934675,
+      "loss": 1.8088,
+      "step": 3626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017468699367930585,
+      "loss": 1.7736,
+      "step": 3627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017466857684215233,
+      "loss": 0.9614,
+      "step": 3628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017465015427929847,
+      "loss": 1.3012,
+      "step": 3629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017463172599215695,
+      "loss": 1.4923,
+      "step": 3630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017461329198214083,
+      "loss": 2.5523,
+      "step": 3631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001745948522506637,
+      "loss": 1.7117,
+      "step": 3632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001745764067991395,
+      "loss": 1.6907,
+      "step": 3633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017455795562898273,
+      "loss": 1.4455,
+      "step": 3634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001745394987416082,
+      "loss": 1.5066,
+      "step": 3635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017452103613843121,
+      "loss": 1.634,
+      "step": 3636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017450256782086752,
+      "loss": 2.0834,
+      "step": 3637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017448409379033328,
+      "loss": 1.3557,
+      "step": 3638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017446561404824514,
+      "loss": 1.1768,
+      "step": 3639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017444712859602016,
+      "loss": 1.6165,
+      "step": 3640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001744286374350758,
+      "loss": 1.8369,
+      "step": 3641
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017441014056683,
+      "loss": 1.1273,
+      "step": 3642
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017439163799270118,
+      "loss": 1.9296,
+      "step": 3643
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017437312971410807,
+      "loss": 1.8754,
+      "step": 3644
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001743546157324699,
+      "loss": 1.591,
+      "step": 3645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017433609604920644,
+      "loss": 1.5218,
+      "step": 3646
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017431757066573776,
+      "loss": 1.3886,
+      "step": 3647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001742990395834844,
+      "loss": 1.4797,
+      "step": 3648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017428050280386738,
+      "loss": 1.5016,
+      "step": 3649
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017426196032830813,
+      "loss": 1.6012,
+      "step": 3650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017424341215822845,
+      "loss": 1.7017,
+      "step": 3651
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017422485829505076,
+      "loss": 0.9624,
+      "step": 3652
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001742062987401977,
+      "loss": 1.2703,
+      "step": 3653
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017418773349509245,
+      "loss": 1.0956,
+      "step": 3654
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017416916256115866,
+      "loss": 1.5548,
+      "step": 3655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017415058593982038,
+      "loss": 1.5031,
+      "step": 3656
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001741320036325021,
+      "loss": 1.4544,
+      "step": 3657
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017411341564062868,
+      "loss": 1.4947,
+      "step": 3658
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017409482196562555,
+      "loss": 1.4025,
+      "step": 3659
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017407622260891845,
+      "loss": 1.9193,
+      "step": 3660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001740576175719336,
+      "loss": 1.5993,
+      "step": 3661
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001740390068560977,
+      "loss": 2.3343,
+      "step": 3662
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017402039046283783,
+      "loss": 1.2731,
+      "step": 3663
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017400176839358155,
+      "loss": 1.6374,
+      "step": 3664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001739831406497568,
+      "loss": 1.6124,
+      "step": 3665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017396450723279196,
+      "loss": 2.0851,
+      "step": 3666
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001739458681441159,
+      "loss": 1.5715,
+      "step": 3667
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001739272233851579,
+      "loss": 1.5038,
+      "step": 3668
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001739085729573477,
+      "loss": 1.5077,
+      "step": 3669
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017388991686211538,
+      "loss": 1.5787,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001738712551008915,
+      "loss": 1.8445,
+      "step": 3671
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017385258767510717,
+      "loss": 1.3435,
+      "step": 3672
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017383391458619375,
+      "loss": 1.6698,
+      "step": 3673
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017381523583558317,
+      "loss": 1.5051,
+      "step": 3674
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017379655142470772,
+      "loss": 0.9701,
+      "step": 3675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017377786135500017,
+      "loss": 1.3088,
+      "step": 3676
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017375916562789364,
+      "loss": 1.8135,
+      "step": 3677
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017374046424482184,
+      "loss": 1.6621,
+      "step": 3678
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017372175720721877,
+      "loss": 1.3176,
+      "step": 3679
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001737030445165189,
+      "loss": 1.5869,
+      "step": 3680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001736843261741572,
+      "loss": 1.9485,
+      "step": 3681
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017366560218156894,
+      "loss": 1.5095,
+      "step": 3682
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017364687254018999,
+      "loss": 1.7719,
+      "step": 3683
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017362813725145654,
+      "loss": 1.2619,
+      "step": 3684
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017360939631680522,
+      "loss": 1.7643,
+      "step": 3685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017359064973767308,
+      "loss": 1.5244,
+      "step": 3686
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017357189751549776,
+      "loss": 1.361,
+      "step": 3687
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001735531396517171,
+      "loss": 1.7779,
+      "step": 3688
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001735343761477695,
+      "loss": 1.7057,
+      "step": 3689
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017351560700509377,
+      "loss": 0.7688,
+      "step": 3690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001734968322251292,
+      "loss": 1.5937,
+      "step": 3691
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017347805180931543,
+      "loss": 1.7565,
+      "step": 3692
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001734592657590926,
+      "loss": 1.7965,
+      "step": 3693
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001734404740759012,
+      "loss": 1.6956,
+      "step": 3694
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017342167676118227,
+      "loss": 1.4691,
+      "step": 3695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017340287381637718,
+      "loss": 1.5276,
+      "step": 3696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017338406524292777,
+      "loss": 1.3999,
+      "step": 3697
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001733652510422763,
+      "loss": 1.7253,
+      "step": 3698
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001733464312158655,
+      "loss": 1.7061,
+      "step": 3699
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017332760576513852,
+      "loss": 1.1365,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017330877469153887,
+      "loss": 1.7553,
+      "step": 3701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017328993799651054,
+      "loss": 1.7634,
+      "step": 3702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017327109568149803,
+      "loss": 1.4367,
+      "step": 3703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017325224774794611,
+      "loss": 1.4246,
+      "step": 3704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017323339419730017,
+      "loss": 1.6213,
+      "step": 3705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017321453503100584,
+      "loss": 1.5559,
+      "step": 3706
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001731956702505093,
+      "loss": 1.6628,
+      "step": 3707
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001731767998572571,
+      "loss": 3.2043,
+      "step": 3708
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001731579238526963,
+      "loss": 1.2231,
+      "step": 3709
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017313904223827437,
+      "loss": 1.9963,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017312015501543907,
+      "loss": 1.4031,
+      "step": 3711
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001731012621856388,
+      "loss": 1.7548,
+      "step": 3712
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017308236375032226,
+      "loss": 2.029,
+      "step": 3713
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017306345971093858,
+      "loss": 1.724,
+      "step": 3714
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017304455006893739,
+      "loss": 1.1253,
+      "step": 3715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001730256348257687,
+      "loss": 1.4865,
+      "step": 3716
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017300671398288292,
+      "loss": 1.5257,
+      "step": 3717
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000172987787541731,
+      "loss": 1.5313,
+      "step": 3718
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001729688555037642,
+      "loss": 1.4751,
+      "step": 3719
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001729499178704343,
+      "loss": 1.4421,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017293097464319342,
+      "loss": 1.5889,
+      "step": 3721
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017291202582349417,
+      "loss": 2.0468,
+      "step": 3722
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017289307141278955,
+      "loss": 1.7204,
+      "step": 3723
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001728741114125331,
+      "loss": 1.5416,
+      "step": 3724
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001728551458241786,
+      "loss": 1.3291,
+      "step": 3725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017283617464918042,
+      "loss": 1.518,
+      "step": 3726
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001728171978889933,
+      "loss": 0.8457,
+      "step": 3727
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017279821554507235,
+      "loss": 1.8189,
+      "step": 3728
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001727792276188732,
+      "loss": 1.4856,
+      "step": 3729
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001727602341118519,
+      "loss": 1.8028,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017274123502546487,
+      "loss": 1.1142,
+      "step": 3731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017272223036116901,
+      "loss": 2.4207,
+      "step": 3732
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017270322012042158,
+      "loss": 1.301,
+      "step": 3733
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001726842043046804,
+      "loss": 0.9915,
+      "step": 3734
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017266518291540354,
+      "loss": 1.5658,
+      "step": 3735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017264615595404966,
+      "loss": 1.5913,
+      "step": 3736
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017262712342207772,
+      "loss": 1.2605,
+      "step": 3737
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017260808532094718,
+      "loss": 1.3452,
+      "step": 3738
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017258904165211794,
+      "loss": 1.163,
+      "step": 3739
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017256999241705025,
+      "loss": 1.3835,
+      "step": 3740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001725509376172049,
+      "loss": 1.6841,
+      "step": 3741
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017253187725404297,
+      "loss": 1.6367,
+      "step": 3742
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017251281132902612,
+      "loss": 2.0379,
+      "step": 3743
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017249373984361624,
+      "loss": 1.0661,
+      "step": 3744
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017247466279927585,
+      "loss": 1.8296,
+      "step": 3745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017245558019746778,
+      "loss": 1.7984,
+      "step": 3746
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017243649203965532,
+      "loss": 1.1284,
+      "step": 3747
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017241739832730214,
+      "loss": 1.8961,
+      "step": 3748
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017239829906187245,
+      "loss": 1.7434,
+      "step": 3749
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001723791942448308,
+      "loss": 1.4114,
+      "step": 3750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017236008387764207,
+      "loss": 1.2836,
+      "step": 3751
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001723409679617718,
+      "loss": 1.4275,
+      "step": 3752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017232184649868573,
+      "loss": 1.0208,
+      "step": 3753
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001723027194898502,
+      "loss": 1.6565,
+      "step": 3754
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017228358693673185,
+      "loss": 1.7234,
+      "step": 3755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017226444884079785,
+      "loss": 1.5358,
+      "step": 3756
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017224530520351566,
+      "loss": 1.9331,
+      "step": 3757
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001722261560263533,
+      "loss": 1.0591,
+      "step": 3758
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017220700131077914,
+      "loss": 1.7794,
+      "step": 3759
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017218784105826199,
+      "loss": 1.5757,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017216867527027106,
+      "loss": 1.542,
+      "step": 3761
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017214950394827606,
+      "loss": 1.5934,
+      "step": 3762
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001721303270937471,
+      "loss": 1.7595,
+      "step": 3763
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001721111447081546,
+      "loss": 2.577,
+      "step": 3764
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017209195679296956,
+      "loss": 1.3216,
+      "step": 3765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017207276334966332,
+      "loss": 1.694,
+      "step": 3766
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017205356437970766,
+      "loss": 1.4901,
+      "step": 3767
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001720343598845748,
+      "loss": 1.5332,
+      "step": 3768
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017201514986573736,
+      "loss": 1.635,
+      "step": 3769
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001719959343246684,
+      "loss": 1.611,
+      "step": 3770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017197671326284138,
+      "loss": 1.9414,
+      "step": 3771
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001719574866817302,
+      "loss": 1.1725,
+      "step": 3772
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001719382545828092,
+      "loss": 1.6962,
+      "step": 3773
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017191901696755313,
+      "loss": 1.8194,
+      "step": 3774
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017189977383743715,
+      "loss": 1.6707,
+      "step": 3775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017188052519393686,
+      "loss": 1.8787,
+      "step": 3776
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017186127103852825,
+      "loss": 2.1375,
+      "step": 3777
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017184201137268777,
+      "loss": 1.7447,
+      "step": 3778
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017182274619789234,
+      "loss": 1.7397,
+      "step": 3779
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017180347551561914,
+      "loss": 1.2947,
+      "step": 3780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017178419932734596,
+      "loss": 1.7495,
+      "step": 3781
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017176491763455087,
+      "loss": 1.717,
+      "step": 3782
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017174563043871247,
+      "loss": 1.7114,
+      "step": 3783
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001717263377413097,
+      "loss": 1.5718,
+      "step": 3784
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001717070395438219,
+      "loss": 2.1534,
+      "step": 3785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000171687735847729,
+      "loss": 1.9808,
+      "step": 3786
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017166842665451118,
+      "loss": 1.7323,
+      "step": 3787
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001716491119656491,
+      "loss": 1.6943,
+      "step": 3788
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001716297917826238,
+      "loss": 2.0052,
+      "step": 3789
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017161046610691686,
+      "loss": 1.4836,
+      "step": 3790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017159113494001017,
+      "loss": 1.5147,
+      "step": 3791
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017157179828338604,
+      "loss": 1.2622,
+      "step": 3792
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017155245613852723,
+      "loss": 1.2202,
+      "step": 3793
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017153310850691702,
+      "loss": 1.0248,
+      "step": 3794
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017151375539003891,
+      "loss": 1.6046,
+      "step": 3795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017149439678937697,
+      "loss": 1.5442,
+      "step": 3796
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017147503270641568,
+      "loss": 1.2424,
+      "step": 3797
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017145566314263983,
+      "loss": 1.3378,
+      "step": 3798
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017143628809953472,
+      "loss": 1.9246,
+      "step": 3799
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017141690757858618,
+      "loss": 1.7951,
+      "step": 3800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017139752158128017,
+      "loss": 1.2247,
+      "step": 3801
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017137813010910334,
+      "loss": 1.7798,
+      "step": 3802
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017135873316354263,
+      "loss": 1.2469,
+      "step": 3803
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017133933074608542,
+      "loss": 1.9075,
+      "step": 3804
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017131992285821955,
+      "loss": 1.5802,
+      "step": 3805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001713005095014332,
+      "loss": 1.396,
+      "step": 3806
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017128109067721502,
+      "loss": 1.3189,
+      "step": 3807
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017126166638705413,
+      "loss": 1.421,
+      "step": 3808
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017124223663243993,
+      "loss": 1.6393,
+      "step": 3809
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017122280141486243,
+      "loss": 1.2745,
+      "step": 3810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017120336073581183,
+      "loss": 2.0386,
+      "step": 3811
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017118391459677897,
+      "loss": 1.3343,
+      "step": 3812
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017116446299925497,
+      "loss": 2.1389,
+      "step": 3813
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001711450059447314,
+      "loss": 1.7378,
+      "step": 3814
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001711255434347003,
+      "loss": 1.5596,
+      "step": 3815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000171106075470654,
+      "loss": 1.1077,
+      "step": 3816
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017108660205408539,
+      "loss": 1.3856,
+      "step": 3817
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017106712318648776,
+      "loss": 1.7391,
+      "step": 3818
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001710476388693547,
+      "loss": 1.6312,
+      "step": 3819
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017102814910418035,
+      "loss": 1.8263,
+      "step": 3820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001710086538924592,
+      "loss": 1.9956,
+      "step": 3821
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017098915323568616,
+      "loss": 1.5347,
+      "step": 3822
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017096964713535657,
+      "loss": 1.2656,
+      "step": 3823
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017095013559296623,
+      "loss": 1.1022,
+      "step": 3824
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017093061861001124,
+      "loss": 1.3901,
+      "step": 3825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017091109618798828,
+      "loss": 1.4347,
+      "step": 3826
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001708915683283943,
+      "loss": 2.1114,
+      "step": 3827
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017087203503272672,
+      "loss": 1.275,
+      "step": 3828
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017085249630248344,
+      "loss": 1.6726,
+      "step": 3829
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017083295213916265,
+      "loss": 1.3069,
+      "step": 3830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017081340254426305,
+      "loss": 0.9226,
+      "step": 3831
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017079384751928376,
+      "loss": 1.3478,
+      "step": 3832
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001707742870657243,
+      "loss": 1.2135,
+      "step": 3833
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001707547211850845,
+      "loss": 1.3917,
+      "step": 3834
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001707351498788648,
+      "loss": 1.6665,
+      "step": 3835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017071557314856596,
+      "loss": 1.4493,
+      "step": 3836
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017069599099568908,
+      "loss": 1.5631,
+      "step": 3837
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001706764034217358,
+      "loss": 1.5228,
+      "step": 3838
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017065681042820812,
+      "loss": 1.5814,
+      "step": 3839
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017063721201660844,
+      "loss": 1.1306,
+      "step": 3840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001706176081884396,
+      "loss": 2.0817,
+      "step": 3841
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017059799894520488,
+      "loss": 1.755,
+      "step": 3842
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001705783842884079,
+      "loss": 1.2655,
+      "step": 3843
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017055876421955278,
+      "loss": 1.2534,
+      "step": 3844
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017053913874014404,
+      "loss": 1.5773,
+      "step": 3845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017051950785168654,
+      "loss": 1.3923,
+      "step": 3846
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017049987155568562,
+      "loss": 1.628,
+      "step": 3847
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017048022985364703,
+      "loss": 1.3862,
+      "step": 3848
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017046058274707693,
+      "loss": 1.4753,
+      "step": 3849
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017044093023748185,
+      "loss": 1.5305,
+      "step": 3850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017042127232636884,
+      "loss": 1.9131,
+      "step": 3851
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017040160901524525,
+      "loss": 1.6732,
+      "step": 3852
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017038194030561893,
+      "loss": 1.8005,
+      "step": 3853
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017036226619899803,
+      "loss": 1.5436,
+      "step": 3854
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001703425866968913,
+      "loss": 1.3579,
+      "step": 3855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017032290180080773,
+      "loss": 1.3616,
+      "step": 3856
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017030321151225678,
+      "loss": 1.4546,
+      "step": 3857
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017028351583274836,
+      "loss": 1.037,
+      "step": 3858
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001702638147637927,
+      "loss": 1.2526,
+      "step": 3859
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017024410830690062,
+      "loss": 1.3527,
+      "step": 3860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001702243964635832,
+      "loss": 1.7727,
+      "step": 3861
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001702046792353519,
+      "loss": 1.5654,
+      "step": 3862
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017018495662371875,
+      "loss": 1.4212,
+      "step": 3863
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017016522863019608,
+      "loss": 2.1158,
+      "step": 3864
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017014549525629668,
+      "loss": 1.7847,
+      "step": 3865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001701257565035337,
+      "loss": 1.1484,
+      "step": 3866
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017010601237342076,
+      "loss": 1.426,
+      "step": 3867
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001700862628674719,
+      "loss": 1.9632,
+      "step": 3868
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017006650798720148,
+      "loss": 1.5838,
+      "step": 3869
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017004674773412438,
+      "loss": 1.6111,
+      "step": 3870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001700269821097558,
+      "loss": 0.9587,
+      "step": 3871
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017000721111561147,
+      "loss": 1.6394,
+      "step": 3872
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016998743475320736,
+      "loss": 1.9266,
+      "step": 3873
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001699676530240601,
+      "loss": 1.265,
+      "step": 3874
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016994786592968643,
+      "loss": 1.2646,
+      "step": 3875
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016992807347160376,
+      "loss": 1.221,
+      "step": 3876
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016990827565132975,
+      "loss": 1.2601,
+      "step": 3877
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016988847247038252,
+      "loss": 1.63,
+      "step": 3878
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016986866393028065,
+      "loss": 1.5749,
+      "step": 3879
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016984885003254308,
+      "loss": 1.538,
+      "step": 3880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016982903077868912,
+      "loss": 1.3592,
+      "step": 3881
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001698092061702386,
+      "loss": 1.1865,
+      "step": 3882
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016978937620871172,
+      "loss": 1.035,
+      "step": 3883
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016976954089562897,
+      "loss": 1.6849,
+      "step": 3884
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016974970023251148,
+      "loss": 1.5469,
+      "step": 3885
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016972985422088055,
+      "loss": 1.1587,
+      "step": 3886
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016971000286225806,
+      "loss": 2.0213,
+      "step": 3887
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016969014615816623,
+      "loss": 1.4337,
+      "step": 3888
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001696702841101277,
+      "loss": 0.949,
+      "step": 3889
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016965041671966553,
+      "loss": 1.2585,
+      "step": 3890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016963054398830317,
+      "loss": 1.5292,
+      "step": 3891
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016961066591756452,
+      "loss": 1.7197,
+      "step": 3892
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016959078250897383,
+      "loss": 1.5102,
+      "step": 3893
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016957089376405578,
+      "loss": 1.6501,
+      "step": 3894
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016955099968433553,
+      "loss": 1.09,
+      "step": 3895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016953110027133854,
+      "loss": 1.294,
+      "step": 3896
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016951119552659072,
+      "loss": 1.4886,
+      "step": 3897
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016949128545161842,
+      "loss": 1.5328,
+      "step": 3898
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016947137004794837,
+      "loss": 1.443,
+      "step": 3899
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016945144931710773,
+      "loss": 1.0704,
+      "step": 3900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016943152326062398,
+      "loss": 1.4946,
+      "step": 3901
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016941159188002514,
+      "loss": 1.8438,
+      "step": 3902
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001693916551768396,
+      "loss": 1.3978,
+      "step": 3903
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016937171315259611,
+      "loss": 1.9206,
+      "step": 3904
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016935176580882386,
+      "loss": 1.5287,
+      "step": 3905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001693318131470524,
+      "loss": 1.7454,
+      "step": 3906
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001693118551688118,
+      "loss": 1.4235,
+      "step": 3907
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016929189187563242,
+      "loss": 1.4573,
+      "step": 3908
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001692719232690451,
+      "loss": 1.0932,
+      "step": 3909
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000169251949350581,
+      "loss": 0.9546,
+      "step": 3910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001692319701217719,
+      "loss": 1.1685,
+      "step": 3911
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016921198558414967,
+      "loss": 2.1555,
+      "step": 3912
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016919199573924685,
+      "loss": 1.2987,
+      "step": 3913
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016917200058859627,
+      "loss": 1.6931,
+      "step": 3914
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016915200013373116,
+      "loss": 1.7782,
+      "step": 3915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016913199437618522,
+      "loss": 1.6008,
+      "step": 3916
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016911198331749254,
+      "loss": 1.5237,
+      "step": 3917
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016909196695918754,
+      "loss": 1.5678,
+      "step": 3918
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016907194530280516,
+      "loss": 1.8029,
+      "step": 3919
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016905191834988068,
+      "loss": 1.422,
+      "step": 3920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001690318861019498,
+      "loss": 1.8522,
+      "step": 3921
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016901184856054857,
+      "loss": 1.6441,
+      "step": 3922
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016899180572721353,
+      "loss": 1.8898,
+      "step": 3923
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016897175760348162,
+      "loss": 1.7557,
+      "step": 3924
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016895170419089017,
+      "loss": 1.9797,
+      "step": 3925
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016893164549097684,
+      "loss": 1.879,
+      "step": 3926
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016891158150527982,
+      "loss": 2.0303,
+      "step": 3927
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016889151223533765,
+      "loss": 1.0365,
+      "step": 3928
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016887143768268923,
+      "loss": 1.194,
+      "step": 3929
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001688513578488739,
+      "loss": 1.4635,
+      "step": 3930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001688312727354315,
+      "loss": 1.4843,
+      "step": 3931
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016881118234390212,
+      "loss": 1.2367,
+      "step": 3932
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001687910866758263,
+      "loss": 1.5085,
+      "step": 3933
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016877098573274505,
+      "loss": 1.6615,
+      "step": 3934
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016875087951619973,
+      "loss": 1.5805,
+      "step": 3935
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001687307680277321,
+      "loss": 1.7581,
+      "step": 3936
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016871065126888434,
+      "loss": 1.3172,
+      "step": 3937
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016869052924119908,
+      "loss": 1.4831,
+      "step": 3938
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016867040194621925,
+      "loss": 1.1083,
+      "step": 3939
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001686502693854883,
+      "loss": 1.7381,
+      "step": 3940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016863013156054996,
+      "loss": 1.5153,
+      "step": 3941
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016860998847294847,
+      "loss": 1.5945,
+      "step": 3942
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016858984012422843,
+      "loss": 0.9899,
+      "step": 3943
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016856968651593484,
+      "loss": 1.8497,
+      "step": 3944
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001685495276496131,
+      "loss": 1.1929,
+      "step": 3945
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016852936352680905,
+      "loss": 1.6462,
+      "step": 3946
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016850919414906885,
+      "loss": 1.2632,
+      "step": 3947
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016848901951793915,
+      "loss": 1.6296,
+      "step": 3948
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016846883963496704,
+      "loss": 1.5545,
+      "step": 3949
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016844865450169982,
+      "loss": 1.6548,
+      "step": 3950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016842846411968543,
+      "loss": 1.8507,
+      "step": 3951
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000168408268490472,
+      "loss": 1.5989,
+      "step": 3952
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001683880676156082,
+      "loss": 1.567,
+      "step": 3953
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016836786149664312,
+      "loss": 1.8893,
+      "step": 3954
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016834765013512613,
+      "loss": 0.8621,
+      "step": 3955
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001683274335326071,
+      "loss": 1.1372,
+      "step": 3956
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016830721169063624,
+      "loss": 1.5072,
+      "step": 3957
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001682869846107642,
+      "loss": 1.4236,
+      "step": 3958
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016826675229454204,
+      "loss": 1.615,
+      "step": 3959
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016824651474352123,
+      "loss": 1.4895,
+      "step": 3960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016822627195925356,
+      "loss": 1.9765,
+      "step": 3961
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001682060239432913,
+      "loss": 1.3581,
+      "step": 3962
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016818577069718706,
+      "loss": 1.5879,
+      "step": 3963
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016816551222249398,
+      "loss": 1.3721,
+      "step": 3964
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016814524852076543,
+      "loss": 2.0738,
+      "step": 3965
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001681249795935553,
+      "loss": 1.7689,
+      "step": 3966
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016810470544241786,
+      "loss": 1.6046,
+      "step": 3967
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001680844260689077,
+      "loss": 2.1687,
+      "step": 3968
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001680641414745799,
+      "loss": 1.868,
+      "step": 3969
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016804385166098996,
+      "loss": 1.5615,
+      "step": 3970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001680235566296937,
+      "loss": 1.9307,
+      "step": 3971
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016800325638224734,
+      "loss": 1.4533,
+      "step": 3972
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001679829509202076,
+      "loss": 1.8774,
+      "step": 3973
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016796264024513148,
+      "loss": 1.2487,
+      "step": 3974
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016794232435857643,
+      "loss": 1.4868,
+      "step": 3975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016792200326210037,
+      "loss": 1.5271,
+      "step": 3976
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001679016769572615,
+      "loss": 1.663,
+      "step": 3977
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016788134544561847,
+      "loss": 1.2055,
+      "step": 3978
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016786100872873033,
+      "loss": 1.6913,
+      "step": 3979
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016784066680815656,
+      "loss": 1.5653,
+      "step": 3980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016782031968545696,
+      "loss": 1.749,
+      "step": 3981
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016779996736219187,
+      "loss": 1.7403,
+      "step": 3982
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016777960983992181,
+      "loss": 1.9367,
+      "step": 3983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016775924712020794,
+      "loss": 1.5179,
+      "step": 3984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016773887920461163,
+      "loss": 0.8495,
+      "step": 3985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016771850609469477,
+      "loss": 1.3669,
+      "step": 3986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016769812779201957,
+      "loss": 1.6524,
+      "step": 3987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001676777442981487,
+      "loss": 1.8718,
+      "step": 3988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016765735561464516,
+      "loss": 1.1111,
+      "step": 3989
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001676369617430724,
+      "loss": 1.5167,
+      "step": 3990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016761656268499426,
+      "loss": 1.3236,
+      "step": 3991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016759615844197495,
+      "loss": 1.5616,
+      "step": 3992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016757574901557915,
+      "loss": 1.8161,
+      "step": 3993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016755533440737183,
+      "loss": 1.6989,
+      "step": 3994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016753491461891847,
+      "loss": 1.3088,
+      "step": 3995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001675144896517848,
+      "loss": 1.8343,
+      "step": 3996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016749405950753713,
+      "loss": 1.5881,
+      "step": 3997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016747362418774206,
+      "loss": 1.3318,
+      "step": 3998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016745318369396653,
+      "loss": 1.2953,
+      "step": 3999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016743273802777804,
+      "loss": 1.12,
+      "step": 4000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016741228719074433,
+      "loss": 1.1119,
+      "step": 4001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016739183118443361,
+      "loss": 1.9955,
+      "step": 4002
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001673713700104145,
+      "loss": 1.5312,
+      "step": 4003
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000167350903670256,
+      "loss": 1.3951,
+      "step": 4004
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001673304321655275,
+      "loss": 1.4001,
+      "step": 4005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016730995549779873,
+      "loss": 1.3952,
+      "step": 4006
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016728947366863995,
+      "loss": 1.4151,
+      "step": 4007
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016726898667962167,
+      "loss": 1.5828,
+      "step": 4008
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016724849453231494,
+      "loss": 1.1572,
+      "step": 4009
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016722799722829106,
+      "loss": 1.6244,
+      "step": 4010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016720749476912183,
+      "loss": 1.7511,
+      "step": 4011
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016718698715637937,
+      "loss": 1.3615,
+      "step": 4012
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016716647439163628,
+      "loss": 1.3891,
+      "step": 4013
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016714595647646552,
+      "loss": 1.4626,
+      "step": 4014
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016712543341244036,
+      "loss": 1.7682,
+      "step": 4015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016710490520113462,
+      "loss": 1.717,
+      "step": 4016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016708437184412243,
+      "loss": 1.5525,
+      "step": 4017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016706383334297824,
+      "loss": 1.2658,
+      "step": 4018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016704328969927707,
+      "loss": 1.875,
+      "step": 4019
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016702274091459418,
+      "loss": 1.6868,
+      "step": 4020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016700218699050527,
+      "loss": 1.0298,
+      "step": 4021
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001669816279285865,
+      "loss": 1.756,
+      "step": 4022
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016696106373041434,
+      "loss": 1.7794,
+      "step": 4023
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001669404943975657,
+      "loss": 2.1811,
+      "step": 4024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016691991993161783,
+      "loss": 1.7594,
+      "step": 4025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016689934033414843,
+      "loss": 1.2903,
+      "step": 4026
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016687875560673562,
+      "loss": 2.0262,
+      "step": 4027
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001668581657509578,
+      "loss": 1.1442,
+      "step": 4028
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016683757076839387,
+      "loss": 1.7444,
+      "step": 4029
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016681697066062308,
+      "loss": 1.5631,
+      "step": 4030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016679636542922505,
+      "loss": 1.5175,
+      "step": 4031
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016677575507577984,
+      "loss": 1.3641,
+      "step": 4032
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016675513960186795,
+      "loss": 1.6624,
+      "step": 4033
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016673451900907008,
+      "loss": 1.2439,
+      "step": 4034
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016671389329896757,
+      "loss": 1.3632,
+      "step": 4035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016669326247314192,
+      "loss": 1.43,
+      "step": 4036
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001666726265331752,
+      "loss": 1.586,
+      "step": 4037
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016665198548064983,
+      "loss": 1.9063,
+      "step": 4038
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001666313393171485,
+      "loss": 1.5163,
+      "step": 4039
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016661068804425447,
+      "loss": 1.2713,
+      "step": 4040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001665900316635513,
+      "loss": 1.1872,
+      "step": 4041
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016656937017662293,
+      "loss": 1.3645,
+      "step": 4042
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016654870358505375,
+      "loss": 1.5141,
+      "step": 4043
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016652803189042844,
+      "loss": 1.6826,
+      "step": 4044
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001665073550943322,
+      "loss": 1.3684,
+      "step": 4045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016648667319835053,
+      "loss": 1.1258,
+      "step": 4046
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001664659862040694,
+      "loss": 1.6053,
+      "step": 4047
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016644529411307502,
+      "loss": 2.0418,
+      "step": 4048
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001664245969269542,
+      "loss": 1.6717,
+      "step": 4049
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016640389464729394,
+      "loss": 1.3412,
+      "step": 4050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016638318727568176,
+      "loss": 1.5681,
+      "step": 4051
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016636247481370556,
+      "loss": 1.9614,
+      "step": 4052
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016634175726295354,
+      "loss": 1.1017,
+      "step": 4053
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016632103462501445,
+      "loss": 1.6995,
+      "step": 4054
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016630030690147724,
+      "loss": 1.5241,
+      "step": 4055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016627957409393138,
+      "loss": 1.2517,
+      "step": 4056
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016625883620396669,
+      "loss": 1.2976,
+      "step": 4057
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016623809323317338,
+      "loss": 1.2585,
+      "step": 4058
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001662173451831421,
+      "loss": 2.2391,
+      "step": 4059
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016619659205546373,
+      "loss": 1.7999,
+      "step": 4060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016617583385172976,
+      "loss": 1.7669,
+      "step": 4061
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016615507057353193,
+      "loss": 1.0904,
+      "step": 4062
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016613430222246238,
+      "loss": 1.1294,
+      "step": 4063
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016611352880011366,
+      "loss": 1.7963,
+      "step": 4064
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016609275030807875,
+      "loss": 2.089,
+      "step": 4065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001660719667479509,
+      "loss": 1.9024,
+      "step": 4066
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001660511781213239,
+      "loss": 1.6869,
+      "step": 4067
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016603038442979183,
+      "loss": 1.0054,
+      "step": 4068
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016600958567494916,
+      "loss": 1.1134,
+      "step": 4069
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016598878185839083,
+      "loss": 1.1739,
+      "step": 4070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016596797298171198,
+      "loss": 1.4476,
+      "step": 4071
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016594715904650841,
+      "loss": 1.457,
+      "step": 4072
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001659263400543761,
+      "loss": 1.3792,
+      "step": 4073
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016590551600691145,
+      "loss": 1.6675,
+      "step": 4074
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016588468690571136,
+      "loss": 1.364,
+      "step": 4075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016586385275237298,
+      "loss": 1.2177,
+      "step": 4076
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001658430135484939,
+      "loss": 1.3619,
+      "step": 4077
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016582216929567217,
+      "loss": 1.2934,
+      "step": 4078
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016580131999550607,
+      "loss": 1.354,
+      "step": 4079
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001657804656495944,
+      "loss": 1.5641,
+      "step": 4080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016575960625953628,
+      "loss": 1.6267,
+      "step": 4081
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016573874182693128,
+      "loss": 1.5305,
+      "step": 4082
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001657178723533793,
+      "loss": 1.0938,
+      "step": 4083
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016569699784048063,
+      "loss": 1.3521,
+      "step": 4084
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016567611828983595,
+      "loss": 1.3322,
+      "step": 4085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016565523370304638,
+      "loss": 1.2119,
+      "step": 4086
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016563434408171335,
+      "loss": 1.8702,
+      "step": 4087
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001656134494274387,
+      "loss": 1.5516,
+      "step": 4088
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001655925497418247,
+      "loss": 1.6302,
+      "step": 4089
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016557164502647393,
+      "loss": 1.2764,
+      "step": 4090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016555073528298943,
+      "loss": 1.4508,
+      "step": 4091
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016552982051297455,
+      "loss": 1.7133,
+      "step": 4092
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001655089007180331,
+      "loss": 1.5352,
+      "step": 4093
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016548797589976925,
+      "loss": 1.5594,
+      "step": 4094
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001654670460597875,
+      "loss": 1.3248,
+      "step": 4095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016544611119969282,
+      "loss": 1.3278,
+      "step": 4096
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016542517132109053,
+      "loss": 1.8791,
+      "step": 4097
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001654042264255863,
+      "loss": 1.2602,
+      "step": 4098
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016538327651478623,
+      "loss": 1.9275,
+      "step": 4099
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016536232159029683,
+      "loss": 1.3353,
+      "step": 4100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016534136165372493,
+      "loss": 1.2731,
+      "step": 4101
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001653203967066777,
+      "loss": 1.9952,
+      "step": 4102
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001652994267507629,
+      "loss": 1.5366,
+      "step": 4103
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016527845178758842,
+      "loss": 1.8348,
+      "step": 4104
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016525747181876272,
+      "loss": 2.0433,
+      "step": 4105
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016523648684589452,
+      "loss": 1.8406,
+      "step": 4106
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016521549687059306,
+      "loss": 0.9093,
+      "step": 4107
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001651945018944678,
+      "loss": 1.2421,
+      "step": 4108
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016517350191912872,
+      "loss": 1.5073,
+      "step": 4109
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016515249694618609,
+      "loss": 1.5014,
+      "step": 4110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016513148697725065,
+      "loss": 1.7786,
+      "step": 4111
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016511047201393347,
+      "loss": 1.226,
+      "step": 4112
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016508945205784595,
+      "loss": 2.0783,
+      "step": 4113
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001650684271106,
+      "loss": 1.4653,
+      "step": 4114
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016504739717380782,
+      "loss": 1.4208,
+      "step": 4115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016502636224908202,
+      "loss": 1.364,
+      "step": 4116
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016500532233803557,
+      "loss": 1.6076,
+      "step": 4117
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016498427744228187,
+      "loss": 1.1318,
+      "step": 4118
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016496322756343466,
+      "loss": 1.5296,
+      "step": 4119
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001649421727031081,
+      "loss": 0.9793,
+      "step": 4120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016492111286291666,
+      "loss": 1.1094,
+      "step": 4121
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016490004804447527,
+      "loss": 1.6084,
+      "step": 4122
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016487897824939927,
+      "loss": 1.6556,
+      "step": 4123
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001648579034793042,
+      "loss": 1.2324,
+      "step": 4124
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016483682373580623,
+      "loss": 1.9155,
+      "step": 4125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016481573902052169,
+      "loss": 1.6588,
+      "step": 4126
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016479464933506744,
+      "loss": 1.1284,
+      "step": 4127
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016477355468106066,
+      "loss": 1.8605,
+      "step": 4128
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001647524550601189,
+      "loss": 1.8372,
+      "step": 4129
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016473135047386013,
+      "loss": 1.4134,
+      "step": 4130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001647102409239027,
+      "loss": 1.9753,
+      "step": 4131
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016468912641186528,
+      "loss": 1.5886,
+      "step": 4132
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016466800693936703,
+      "loss": 0.9727,
+      "step": 4133
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016464688250802733,
+      "loss": 1.8126,
+      "step": 4134
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001646257531194661,
+      "loss": 1.6012,
+      "step": 4135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016460461877530356,
+      "loss": 1.1623,
+      "step": 4136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016458347947716033,
+      "loss": 1.4817,
+      "step": 4137
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016456233522665737,
+      "loss": 1.7243,
+      "step": 4138
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001645411860254161,
+      "loss": 1.2676,
+      "step": 4139
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016452003187505823,
+      "loss": 1.8546,
+      "step": 4140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016449887277720596,
+      "loss": 1.4384,
+      "step": 4141
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001644777087334817,
+      "loss": 1.7021,
+      "step": 4142
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001644565397455084,
+      "loss": 1.3638,
+      "step": 4143
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016443536581490936,
+      "loss": 1.5556,
+      "step": 4144
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016441418694330818,
+      "loss": 1.5182,
+      "step": 4145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001643930031323289,
+      "loss": 2.0858,
+      "step": 4146
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016437181438359592,
+      "loss": 1.7436,
+      "step": 4147
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016435062069873407,
+      "loss": 1.4292,
+      "step": 4148
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016432942207936842,
+      "loss": 1.312,
+      "step": 4149
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016430821852712465,
+      "loss": 0.2409,
+      "step": 4150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016428701004362855,
+      "loss": 0.9955,
+      "step": 4151
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016426579663050646,
+      "loss": 1.7941,
+      "step": 4152
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001642445782893851,
+      "loss": 1.2914,
+      "step": 4153
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001642233550218915,
+      "loss": 1.9238,
+      "step": 4154
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016420212682965307,
+      "loss": 1.5274,
+      "step": 4155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016418089371429763,
+      "loss": 1.8961,
+      "step": 4156
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016415965567745342,
+      "loss": 1.6366,
+      "step": 4157
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001641384127207489,
+      "loss": 1.6833,
+      "step": 4158
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001641171648458131,
+      "loss": 1.4922,
+      "step": 4159
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016409591205427532,
+      "loss": 1.1931,
+      "step": 4160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016407465434776524,
+      "loss": 1.9356,
+      "step": 4161
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016405339172791293,
+      "loss": 1.762,
+      "step": 4162
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016403212419634886,
+      "loss": 1.1528,
+      "step": 4163
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016401085175470385,
+      "loss": 1.6083,
+      "step": 4164
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001639895744046091,
+      "loss": 1.6775,
+      "step": 4165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001639682921476962,
+      "loss": 1.5188,
+      "step": 4166
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016394700498559707,
+      "loss": 1.4443,
+      "step": 4167
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016392571291994407,
+      "loss": 1.2263,
+      "step": 4168
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016390441595236994,
+      "loss": 1.4349,
+      "step": 4169
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001638831140845077,
+      "loss": 1.0992,
+      "step": 4170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016386180731799087,
+      "loss": 1.7504,
+      "step": 4171
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016384049565445322,
+      "loss": 1.5451,
+      "step": 4172
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016381917909552901,
+      "loss": 1.2476,
+      "step": 4173
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001637978576428528,
+      "loss": 1.9403,
+      "step": 4174
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001637765312980596,
+      "loss": 1.2119,
+      "step": 4175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016375520006278472,
+      "loss": 1.9593,
+      "step": 4176
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001637338639386638,
+      "loss": 0.9408,
+      "step": 4177
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016371252292733302,
+      "loss": 1.9576,
+      "step": 4178
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016369117703042885,
+      "loss": 1.81,
+      "step": 4179
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016366982624958803,
+      "loss": 1.8504,
+      "step": 4180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016364847058644786,
+      "loss": 1.8193,
+      "step": 4181
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016362711004264585,
+      "loss": 1.3557,
+      "step": 4182
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016360574461982003,
+      "loss": 0.9948,
+      "step": 4183
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016358437431960874,
+      "loss": 1.5648,
+      "step": 4184
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016356299914365062,
+      "loss": 1.5839,
+      "step": 4185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016354161909358476,
+      "loss": 1.9816,
+      "step": 4186
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016352023417105067,
+      "loss": 1.5328,
+      "step": 4187
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001634988443776881,
+      "loss": 1.4259,
+      "step": 4188
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016347744971513734,
+      "loss": 1.5451,
+      "step": 4189
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001634560501850389,
+      "loss": 1.2198,
+      "step": 4190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001634346457890338,
+      "loss": 1.518,
+      "step": 4191
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016341323652876324,
+      "loss": 1.2515,
+      "step": 4192
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016339182240586903,
+      "loss": 1.5902,
+      "step": 4193
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016337040342199322,
+      "loss": 1.3995,
+      "step": 4194
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016334897957877818,
+      "loss": 1.3338,
+      "step": 4195
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016332755087786676,
+      "loss": 1.3312,
+      "step": 4196
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001633061173209022,
+      "loss": 1.5074,
+      "step": 4197
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000163284678909528,
+      "loss": 1.2445,
+      "step": 4198
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016326323564538808,
+      "loss": 1.6562,
+      "step": 4199
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016324178753012678,
+      "loss": 1.7409,
+      "step": 4200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016322033456538877,
+      "loss": 1.6525,
+      "step": 4201
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016319887675281906,
+      "loss": 1.8813,
+      "step": 4202
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001631774140940631,
+      "loss": 1.5614,
+      "step": 4203
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001631559465907667,
+      "loss": 1.1121,
+      "step": 4204
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016313447424457596,
+      "loss": 0.9694,
+      "step": 4205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016311299705713746,
+      "loss": 1.485,
+      "step": 4206
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016309151503009808,
+      "loss": 1.9271,
+      "step": 4207
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001630700281651051,
+      "loss": 1.7,
+      "step": 4208
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016304853646380619,
+      "loss": 1.9077,
+      "step": 4209
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016302703992784933,
+      "loss": 2.106,
+      "step": 4210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001630055385588829,
+      "loss": 2.0878,
+      "step": 4211
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001629840323585557,
+      "loss": 1.3098,
+      "step": 4212
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016296252132851687,
+      "loss": 1.4225,
+      "step": 4213
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001629410054704158,
+      "loss": 1.2827,
+      "step": 4214
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016291948478590253,
+      "loss": 1.7047,
+      "step": 4215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016289795927662714,
+      "loss": 1.8174,
+      "step": 4216
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016287642894424035,
+      "loss": 1.6246,
+      "step": 4217
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016285489379039305,
+      "loss": 1.9514,
+      "step": 4218
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016283335381673663,
+      "loss": 2.0495,
+      "step": 4219
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016281180902492283,
+      "loss": 1.4718,
+      "step": 4220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016279025941660374,
+      "loss": 1.244,
+      "step": 4221
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016276870499343175,
+      "loss": 1.5895,
+      "step": 4222
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016274714575705975,
+      "loss": 1.5063,
+      "step": 4223
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001627255817091409,
+      "loss": 1.132,
+      "step": 4224
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016270401285132876,
+      "loss": 1.3767,
+      "step": 4225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016268243918527735,
+      "loss": 0.8568,
+      "step": 4226
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016266086071264083,
+      "loss": 2.4428,
+      "step": 4227
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016263927743507398,
+      "loss": 1.3307,
+      "step": 4228
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001626176893542318,
+      "loss": 1.7523,
+      "step": 4229
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001625960964717697,
+      "loss": 1.0308,
+      "step": 4230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016257449878934345,
+      "loss": 1.5499,
+      "step": 4231
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001625528963086092,
+      "loss": 1.7359,
+      "step": 4232
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016253128903122343,
+      "loss": 1.6947,
+      "step": 4233
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001625096769588431,
+      "loss": 1.7549,
+      "step": 4234
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016248806009312537,
+      "loss": 1.5994,
+      "step": 4235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001624664384357279,
+      "loss": 0.8518,
+      "step": 4236
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016244481198830863,
+      "loss": 1.4071,
+      "step": 4237
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016242318075252596,
+      "loss": 1.3713,
+      "step": 4238
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016240154473003862,
+      "loss": 1.2612,
+      "step": 4239
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016237990392250562,
+      "loss": 1.6741,
+      "step": 4240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016235825833158643,
+      "loss": 1.6972,
+      "step": 4241
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016233660795894093,
+      "loss": 1.2359,
+      "step": 4242
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016231495280622922,
+      "loss": 1.646,
+      "step": 4243
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016229329287511193,
+      "loss": 1.2241,
+      "step": 4244
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016227162816724987,
+      "loss": 2.2764,
+      "step": 4245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016224995868430444,
+      "loss": 1.0607,
+      "step": 4246
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016222828442793725,
+      "loss": 1.4384,
+      "step": 4247
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016220660539981028,
+      "loss": 2.0111,
+      "step": 4248
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001621849216015859,
+      "loss": 1.1745,
+      "step": 4249
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016216323303492697,
+      "loss": 2.0944,
+      "step": 4250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016214153970149647,
+      "loss": 1.1858,
+      "step": 4251
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016211984160295793,
+      "loss": 1.5305,
+      "step": 4252
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001620981387409752,
+      "loss": 2.3911,
+      "step": 4253
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016207643111721247,
+      "loss": 1.127,
+      "step": 4254
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016205471873333434,
+      "loss": 1.9126,
+      "step": 4255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016203300159100574,
+      "loss": 1.5537,
+      "step": 4256
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016201127969189195,
+      "loss": 1.5031,
+      "step": 4257
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001619895530376587,
+      "loss": 1.2618,
+      "step": 4258
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001619678216299719,
+      "loss": 1.419,
+      "step": 4259
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016194608547049808,
+      "loss": 1.5569,
+      "step": 4260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016192434456090396,
+      "loss": 1.5264,
+      "step": 4261
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016190259890285662,
+      "loss": 0.9215,
+      "step": 4262
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001618808484980236,
+      "loss": 2.0176,
+      "step": 4263
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016185909334807274,
+      "loss": 1.856,
+      "step": 4264
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016183733345467226,
+      "loss": 1.6472,
+      "step": 4265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016181556881949077,
+      "loss": 2.1209,
+      "step": 4266
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016179379944419712,
+      "loss": 1.5088,
+      "step": 4267
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016177202533046073,
+      "loss": 1.3936,
+      "step": 4268
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016175024647995122,
+      "loss": 0.7817,
+      "step": 4269
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016172846289433862,
+      "loss": 1.8262,
+      "step": 4270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016170667457529334,
+      "loss": 1.208,
+      "step": 4271
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016168488152448616,
+      "loss": 1.4042,
+      "step": 4272
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016166308374358814,
+      "loss": 1.846,
+      "step": 4273
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016164128123427086,
+      "loss": 1.3377,
+      "step": 4274
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016161947399820608,
+      "loss": 1.6662,
+      "step": 4275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001615976620370661,
+      "loss": 1.864,
+      "step": 4276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016157584535252343,
+      "loss": 1.8105,
+      "step": 4277
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016155402394625102,
+      "loss": 1.7227,
+      "step": 4278
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016153219781992218,
+      "loss": 1.1004,
+      "step": 4279
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016151036697521057,
+      "loss": 1.5094,
+      "step": 4280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016148853141379022,
+      "loss": 1.3595,
+      "step": 4281
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001614666911373355,
+      "loss": 1.5332,
+      "step": 4282
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016144484614752112,
+      "loss": 1.1952,
+      "step": 4283
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016142299644602223,
+      "loss": 2.6078,
+      "step": 4284
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016140114203451434,
+      "loss": 1.4578,
+      "step": 4285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016137928291467318,
+      "loss": 1.6943,
+      "step": 4286
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016135741908817505,
+      "loss": 1.7986,
+      "step": 4287
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016133555055669637,
+      "loss": 1.2063,
+      "step": 4288
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001613136773219142,
+      "loss": 1.958,
+      "step": 4289
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001612917993855057,
+      "loss": 1.6454,
+      "step": 4290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016126991674914854,
+      "loss": 1.8261,
+      "step": 4291
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016124802941452072,
+      "loss": 1.2109,
+      "step": 4292
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001612261373833006,
+      "loss": 1.7536,
+      "step": 4293
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016120424065716688,
+      "loss": 1.3697,
+      "step": 4294
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016118233923779866,
+      "loss": 1.3832,
+      "step": 4295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016116043312687533,
+      "loss": 1.1628,
+      "step": 4296
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016113852232607675,
+      "loss": 1.3839,
+      "step": 4297
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000161116606837083,
+      "loss": 1.3721,
+      "step": 4298
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001610946866615746,
+      "loss": 1.4066,
+      "step": 4299
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001610727618012325,
+      "loss": 1.6096,
+      "step": 4300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016105083225773786,
+      "loss": 1.3801,
+      "step": 4301
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016102889803277227,
+      "loss": 1.5089,
+      "step": 4302
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016100695912801773,
+      "loss": 1.2352,
+      "step": 4303
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016098501554515652,
+      "loss": 1.3088,
+      "step": 4304
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001609630672858713,
+      "loss": 1.5421,
+      "step": 4305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016094111435184508,
+      "loss": 1.4136,
+      "step": 4306
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016091915674476128,
+      "loss": 1.3784,
+      "step": 4307
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016089719446630362,
+      "loss": 1.2983,
+      "step": 4308
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016087522751815625,
+      "loss": 1.1158,
+      "step": 4309
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016085325590200355,
+      "loss": 1.5686,
+      "step": 4310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016083127961953036,
+      "loss": 1.9779,
+      "step": 4311
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016080929867242193,
+      "loss": 0.936,
+      "step": 4312
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001607873130623637,
+      "loss": 1.4634,
+      "step": 4313
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001607653227910416,
+      "loss": 1.9132,
+      "step": 4314
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016074332786014187,
+      "loss": 1.8591,
+      "step": 4315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001607213282713511,
+      "loss": 1.1763,
+      "step": 4316
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001606993240263563,
+      "loss": 1.8848,
+      "step": 4317
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016067731512684475,
+      "loss": 1.2738,
+      "step": 4318
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016065530157450414,
+      "loss": 1.4822,
+      "step": 4319
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016063328337102248,
+      "loss": 1.3827,
+      "step": 4320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016061126051808818,
+      "loss": 1.5131,
+      "step": 4321
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016058923301739,
+      "loss": 1.9826,
+      "step": 4322
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016056720087061698,
+      "loss": 1.2868,
+      "step": 4323
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016054516407945868,
+      "loss": 1.3091,
+      "step": 4324
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016052312264560483,
+      "loss": 1.8158,
+      "step": 4325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016050107657074563,
+      "loss": 1.4204,
+      "step": 4326
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001604790258565716,
+      "loss": 1.5272,
+      "step": 4327
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016045697050477361,
+      "loss": 0.6002,
+      "step": 4328
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016043491051704296,
+      "loss": 0.8522,
+      "step": 4329
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016041284589507116,
+      "loss": 1.6724,
+      "step": 4330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001603907766405502,
+      "loss": 1.6392,
+      "step": 4331
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016036870275517242,
+      "loss": 1.7448,
+      "step": 4332
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016034662424063042,
+      "loss": 1.3262,
+      "step": 4333
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016032454109861724,
+      "loss": 1.3153,
+      "step": 4334
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016030245333082622,
+      "loss": 1.165,
+      "step": 4335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016028036093895112,
+      "loss": 1.6199,
+      "step": 4336
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016025826392468604,
+      "loss": 1.7361,
+      "step": 4337
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016023616228972532,
+      "loss": 1.1824,
+      "step": 4338
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016021405603576383,
+      "loss": 1.3187,
+      "step": 4339
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001601919451644967,
+      "loss": 2.1939,
+      "step": 4340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016016982967761943,
+      "loss": 1.9405,
+      "step": 4341
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016014770957682782,
+      "loss": 1.1911,
+      "step": 4342
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001601255848638181,
+      "loss": 1.6891,
+      "step": 4343
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016010345554028684,
+      "loss": 1.6075,
+      "step": 4344
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016008132160793097,
+      "loss": 1.7262,
+      "step": 4345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016005918306844768,
+      "loss": 1.1615,
+      "step": 4346
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016003703992353464,
+      "loss": 1.7696,
+      "step": 4347
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016001489217488986,
+      "loss": 1.2265,
+      "step": 4348
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001599927398242116,
+      "loss": 1.7458,
+      "step": 4349
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001599705828731985,
+      "loss": 2.6419,
+      "step": 4350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015994842132354968,
+      "loss": 1.5204,
+      "step": 4351
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015992625517696448,
+      "loss": 1.4865,
+      "step": 4352
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015990408443514265,
+      "loss": 1.6421,
+      "step": 4353
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015988190909978423,
+      "loss": 1.6751,
+      "step": 4354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001598597291725897,
+      "loss": 1.3137,
+      "step": 4355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015983754465525988,
+      "loss": 1.4511,
+      "step": 4356
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015981535554949585,
+      "loss": 1.5123,
+      "step": 4357
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015979316185699914,
+      "loss": 1.5074,
+      "step": 4358
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015977096357947155,
+      "loss": 1.7185,
+      "step": 4359
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015974876071861535,
+      "loss": 1.1233,
+      "step": 4360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015972655327613303,
+      "loss": 1.278,
+      "step": 4361
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015970434125372752,
+      "loss": 1.5883,
+      "step": 4362
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015968212465310207,
+      "loss": 1.0159,
+      "step": 4363
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015965990347596026,
+      "loss": 1.784,
+      "step": 4364
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015963767772400606,
+      "loss": 1.1921,
+      "step": 4365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001596154473989438,
+      "loss": 1.1335,
+      "step": 4366
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001595932125024781,
+      "loss": 1.0423,
+      "step": 4367
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015957097303631398,
+      "loss": 1.5274,
+      "step": 4368
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015954872900215676,
+      "loss": 1.8946,
+      "step": 4369
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001595264804017122,
+      "loss": 1.5208,
+      "step": 4370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015950422723668637,
+      "loss": 1.2712,
+      "step": 4371
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001594819695087856,
+      "loss": 2.1641,
+      "step": 4372
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015945970721971671,
+      "loss": 1.3823,
+      "step": 4373
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015943744037118677,
+      "loss": 1.2491,
+      "step": 4374
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015941516896490328,
+      "loss": 1.5668,
+      "step": 4375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015939289300257397,
+      "loss": 1.5813,
+      "step": 4376
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015937061248590709,
+      "loss": 2.4632,
+      "step": 4377
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015934832741661107,
+      "loss": 1.4583,
+      "step": 4378
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015932603779639478,
+      "loss": 1.8777,
+      "step": 4379
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015930374362696745,
+      "loss": 1.9137,
+      "step": 4380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001592814449100386,
+      "loss": 1.3439,
+      "step": 4381
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015925914164731814,
+      "loss": 1.5018,
+      "step": 4382
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015923683384051632,
+      "loss": 1.6126,
+      "step": 4383
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015921452149134373,
+      "loss": 1.6973,
+      "step": 4384
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015919220460151133,
+      "loss": 1.2346,
+      "step": 4385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015916988317273042,
+      "loss": 1.7421,
+      "step": 4386
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015914755720671263,
+      "loss": 1.2799,
+      "step": 4387
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001591252267051699,
+      "loss": 1.5363,
+      "step": 4388
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001591028916698146,
+      "loss": 1.8248,
+      "step": 4389
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001590805521023595,
+      "loss": 1.4261,
+      "step": 4390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001590582080045175,
+      "loss": 1.8987,
+      "step": 4391
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015903585937800205,
+      "loss": 1.6323,
+      "step": 4392
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015901350622452687,
+      "loss": 1.5063,
+      "step": 4393
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015899114854580604,
+      "loss": 1.477,
+      "step": 4394
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015896878634355396,
+      "loss": 1.0431,
+      "step": 4395
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015894641961948538,
+      "loss": 1.3185,
+      "step": 4396
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015892404837531548,
+      "loss": 1.4588,
+      "step": 4397
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015890167261275968,
+      "loss": 1.5668,
+      "step": 4398
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015887929233353374,
+      "loss": 1.8628,
+      "step": 4399
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015885690753935391,
+      "loss": 1.4476,
+      "step": 4400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015883451823193664,
+      "loss": 1.1708,
+      "step": 4401
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015881212441299877,
+      "loss": 1.6507,
+      "step": 4402
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001587897260842575,
+      "loss": 1.6504,
+      "step": 4403
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001587673232474304,
+      "loss": 1.4403,
+      "step": 4404
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001587449159042353,
+      "loss": 1.069,
+      "step": 4405
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015872250405639043,
+      "loss": 1.5751,
+      "step": 4406
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015870008770561443,
+      "loss": 1.6222,
+      "step": 4407
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015867766685362616,
+      "loss": 1.4776,
+      "step": 4408
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001586552415021449,
+      "loss": 1.3556,
+      "step": 4409
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015863281165289027,
+      "loss": 1.5926,
+      "step": 4410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001586103773075822,
+      "loss": 1.3817,
+      "step": 4411
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015858793846794104,
+      "loss": 0.5373,
+      "step": 4412
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015856549513568738,
+      "loss": 1.2084,
+      "step": 4413
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015854304731254226,
+      "loss": 0.9764,
+      "step": 4414
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015852059500022695,
+      "loss": 1.3996,
+      "step": 4415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001584981382004632,
+      "loss": 1.2423,
+      "step": 4416
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015847567691497293,
+      "loss": 1.5685,
+      "step": 4417
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015845321114547862,
+      "loss": 2.4239,
+      "step": 4418
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015843074089370293,
+      "loss": 1.5211,
+      "step": 4419
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001584082661613689,
+      "loss": 1.8346,
+      "step": 4420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015838578695019995,
+      "loss": 2.2869,
+      "step": 4421
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001583633032619198,
+      "loss": 1.7803,
+      "step": 4422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015834081509825254,
+      "loss": 0.989,
+      "step": 4423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015831832246092259,
+      "loss": 1.3411,
+      "step": 4424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015829582535165475,
+      "loss": 1.5959,
+      "step": 4425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001582733237721741,
+      "loss": 1.7267,
+      "step": 4426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015825081772420614,
+      "loss": 1.2891,
+      "step": 4427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015822830720947658,
+      "loss": 1.4523,
+      "step": 4428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015820579222971162,
+      "loss": 1.5557,
+      "step": 4429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001581832727866378,
+      "loss": 1.178,
+      "step": 4430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015816074888198183,
+      "loss": 1.942,
+      "step": 4431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015813822051747095,
+      "loss": 1.4263,
+      "step": 4432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015811568769483265,
+      "loss": 0.9643,
+      "step": 4433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015809315041579477,
+      "loss": 1.5576,
+      "step": 4434
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015807060868208554,
+      "loss": 1.3554,
+      "step": 4435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015804806249543346,
+      "loss": 1.9649,
+      "step": 4436
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001580255118575674,
+      "loss": 1.8248,
+      "step": 4437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015800295677021662,
+      "loss": 1.4934,
+      "step": 4438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001579803972351106,
+      "loss": 1.5304,
+      "step": 4439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015795783325397937,
+      "loss": 1.3991,
+      "step": 4440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015793526482855302,
+      "loss": 1.5563,
+      "step": 4441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015791269196056224,
+      "loss": 1.6234,
+      "step": 4442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015789011465173794,
+      "loss": 0.8958,
+      "step": 4443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001578675329038113,
+      "loss": 1.5168,
+      "step": 4444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000157844946718514,
+      "loss": 1.2586,
+      "step": 4445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015782235609757797,
+      "loss": 1.3792,
+      "step": 4446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015779976104273548,
+      "loss": 1.5484,
+      "step": 4447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015777716155571916,
+      "loss": 1.6614,
+      "step": 4448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015775455763826198,
+      "loss": 1.7849,
+      "step": 4449
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015773194929209723,
+      "loss": 1.2895,
+      "step": 4450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015770933651895855,
+      "loss": 1.4944,
+      "step": 4451
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015768671932057994,
+      "loss": 1.5027,
+      "step": 4452
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015766409769869572,
+      "loss": 1.5789,
+      "step": 4453
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015764147165504052,
+      "loss": 1.7166,
+      "step": 4454
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015761884119134937,
+      "loss": 1.207,
+      "step": 4455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015759620630935762,
+      "loss": 1.5493,
+      "step": 4456
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001575735670108009,
+      "loss": 1.1795,
+      "step": 4457
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015755092329741528,
+      "loss": 1.4734,
+      "step": 4458
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015752827517093708,
+      "loss": 1.6432,
+      "step": 4459
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015750562263310298,
+      "loss": 1.7199,
+      "step": 4460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015748296568565006,
+      "loss": 2.0726,
+      "step": 4461
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015746030433031564,
+      "loss": 1.2484,
+      "step": 4462
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015743763856883745,
+      "loss": 1.4268,
+      "step": 4463
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015741496840295354,
+      "loss": 1.7093,
+      "step": 4464
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015739229383440227,
+      "loss": 1.7807,
+      "step": 4465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015736961486492238,
+      "loss": 2.0387,
+      "step": 4466
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015734693149625296,
+      "loss": 1.6562,
+      "step": 4467
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015732424373013335,
+      "loss": 1.7002,
+      "step": 4468
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015730155156830328,
+      "loss": 1.497,
+      "step": 4469
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015727885501250286,
+      "loss": 1.4924,
+      "step": 4470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015725615406447248,
+      "loss": 1.3594,
+      "step": 4471
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015723344872595288,
+      "loss": 1.6048,
+      "step": 4472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015721073899868513,
+      "loss": 1.9431,
+      "step": 4473
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015718802488441067,
+      "loss": 1.4548,
+      "step": 4474
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015716530638487127,
+      "loss": 1.6359,
+      "step": 4475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015714258350180892,
+      "loss": 1.4004,
+      "step": 4476
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001571198562369662,
+      "loss": 1.2807,
+      "step": 4477
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015709712459208573,
+      "loss": 1.6183,
+      "step": 4478
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015707438856891066,
+      "loss": 1.4806,
+      "step": 4479
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015705164816918446,
+      "loss": 1.1307,
+      "step": 4480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015702890339465084,
+      "loss": 1.0842,
+      "step": 4481
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015700615424705394,
+      "loss": 1.7382,
+      "step": 4482
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001569834007281382,
+      "loss": 1.5864,
+      "step": 4483
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015696064283964836,
+      "loss": 1.4605,
+      "step": 4484
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001569378805833296,
+      "loss": 1.7883,
+      "step": 4485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015691511396092727,
+      "loss": 1.6772,
+      "step": 4486
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015689234297418722,
+      "loss": 1.4571,
+      "step": 4487
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015686956762485553,
+      "loss": 2.042,
+      "step": 4488
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015684678791467866,
+      "loss": 1.3266,
+      "step": 4489
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001568240038454034,
+      "loss": 1.6707,
+      "step": 4490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015680121541877684,
+      "loss": 2.3687,
+      "step": 4491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015677842263654647,
+      "loss": 1.3428,
+      "step": 4492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001567556255004601,
+      "loss": 1.6025,
+      "step": 4493
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015673282401226573,
+      "loss": 1.3463,
+      "step": 4494
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015671001817371192,
+      "loss": 1.7406,
+      "step": 4495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015668720798654744,
+      "loss": 1.4816,
+      "step": 4496
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015666439345252134,
+      "loss": 1.312,
+      "step": 4497
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015664157457338318,
+      "loss": 1.0285,
+      "step": 4498
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015661875135088268,
+      "loss": 1.5621,
+      "step": 4499
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015659592378676997,
+      "loss": 1.0559,
+      "step": 4500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015657309188279553,
+      "loss": 1.6948,
+      "step": 4501
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015655025564071008,
+      "loss": 1.4782,
+      "step": 4502
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001565274150622648,
+      "loss": 1.0113,
+      "step": 4503
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015650457014921115,
+      "loss": 1.2232,
+      "step": 4504
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015648172090330085,
+      "loss": 1.5552,
+      "step": 4505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015645886732628604,
+      "loss": 1.4872,
+      "step": 4506
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001564360094199192,
+      "loss": 1.7686,
+      "step": 4507
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015641314718595306,
+      "loss": 1.8102,
+      "step": 4508
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015639028062614077,
+      "loss": 1.0911,
+      "step": 4509
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015636740974223575,
+      "loss": 1.7141,
+      "step": 4510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015634453453599177,
+      "loss": 1.2505,
+      "step": 4511
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000156321655009163,
+      "loss": 1.0364,
+      "step": 4512
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015629877116350377,
+      "loss": 1.3238,
+      "step": 4513
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015627588300076893,
+      "loss": 1.9727,
+      "step": 4514
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015625299052271355,
+      "loss": 1.7568,
+      "step": 4515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015623009373109305,
+      "loss": 1.016,
+      "step": 4516
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015620719262766323,
+      "loss": 2.3242,
+      "step": 4517
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001561842872141801,
+      "loss": 2.1082,
+      "step": 4518
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001561613774924002,
+      "loss": 1.1674,
+      "step": 4519
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015613846346408018,
+      "loss": 1.3827,
+      "step": 4520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015611554513097716,
+      "loss": 1.547,
+      "step": 4521
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001560926224948486,
+      "loss": 1.7261,
+      "step": 4522
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015606969555745216,
+      "loss": 1.6221,
+      "step": 4523
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015604676432054592,
+      "loss": 1.0517,
+      "step": 4524
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015602382878588837,
+      "loss": 1.3438,
+      "step": 4525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015600088895523814,
+      "loss": 1.1344,
+      "step": 4526
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015597794483035438,
+      "loss": 1.0178,
+      "step": 4527
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001559549964129964,
+      "loss": 1.4322,
+      "step": 4528
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015593204370492397,
+      "loss": 1.2307,
+      "step": 4529
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015590908670789713,
+      "loss": 1.5541,
+      "step": 4530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001558861254236762,
+      "loss": 1.4731,
+      "step": 4531
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000155863159854022,
+      "loss": 2.1725,
+      "step": 4532
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001558401900006955,
+      "loss": 1.3758,
+      "step": 4533
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015581721586545804,
+      "loss": 1.351,
+      "step": 4534
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015579423745007133,
+      "loss": 1.0138,
+      "step": 4535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001557712547562974,
+      "loss": 1.7949,
+      "step": 4536
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001557482677858986,
+      "loss": 1.7099,
+      "step": 4537
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015572527654063764,
+      "loss": 1.4607,
+      "step": 4538
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015570228102227746,
+      "loss": 0.5413,
+      "step": 4539
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001556792812325814,
+      "loss": 1.3106,
+      "step": 4540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015565627717331314,
+      "loss": 1.8232,
+      "step": 4541
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001556332688462367,
+      "loss": 1.5135,
+      "step": 4542
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001556102562531163,
+      "loss": 1.7169,
+      "step": 4543
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015558723939571666,
+      "loss": 1.5397,
+      "step": 4544
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015556421827580272,
+      "loss": 1.6966,
+      "step": 4545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001555411928951398,
+      "loss": 1.2565,
+      "step": 4546
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015551816325549346,
+      "loss": 1.7465,
+      "step": 4547
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001554951293586297,
+      "loss": 1.6136,
+      "step": 4548
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015547209120631483,
+      "loss": 1.8683,
+      "step": 4549
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015544904880031535,
+      "loss": 1.5768,
+      "step": 4550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015542600214239825,
+      "loss": 1.4933,
+      "step": 4551
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001554029512343308,
+      "loss": 1.4635,
+      "step": 4552
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001553798960778805,
+      "loss": 0.5798,
+      "step": 4553
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015535683667481536,
+      "loss": 2.1196,
+      "step": 4554
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015533377302690354,
+      "loss": 1.1486,
+      "step": 4555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015531070513591358,
+      "loss": 1.4364,
+      "step": 4556
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015528763300361442,
+      "loss": 1.5108,
+      "step": 4557
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015526455663177524,
+      "loss": 1.6714,
+      "step": 4558
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015524147602216556,
+      "loss": 1.5754,
+      "step": 4559
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015521839117655525,
+      "loss": 1.3431,
+      "step": 4560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015519530209671449,
+      "loss": 1.4931,
+      "step": 4561
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001551722087844138,
+      "loss": 0.9734,
+      "step": 4562
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015514911124142396,
+      "loss": 1.3714,
+      "step": 4563
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001551260094695162,
+      "loss": 1.5897,
+      "step": 4564
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015510290347046192,
+      "loss": 1.5711,
+      "step": 4565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015507979324603296,
+      "loss": 1.1098,
+      "step": 4566
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001550566787980015,
+      "loss": 1.0881,
+      "step": 4567
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001550335601281399,
+      "loss": 1.3215,
+      "step": 4568
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015501043723822097,
+      "loss": 1.8069,
+      "step": 4569
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015498731013001784,
+      "loss": 1.5104,
+      "step": 4570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015496417880530387,
+      "loss": 1.2,
+      "step": 4571
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015494104326585286,
+      "loss": 1.3374,
+      "step": 4572
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001549179035134389,
+      "loss": 1.7464,
+      "step": 4573
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001548947595498363,
+      "loss": 1.5667,
+      "step": 4574
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015487161137681978,
+      "loss": 1.4489,
+      "step": 4575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015484845899616447,
+      "loss": 1.9969,
+      "step": 4576
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015482530240964565,
+      "loss": 1.1675,
+      "step": 4577
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000154802141619039,
+      "loss": 1.2437,
+      "step": 4578
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015477897662612058,
+      "loss": 1.1291,
+      "step": 4579
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015475580743266667,
+      "loss": 1.4656,
+      "step": 4580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015473263404045394,
+      "loss": 1.2941,
+      "step": 4581
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015470945645125936,
+      "loss": 1.3969,
+      "step": 4582
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001546862746668602,
+      "loss": 1.6042,
+      "step": 4583
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015466308868903412,
+      "loss": 1.4941,
+      "step": 4584
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015463989851955903,
+      "loss": 1.285,
+      "step": 4585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015461670416021318,
+      "loss": 1.5359,
+      "step": 4586
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015459350561277513,
+      "loss": 1.9752,
+      "step": 4587
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015457030287902387,
+      "loss": 1.8164,
+      "step": 4588
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001545470959607385,
+      "loss": 1.4696,
+      "step": 4589
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015452388485969864,
+      "loss": 1.9864,
+      "step": 4590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015450066957768414,
+      "loss": 0.821,
+      "step": 4591
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015447745011647513,
+      "loss": 1.6753,
+      "step": 4592
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015445422647785223,
+      "loss": 1.7837,
+      "step": 4593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015443099866359612,
+      "loss": 1.4858,
+      "step": 4594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015440776667548806,
+      "loss": 1.2194,
+      "step": 4595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015438453051530945,
+      "loss": 1.866,
+      "step": 4596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015436129018484213,
+      "loss": 1.6654,
+      "step": 4597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015433804568586813,
+      "loss": 1.8162,
+      "step": 4598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015431479702016993,
+      "loss": 1.7901,
+      "step": 4599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015429154418953025,
+      "loss": 1.893,
+      "step": 4600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015426828719573217,
+      "loss": 2.1826,
+      "step": 4601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015424502604055907,
+      "loss": 1.3261,
+      "step": 4602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015422176072579463,
+      "loss": 1.6485,
+      "step": 4603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015419849125322286,
+      "loss": 1.2572,
+      "step": 4604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015417521762462814,
+      "loss": 1.4141,
+      "step": 4605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001541519398417951,
+      "loss": 1.3885,
+      "step": 4606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015412865790650874,
+      "loss": 0.387,
+      "step": 4607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015410537182055433,
+      "loss": 1.3311,
+      "step": 4608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015408208158571748,
+      "loss": 2.2831,
+      "step": 4609
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015405878720378414,
+      "loss": 1.4865,
+      "step": 4610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015403548867654054,
+      "loss": 1.9198,
+      "step": 4611
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015401218600577329,
+      "loss": 1.3809,
+      "step": 4612
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015398887919326914,
+      "loss": 1.3483,
+      "step": 4613
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015396556824081548,
+      "loss": 1.2217,
+      "step": 4614
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015394225315019973,
+      "loss": 0.9122,
+      "step": 4615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015391893392320972,
+      "loss": 1.0221,
+      "step": 4616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015389561056163364,
+      "loss": 1.4545,
+      "step": 4617
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015387228306725995,
+      "loss": 1.387,
+      "step": 4618
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001538489514418774,
+      "loss": 1.5346,
+      "step": 4619
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015382561568727513,
+      "loss": 1.4877,
+      "step": 4620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001538022758052426,
+      "loss": 1.2421,
+      "step": 4621
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015377893179756945,
+      "loss": 1.2623,
+      "step": 4622
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015375558366604583,
+      "loss": 1.5067,
+      "step": 4623
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015373223141246207,
+      "loss": 1.7379,
+      "step": 4624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015370887503860883,
+      "loss": 1.917,
+      "step": 4625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015368551454627714,
+      "loss": 1.3158,
+      "step": 4626
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015366214993725832,
+      "loss": 1.6106,
+      "step": 4627
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000153638781213344,
+      "loss": 1.8755,
+      "step": 4628
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015361540837632612,
+      "loss": 1.9538,
+      "step": 4629
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015359203142799695,
+      "loss": 1.0322,
+      "step": 4630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015356865037014912,
+      "loss": 1.261,
+      "step": 4631
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015354526520457542,
+      "loss": 1.5759,
+      "step": 4632
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015352187593306918,
+      "loss": 2.2385,
+      "step": 4633
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001534984825574238,
+      "loss": 1.2923,
+      "step": 4634
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015347508507943322,
+      "loss": 1.4868,
+      "step": 4635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015345168350089158,
+      "loss": 1.763,
+      "step": 4636
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015342827782359328,
+      "loss": 1.4421,
+      "step": 4637
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015340486804933318,
+      "loss": 1.5981,
+      "step": 4638
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015338145417990632,
+      "loss": 1.4597,
+      "step": 4639
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015335803621710815,
+      "loss": 1.2751,
+      "step": 4640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001533346141627344,
+      "loss": 1.4737,
+      "step": 4641
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001533111880185811,
+      "loss": 1.4064,
+      "step": 4642
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015328775778644457,
+      "loss": 1.2845,
+      "step": 4643
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001532643234681215,
+      "loss": 1.3532,
+      "step": 4644
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001532408850654089,
+      "loss": 1.55,
+      "step": 4645
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015321744258010397,
+      "loss": 2.2393,
+      "step": 4646
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015319399601400444,
+      "loss": 1.8028,
+      "step": 4647
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015317054536890815,
+      "loss": 1.476,
+      "step": 4648
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015314709064661333,
+      "loss": 1.6213,
+      "step": 4649
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015312363184891858,
+      "loss": 1.6969,
+      "step": 4650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015310016897762267,
+      "loss": 1.7875,
+      "step": 4651
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015307670203452487,
+      "loss": 1.6771,
+      "step": 4652
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015305323102142454,
+      "loss": 1.5113,
+      "step": 4653
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015302975594012162,
+      "loss": 1.4034,
+      "step": 4654
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015300627679241607,
+      "loss": 1.1959,
+      "step": 4655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001529827935801084,
+      "loss": 1.3611,
+      "step": 4656
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001529593063049993,
+      "loss": 1.6536,
+      "step": 4657
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015293581496888986,
+      "loss": 1.8904,
+      "step": 4658
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015291231957358133,
+      "loss": 1.8314,
+      "step": 4659
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015288882012087548,
+      "loss": 1.4708,
+      "step": 4660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015286531661257423,
+      "loss": 1.5505,
+      "step": 4661
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001528418090504799,
+      "loss": 1.1394,
+      "step": 4662
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015281829743639503,
+      "loss": 1.4323,
+      "step": 4663
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015279478177212253,
+      "loss": 1.7149,
+      "step": 4664
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001527712620594657,
+      "loss": 1.381,
+      "step": 4665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015274773830022798,
+      "loss": 1.4259,
+      "step": 4666
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015272421049621326,
+      "loss": 1.433,
+      "step": 4667
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015270067864922563,
+      "loss": 1.5257,
+      "step": 4668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015267714276106958,
+      "loss": 1.4371,
+      "step": 4669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015265360283354991,
+      "loss": 1.5047,
+      "step": 4670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001526300588684717,
+      "loss": 1.1734,
+      "step": 4671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015260651086764027,
+      "loss": 1.4097,
+      "step": 4672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015258295883286137,
+      "loss": 1.2233,
+      "step": 4673
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000152559402765941,
+      "loss": 1.257,
+      "step": 4674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001525358426686854,
+      "loss": 2.2574,
+      "step": 4675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001525122785429013,
+      "loss": 2.1811,
+      "step": 4676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015248871039039562,
+      "loss": 1.6172,
+      "step": 4677
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015246513821297555,
+      "loss": 1.4424,
+      "step": 4678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015244156201244864,
+      "loss": 1.2281,
+      "step": 4679
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001524179817906228,
+      "loss": 1.6958,
+      "step": 4680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015239439754930614,
+      "loss": 1.6645,
+      "step": 4681
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001523708092903072,
+      "loss": 1.6325,
+      "step": 4682
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015234721701543473,
+      "loss": 1.3234,
+      "step": 4683
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015232362072649784,
+      "loss": 1.0664,
+      "step": 4684
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015230002042530586,
+      "loss": 1.4849,
+      "step": 4685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015227641611366858,
+      "loss": 1.8037,
+      "step": 4686
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000152252807793396,
+      "loss": 1.7854,
+      "step": 4687
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015222919546629838,
+      "loss": 1.4171,
+      "step": 4688
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015220557913418642,
+      "loss": 1.6084,
+      "step": 4689
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015218195879887104,
+      "loss": 1.3536,
+      "step": 4690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015215833446216345,
+      "loss": 1.261,
+      "step": 4691
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015213470612587522,
+      "loss": 2.276,
+      "step": 4692
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015211107379181824,
+      "loss": 1.6647,
+      "step": 4693
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015208743746180462,
+      "loss": 0.8283,
+      "step": 4694
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015206379713764688,
+      "loss": 1.6609,
+      "step": 4695
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015204015282115773,
+      "loss": 1.9598,
+      "step": 4696
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015201650451415034,
+      "loss": 1.1801,
+      "step": 4697
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000151992852218438,
+      "loss": 1.3456,
+      "step": 4698
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001519691959358345,
+      "loss": 1.839,
+      "step": 4699
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015194553566815375,
+      "loss": 1.7065,
+      "step": 4700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015192187141721016,
+      "loss": 1.3495,
+      "step": 4701
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015189820318481826,
+      "loss": 1.1218,
+      "step": 4702
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015187453097279297,
+      "loss": 1.7372,
+      "step": 4703
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001518508547829495,
+      "loss": 1.7345,
+      "step": 4704
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015182717461710349,
+      "loss": 1.7872,
+      "step": 4705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001518034904770706,
+      "loss": 1.0762,
+      "step": 4706
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015177980236466706,
+      "loss": 1.1695,
+      "step": 4707
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015175611028170932,
+      "loss": 1.6127,
+      "step": 4708
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015173241423001413,
+      "loss": 1.4233,
+      "step": 4709
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001517087142113985,
+      "loss": 1.8077,
+      "step": 4710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001516850102276798,
+      "loss": 0.857,
+      "step": 4711
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001516613022806757,
+      "loss": 1.1947,
+      "step": 4712
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015163759037220412,
+      "loss": 0.9244,
+      "step": 4713
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001516138745040834,
+      "loss": 1.3658,
+      "step": 4714
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015159015467813201,
+      "loss": 1.6828,
+      "step": 4715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015156643089616893,
+      "loss": 1.4436,
+      "step": 4716
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015154270316001327,
+      "loss": 0.8737,
+      "step": 4717
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001515189714714845,
+      "loss": 1.4781,
+      "step": 4718
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015149523583240243,
+      "loss": 1.1839,
+      "step": 4719
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015147149624458712,
+      "loss": 2.4023,
+      "step": 4720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015144775270985902,
+      "loss": 0.6774,
+      "step": 4721
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015142400523003878,
+      "loss": 1.5542,
+      "step": 4722
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015140025380694735,
+      "loss": 2.0331,
+      "step": 4723
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015137649844240608,
+      "loss": 1.4657,
+      "step": 4724
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015135273913823653,
+      "loss": 1.8028,
+      "step": 4725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015132897589626064,
+      "loss": 1.3433,
+      "step": 4726
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015130520871830058,
+      "loss": 1.7156,
+      "step": 4727
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001512814376061789,
+      "loss": 1.264,
+      "step": 4728
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015125766256171837,
+      "loss": 2.1039,
+      "step": 4729
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001512338835867421,
+      "loss": 1.4774,
+      "step": 4730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015121010068307346,
+      "loss": 1.1577,
+      "step": 4731
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001511863138525362,
+      "loss": 1.2206,
+      "step": 4732
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015116252309695436,
+      "loss": 1.1856,
+      "step": 4733
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001511387284181522,
+      "loss": 1.4894,
+      "step": 4734
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015111492981795434,
+      "loss": 2.1731,
+      "step": 4735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015109112729818573,
+      "loss": 1.6714,
+      "step": 4736
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015106732086067154,
+      "loss": 1.1035,
+      "step": 4737
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001510435105072373,
+      "loss": 1.394,
+      "step": 4738
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015101969623970883,
+      "loss": 1.3623,
+      "step": 4739
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015099587805991226,
+      "loss": 1.6338,
+      "step": 4740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015097205596967394,
+      "loss": 1.196,
+      "step": 4741
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015094822997082066,
+      "loss": 1.5764,
+      "step": 4742
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001509244000651794,
+      "loss": 1.3194,
+      "step": 4743
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015090056625457744,
+      "loss": 1.8308,
+      "step": 4744
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015087672854084246,
+      "loss": 1.5751,
+      "step": 4745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015085288692580234,
+      "loss": 1.8294,
+      "step": 4746
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015082904141128532,
+      "loss": 1.3183,
+      "step": 4747
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015080519199911984,
+      "loss": 1.7757,
+      "step": 4748
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015078133869113477,
+      "loss": 1.4279,
+      "step": 4749
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015075748148915915,
+      "loss": 1.5988,
+      "step": 4750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001507336203950225,
+      "loss": 0.8395,
+      "step": 4751
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015070975541055443,
+      "loss": 1.8717,
+      "step": 4752
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000150685886537585,
+      "loss": 1.914,
+      "step": 4753
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015066201377794447,
+      "loss": 1.7806,
+      "step": 4754
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001506381371334635,
+      "loss": 1.3429,
+      "step": 4755
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015061425660597286,
+      "loss": 0.9466,
+      "step": 4756
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015059037219730388,
+      "loss": 1.7809,
+      "step": 4757
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000150566483909288,
+      "loss": 1.6091,
+      "step": 4758
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015054259174375698,
+      "loss": 1.2871,
+      "step": 4759
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015051869570254298,
+      "loss": 1.4883,
+      "step": 4760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015049479578747833,
+      "loss": 1.2683,
+      "step": 4761
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015047089200039573,
+      "loss": 1.4562,
+      "step": 4762
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015044698434312816,
+      "loss": 1.3712,
+      "step": 4763
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015042307281750886,
+      "loss": 1.0298,
+      "step": 4764
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015039915742537144,
+      "loss": 1.2657,
+      "step": 4765
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001503752381685498,
+      "loss": 1.7694,
+      "step": 4766
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015035131504887806,
+      "loss": 1.0889,
+      "step": 4767
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015032738806819066,
+      "loss": 1.8006,
+      "step": 4768
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001503034572283224,
+      "loss": 1.6766,
+      "step": 4769
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015027952253110832,
+      "loss": 1.0341,
+      "step": 4770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015025558397838376,
+      "loss": 1.2193,
+      "step": 4771
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001502316415719844,
+      "loss": 1.5848,
+      "step": 4772
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015020769531374613,
+      "loss": 1.605,
+      "step": 4773
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001501837452055052,
+      "loss": 1.2191,
+      "step": 4774
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001501597912490982,
+      "loss": 1.7056,
+      "step": 4775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015013583344636186,
+      "loss": 1.5143,
+      "step": 4776
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015011187179913335,
+      "loss": 1.2686,
+      "step": 4777
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001500879063092501,
+      "loss": 1.6351,
+      "step": 4778
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001500639369785498,
+      "loss": 1.4716,
+      "step": 4779
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015003996380887045,
+      "loss": 1.4138,
+      "step": 4780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015001598680205034,
+      "loss": 1.5902,
+      "step": 4781
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014999200595992808,
+      "loss": 1.5539,
+      "step": 4782
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014996802128434259,
+      "loss": 1.0335,
+      "step": 4783
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000149944032777133,
+      "loss": 1.2645,
+      "step": 4784
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014992004044013881,
+      "loss": 1.6362,
+      "step": 4785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014989604427519977,
+      "loss": 1.7673,
+      "step": 4786
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014987204428415596,
+      "loss": 1.1059,
+      "step": 4787
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014984804046884773,
+      "loss": 1.7674,
+      "step": 4788
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014982403283111574,
+      "loss": 1.6047,
+      "step": 4789
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014980002137280092,
+      "loss": 1.3845,
+      "step": 4790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001497760060957445,
+      "loss": 1.8646,
+      "step": 4791
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014975198700178805,
+      "loss": 1.1455,
+      "step": 4792
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001497279640927733,
+      "loss": 1.606,
+      "step": 4793
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014970393737054243,
+      "loss": 1.4567,
+      "step": 4794
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001496799068369379,
+      "loss": 1.8518,
+      "step": 4795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014965587249380228,
+      "loss": 1.2769,
+      "step": 4796
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014963183434297866,
+      "loss": 1.1211,
+      "step": 4797
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014960779238631027,
+      "loss": 1.5189,
+      "step": 4798
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014958374662564072,
+      "loss": 0.8292,
+      "step": 4799
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014955969706281384,
+      "loss": 1.5958,
+      "step": 4800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001495356436996738,
+      "loss": 1.7521,
+      "step": 4801
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014951158653806506,
+      "loss": 1.6814,
+      "step": 4802
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014948752557983236,
+      "loss": 1.036,
+      "step": 4803
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001494634608268207,
+      "loss": 1.3153,
+      "step": 4804
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014943939228087542,
+      "loss": 1.7139,
+      "step": 4805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014941531994384214,
+      "loss": 1.2255,
+      "step": 4806
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014939124381756677,
+      "loss": 1.4609,
+      "step": 4807
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014936716390389552,
+      "loss": 1.8981,
+      "step": 4808
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014934308020467482,
+      "loss": 1.5184,
+      "step": 4809
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014931899272175146,
+      "loss": 1.1514,
+      "step": 4810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014929490145697258,
+      "loss": 1.0146,
+      "step": 4811
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001492708064121854,
+      "loss": 1.4942,
+      "step": 4812
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014924670758923768,
+      "loss": 1.9413,
+      "step": 4813
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014922260498997732,
+      "loss": 1.4136,
+      "step": 4814
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001491984986162525,
+      "loss": 1.4965,
+      "step": 4815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001491743884699118,
+      "loss": 1.7398,
+      "step": 4816
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000149150274552804,
+      "loss": 2.0423,
+      "step": 4817
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014912615686677816,
+      "loss": 1.1979,
+      "step": 4818
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001491020354136837,
+      "loss": 1.302,
+      "step": 4819
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014907791019537028,
+      "loss": 1.8263,
+      "step": 4820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014905378121368785,
+      "loss": 1.2965,
+      "step": 4821
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014902964847048666,
+      "loss": 1.7041,
+      "step": 4822
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014900551196761724,
+      "loss": 1.5517,
+      "step": 4823
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014898137170693042,
+      "loss": 2.2021,
+      "step": 4824
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014895722769027732,
+      "loss": 1.6395,
+      "step": 4825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014893307991950934,
+      "loss": 1.6969,
+      "step": 4826
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001489089283964782,
+      "loss": 1.6423,
+      "step": 4827
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014888477312303576,
+      "loss": 1.1682,
+      "step": 4828
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014886061410103444,
+      "loss": 1.4403,
+      "step": 4829
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014883645133232664,
+      "loss": 1.4123,
+      "step": 4830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001488122848187653,
+      "loss": 1.8824,
+      "step": 4831
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014878811456220349,
+      "loss": 1.7653,
+      "step": 4832
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014876394056449467,
+      "loss": 0.9432,
+      "step": 4833
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014873976282749253,
+      "loss": 1.7917,
+      "step": 4834
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014871558135305101,
+      "loss": 1.7893,
+      "step": 4835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014869139614302442,
+      "loss": 1.4487,
+      "step": 4836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014866720719926732,
+      "loss": 1.8274,
+      "step": 4837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014864301452363453,
+      "loss": 1.8328,
+      "step": 4838
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014861881811798123,
+      "loss": 1.3832,
+      "step": 4839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014859461798416278,
+      "loss": 1.5343,
+      "step": 4840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014857041412403492,
+      "loss": 1.5413,
+      "step": 4841
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014854620653945361,
+      "loss": 1.2857,
+      "step": 4842
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014852199523227516,
+      "loss": 1.6217,
+      "step": 4843
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001484977802043561,
+      "loss": 1.5868,
+      "step": 4844
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014847356145755331,
+      "loss": 1.3305,
+      "step": 4845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001484493389937239,
+      "loss": 1.481,
+      "step": 4846
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014842511281472524,
+      "loss": 1.2377,
+      "step": 4847
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001484008829224151,
+      "loss": 1.3622,
+      "step": 4848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014837664931865145,
+      "loss": 1.4768,
+      "step": 4849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014835241200529252,
+      "loss": 1.52,
+      "step": 4850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014832817098419691,
+      "loss": 1.4834,
+      "step": 4851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014830392625722345,
+      "loss": 1.8603,
+      "step": 4852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014827967782623125,
+      "loss": 1.216,
+      "step": 4853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014825542569307971,
+      "loss": 1.0648,
+      "step": 4854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014823116985962856,
+      "loss": 1.1882,
+      "step": 4855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001482069103277377,
+      "loss": 1.4401,
+      "step": 4856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014818264709926747,
+      "loss": 1.5405,
+      "step": 4857
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014815838017607835,
+      "loss": 1.5045,
+      "step": 4858
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001481341095600312,
+      "loss": 1.0268,
+      "step": 4859
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014810983525298708,
+      "loss": 1.6531,
+      "step": 4860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001480855572568075,
+      "loss": 1.3916,
+      "step": 4861
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014806127557335396,
+      "loss": 1.0411,
+      "step": 4862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014803699020448854,
+      "loss": 1.5613,
+      "step": 4863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001480127011520735,
+      "loss": 1.5272,
+      "step": 4864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014798840841797126,
+      "loss": 1.9167,
+      "step": 4865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001479641120040447,
+      "loss": 2.1002,
+      "step": 4866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014793981191215684,
+      "loss": 1.1565,
+      "step": 4867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001479155081441711,
+      "loss": 1.18,
+      "step": 4868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014789120070195114,
+      "loss": 1.3074,
+      "step": 4869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014786688958736086,
+      "loss": 1.4698,
+      "step": 4870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014784257480226443,
+      "loss": 1.3806,
+      "step": 4871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014781825634852647,
+      "loss": 1.5763,
+      "step": 4872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014779393422801167,
+      "loss": 1.5369,
+      "step": 4873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014776960844258507,
+      "loss": 1.5391,
+      "step": 4874
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014774527899411206,
+      "loss": 1.8871,
+      "step": 4875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014772094588445822,
+      "loss": 1.4147,
+      "step": 4876
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001476966091154895,
+      "loss": 1.6539,
+      "step": 4877
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014767226868907203,
+      "loss": 1.4235,
+      "step": 4878
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001476479246070723,
+      "loss": 1.8561,
+      "step": 4879
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000147623576871357,
+      "loss": 1.3284,
+      "step": 4880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014759922548379323,
+      "loss": 1.7057,
+      "step": 4881
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014757487044624824,
+      "loss": 1.5545,
+      "step": 4882
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014755051176058962,
+      "loss": 0.3218,
+      "step": 4883
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014752614942868526,
+      "loss": 1.5169,
+      "step": 4884
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001475017834524033,
+      "loss": 1.4338,
+      "step": 4885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014747741383361207,
+      "loss": 1.0606,
+      "step": 4886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014745304057418035,
+      "loss": 1.924,
+      "step": 4887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014742866367597714,
+      "loss": 1.5724,
+      "step": 4888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014740428314087166,
+      "loss": 1.3516,
+      "step": 4889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014737989897073344,
+      "loss": 1.4251,
+      "step": 4890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001473555111674323,
+      "loss": 1.4591,
+      "step": 4891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014733111973283834,
+      "loss": 1.2506,
+      "step": 4892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014730672466882193,
+      "loss": 1.1186,
+      "step": 4893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001472823259772537,
+      "loss": 1.8051,
+      "step": 4894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001472579236600046,
+      "loss": 1.4105,
+      "step": 4895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001472335177189459,
+      "loss": 1.1641,
+      "step": 4896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014720910815594897,
+      "loss": 1.7131,
+      "step": 4897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014718469497288568,
+      "loss": 1.5576,
+      "step": 4898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014716027817162795,
+      "loss": 1.5085,
+      "step": 4899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014713585775404816,
+      "loss": 1.0409,
+      "step": 4900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014711143372201897,
+      "loss": 1.3012,
+      "step": 4901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014708700607741316,
+      "loss": 1.0726,
+      "step": 4902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014706257482210392,
+      "loss": 1.8149,
+      "step": 4903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014703813995796465,
+      "loss": 1.3435,
+      "step": 4904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014701370148686906,
+      "loss": 1.4975,
+      "step": 4905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014698925941069115,
+      "loss": 1.6219,
+      "step": 4906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014696481373130515,
+      "loss": 1.4397,
+      "step": 4907
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014694036445058562,
+      "loss": 1.0633,
+      "step": 4908
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014691591157040736,
+      "loss": 1.2614,
+      "step": 4909
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014689145509264544,
+      "loss": 1.6759,
+      "step": 4910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014686699501917523,
+      "loss": 1.4141,
+      "step": 4911
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014684253135187233,
+      "loss": 1.6371,
+      "step": 4912
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014681806409261273,
+      "loss": 1.4293,
+      "step": 4913
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014679359324327253,
+      "loss": 1.5071,
+      "step": 4914
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014676911880572827,
+      "loss": 1.5951,
+      "step": 4915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014674464078185662,
+      "loss": 1.7818,
+      "step": 4916
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014672015917353463,
+      "loss": 1.4207,
+      "step": 4917
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014669567398263957,
+      "loss": 1.4748,
+      "step": 4918
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014667118521104904,
+      "loss": 1.2545,
+      "step": 4919
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014664669286064083,
+      "loss": 1.7408,
+      "step": 4920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014662219693329306,
+      "loss": 1.1756,
+      "step": 4921
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014659769743088414,
+      "loss": 1.3101,
+      "step": 4922
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001465731943552927,
+      "loss": 1.1729,
+      "step": 4923
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001465486877083977,
+      "loss": 1.3518,
+      "step": 4924
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001465241774920783,
+      "loss": 1.5628,
+      "step": 4925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014649966370821403,
+      "loss": 1.7157,
+      "step": 4926
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014647514635868464,
+      "loss": 1.6025,
+      "step": 4927
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014645062544537017,
+      "loss": 1.5827,
+      "step": 4928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014642610097015084,
+      "loss": 1.095,
+      "step": 4929
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001464015729349073,
+      "loss": 1.4096,
+      "step": 4930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001463770413415204,
+      "loss": 1.2299,
+      "step": 4931
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001463525061918712,
+      "loss": 1.627,
+      "step": 4932
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014632796748784117,
+      "loss": 1.3661,
+      "step": 4933
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001463034252313119,
+      "loss": 1.4688,
+      "step": 4934
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001462788794241654,
+      "loss": 1.3401,
+      "step": 4935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001462543300682838,
+      "loss": 1.524,
+      "step": 4936
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014622977716554966,
+      "loss": 1.6008,
+      "step": 4937
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001462052207178457,
+      "loss": 1.4948,
+      "step": 4938
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014618066072705495,
+      "loss": 1.5732,
+      "step": 4939
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014615609719506066,
+      "loss": 1.1141,
+      "step": 4940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001461315301237465,
+      "loss": 1.5954,
+      "step": 4941
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001461069595149962,
+      "loss": 1.5218,
+      "step": 4942
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014608238537069395,
+      "loss": 1.425,
+      "step": 4943
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001460578076927241,
+      "loss": 1.3668,
+      "step": 4944
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001460332264829713,
+      "loss": 1.6247,
+      "step": 4945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014600864174332053,
+      "loss": 1.6922,
+      "step": 4946
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014598405347565692,
+      "loss": 0.9842,
+      "step": 4947
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014595946168186594,
+      "loss": 1.4039,
+      "step": 4948
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014593486636383334,
+      "loss": 1.6815,
+      "step": 4949
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014591026752344512,
+      "loss": 0.5886,
+      "step": 4950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001458856651625876,
+      "loss": 1.3348,
+      "step": 4951
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014586105928314727,
+      "loss": 1.8229,
+      "step": 4952
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000145836449887011,
+      "loss": 1.6723,
+      "step": 4953
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014581183697606585,
+      "loss": 1.4054,
+      "step": 4954
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014578722055219912,
+      "loss": 1.4858,
+      "step": 4955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014576260061729848,
+      "loss": 1.9556,
+      "step": 4956
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014573797717325188,
+      "loss": 1.5478,
+      "step": 4957
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001457133502219474,
+      "loss": 1.4895,
+      "step": 4958
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014568871976527355,
+      "loss": 1.9144,
+      "step": 4959
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014566408580511894,
+      "loss": 1.5653,
+      "step": 4960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001456394483433726,
+      "loss": 1.4727,
+      "step": 4961
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014561480738192373,
+      "loss": 1.1782,
+      "step": 4962
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014559016292266187,
+      "loss": 1.6307,
+      "step": 4963
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014556551496747684,
+      "loss": 1.7408,
+      "step": 4964
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014554086351825857,
+      "loss": 1.5306,
+      "step": 4965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014551620857689742,
+      "loss": 1.084,
+      "step": 4966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014549155014528402,
+      "loss": 1.8087,
+      "step": 4967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014546688822530913,
+      "loss": 1.3647,
+      "step": 4968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014544222281886394,
+      "loss": 1.9474,
+      "step": 4969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014541755392783983,
+      "loss": 0.5861,
+      "step": 4970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014539288155412835,
+      "loss": 1.7533,
+      "step": 4971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014536820569962153,
+      "loss": 1.4751,
+      "step": 4972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014534352636621148,
+      "loss": 1.2644,
+      "step": 4973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014531884355579064,
+      "loss": 1.6251,
+      "step": 4974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001452941572702518,
+      "loss": 1.2061,
+      "step": 4975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001452694675114879,
+      "loss": 1.5895,
+      "step": 4976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014524477428139217,
+      "loss": 1.5682,
+      "step": 4977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014522007758185817,
+      "loss": 1.5282,
+      "step": 4978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001451953774147796,
+      "loss": 1.649,
+      "step": 4979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014517067378205063,
+      "loss": 1.3762,
+      "step": 4980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014514596668556545,
+      "loss": 1.3783,
+      "step": 4981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001451212561272187,
+      "loss": 1.0237,
+      "step": 4982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014509654210890523,
+      "loss": 1.3127,
+      "step": 4983
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014507182463252013,
+      "loss": 1.3606,
+      "step": 4984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014504710369995876,
+      "loss": 1.3444,
+      "step": 4985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014502237931311677,
+      "loss": 1.3252,
+      "step": 4986
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001449976514738901,
+      "loss": 1.6018,
+      "step": 4987
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014497292018417486,
+      "loss": 1.4641,
+      "step": 4988
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014494818544586753,
+      "loss": 1.0705,
+      "step": 4989
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014492344726086475,
+      "loss": 1.6672,
+      "step": 4990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014489870563106356,
+      "loss": 2.2187,
+      "step": 4991
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014487396055836113,
+      "loss": 1.2567,
+      "step": 4992
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014484921204465496,
+      "loss": 1.8324,
+      "step": 4993
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014482446009184282,
+      "loss": 1.5365,
+      "step": 4994
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001447997047018227,
+      "loss": 1.5354,
+      "step": 4995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014477494587649293,
+      "loss": 1.3915,
+      "step": 4996
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014475018361775197,
+      "loss": 1.4871,
+      "step": 4997
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001447254179274987,
+      "loss": 1.5966,
+      "step": 4998
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014470064880763218,
+      "loss": 1.4559,
+      "step": 4999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014467587626005172,
+      "loss": 2.1806,
+      "step": 5000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014465110028665697,
+      "loss": 1.0606,
+      "step": 5001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014462632088934772,
+      "loss": 1.8036,
+      "step": 5002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001446015380700241,
+      "loss": 1.1689,
+      "step": 5003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014457675183058655,
+      "loss": 1.2098,
+      "step": 5004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014455196217293565,
+      "loss": 2.2356,
+      "step": 5005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014452716909897238,
+      "loss": 1.3508,
+      "step": 5006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014450237261059786,
+      "loss": 1.1972,
+      "step": 5007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014447757270971354,
+      "loss": 1.2175,
+      "step": 5008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001444527693982211,
+      "loss": 1.7788,
+      "step": 5009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001444279626780225,
+      "loss": 1.7292,
+      "step": 5010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014440315255101995,
+      "loss": 1.3042,
+      "step": 5011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014437833901911597,
+      "loss": 1.1358,
+      "step": 5012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014435352208421327,
+      "loss": 1.7213,
+      "step": 5013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014432870174821484,
+      "loss": 1.3238,
+      "step": 5014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014430387801302397,
+      "loss": 1.5099,
+      "step": 5015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014427905088054414,
+      "loss": 1.9152,
+      "step": 5016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014425422035267916,
+      "loss": 1.3404,
+      "step": 5017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014422938643133306,
+      "loss": 2.1296,
+      "step": 5018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014420454911841018,
+      "loss": 1.8873,
+      "step": 5019
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014417970841581505,
+      "loss": 1.4047,
+      "step": 5020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014415486432545245,
+      "loss": 1.0901,
+      "step": 5021
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014413001684922757,
+      "loss": 0.9388,
+      "step": 5022
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014410516598904567,
+      "loss": 1.197,
+      "step": 5023
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014408031174681235,
+      "loss": 1.4939,
+      "step": 5024
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014405545412443356,
+      "loss": 1.1808,
+      "step": 5025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001440305931238153,
+      "loss": 1.2646,
+      "step": 5026
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014400572874686402,
+      "loss": 1.7964,
+      "step": 5027
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014398086099548635,
+      "loss": 1.705,
+      "step": 5028
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014395598987158918,
+      "loss": 1.2888,
+      "step": 5029
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014393111537707966,
+      "loss": 0.9232,
+      "step": 5030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014390623751386523,
+      "loss": 1.3851,
+      "step": 5031
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014388135628385352,
+      "loss": 1.77,
+      "step": 5032
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001438564716889525,
+      "loss": 1.6159,
+      "step": 5033
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014383158373107034,
+      "loss": 1.5181,
+      "step": 5034
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014380669241211548,
+      "loss": 1.4976,
+      "step": 5035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014378179773399664,
+      "loss": 1.4196,
+      "step": 5036
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014375689969862277,
+      "loss": 1.3688,
+      "step": 5037
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001437319983079031,
+      "loss": 1.3199,
+      "step": 5038
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001437070935637471,
+      "loss": 1.0279,
+      "step": 5039
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014368218546806453,
+      "loss": 1.6957,
+      "step": 5040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014365727402276533,
+      "loss": 1.6415,
+      "step": 5041
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014363235922975977,
+      "loss": 1.2289,
+      "step": 5042
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014360744109095838,
+      "loss": 1.0627,
+      "step": 5043
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001435825196082719,
+      "loss": 1.7515,
+      "step": 5044
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014355759478361133,
+      "loss": 0.7826,
+      "step": 5045
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014353266661888797,
+      "loss": 1.3639,
+      "step": 5046
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014350773511601333,
+      "loss": 1.4214,
+      "step": 5047
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001434828002768992,
+      "loss": 1.8012,
+      "step": 5048
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014345786210345765,
+      "loss": 1.6219,
+      "step": 5049
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014343292059760092,
+      "loss": 1.2353,
+      "step": 5050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001434079757612416,
+      "loss": 1.4737,
+      "step": 5051
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014338302759629254,
+      "loss": 1.7889,
+      "step": 5052
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001433580761046667,
+      "loss": 2.1766,
+      "step": 5053
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014333312128827748,
+      "loss": 1.184,
+      "step": 5054
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014330816314903845,
+      "loss": 1.6881,
+      "step": 5055
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001432832016888634,
+      "loss": 1.5169,
+      "step": 5056
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014325823690966645,
+      "loss": 1.2156,
+      "step": 5057
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014323326881336189,
+      "loss": 1.352,
+      "step": 5058
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014320829740186437,
+      "loss": 1.3435,
+      "step": 5059
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014318332267708868,
+      "loss": 1.7855,
+      "step": 5060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014315834464094995,
+      "loss": 1.6982,
+      "step": 5061
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014313336329536356,
+      "loss": 1.6309,
+      "step": 5062
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014310837864224505,
+      "loss": 1.4924,
+      "step": 5063
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014308339068351032,
+      "loss": 1.2865,
+      "step": 5064
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001430583994210755,
+      "loss": 1.6682,
+      "step": 5065
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014303340485685695,
+      "loss": 1.1343,
+      "step": 5066
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014300840699277124,
+      "loss": 1.7709,
+      "step": 5067
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014298340583073532,
+      "loss": 1.3569,
+      "step": 5068
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001429584013726663,
+      "loss": 1.51,
+      "step": 5069
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014293339362048151,
+      "loss": 1.6292,
+      "step": 5070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014290838257609867,
+      "loss": 1.5465,
+      "step": 5071
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014288336824143556,
+      "loss": 1.4806,
+      "step": 5072
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014285835061841038,
+      "loss": 1.251,
+      "step": 5073
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001428333297089415,
+      "loss": 1.6241,
+      "step": 5074
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014280830551494756,
+      "loss": 1.7141,
+      "step": 5075
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014278327803834753,
+      "loss": 1.4119,
+      "step": 5076
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014275824728106045,
+      "loss": 1.4152,
+      "step": 5077
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014273321324500575,
+      "loss": 1.3799,
+      "step": 5078
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014270817593210308,
+      "loss": 1.328,
+      "step": 5079
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014268313534427233,
+      "loss": 1.8779,
+      "step": 5080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001426580914834337,
+      "loss": 1.3003,
+      "step": 5081
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014263304435150755,
+      "loss": 1.7711,
+      "step": 5082
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014260799395041454,
+      "loss": 1.6023,
+      "step": 5083
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014258294028207552,
+      "loss": 1.6546,
+      "step": 5084
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014255788334841175,
+      "loss": 1.5842,
+      "step": 5085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014253282315134454,
+      "loss": 1.3664,
+      "step": 5086
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001425077596927956,
+      "loss": 1.4016,
+      "step": 5087
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001424826929746868,
+      "loss": 1.5105,
+      "step": 5088
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014245762299894035,
+      "loss": 1.7539,
+      "step": 5089
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001424325497674786,
+      "loss": 1.9051,
+      "step": 5090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001424074732822242,
+      "loss": 1.4813,
+      "step": 5091
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014238239354510005,
+      "loss": 1.2411,
+      "step": 5092
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014235731055802934,
+      "loss": 0.9343,
+      "step": 5093
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001423322243229355,
+      "loss": 1.2985,
+      "step": 5094
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001423071348417421,
+      "loss": 1.1862,
+      "step": 5095
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014228204211637306,
+      "loss": 1.494,
+      "step": 5096
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014225694614875258,
+      "loss": 1.2164,
+      "step": 5097
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000142231846940805,
+      "loss": 1.2808,
+      "step": 5098
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014220674449445496,
+      "loss": 1.2834,
+      "step": 5099
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014218163881162742,
+      "loss": 1.6024,
+      "step": 5100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014215652989424746,
+      "loss": 1.8412,
+      "step": 5101
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014213141774424052,
+      "loss": 1.6369,
+      "step": 5102
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014210630236353219,
+      "loss": 1.6643,
+      "step": 5103
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014208118375404836,
+      "loss": 1.3569,
+      "step": 5104
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001420560619177152,
+      "loss": 1.3349,
+      "step": 5105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014203093685645908,
+      "loss": 1.2661,
+      "step": 5106
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014200580857220662,
+      "loss": 1.2841,
+      "step": 5107
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014198067706688467,
+      "loss": 1.7825,
+      "step": 5108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014195554234242038,
+      "loss": 1.6859,
+      "step": 5109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014193040440074113,
+      "loss": 1.4914,
+      "step": 5110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014190526324377452,
+      "loss": 1.2618,
+      "step": 5111
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014188011887344837,
+      "loss": 1.6756,
+      "step": 5112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001418549712916909,
+      "loss": 1.603,
+      "step": 5113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014182982050043035,
+      "loss": 1.5378,
+      "step": 5114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014180466650159538,
+      "loss": 1.1569,
+      "step": 5115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014177950929711484,
+      "loss": 1.5163,
+      "step": 5116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014175434888891777,
+      "loss": 1.1224,
+      "step": 5117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014172918527893358,
+      "loss": 1.8612,
+      "step": 5118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014170401846909177,
+      "loss": 1.1039,
+      "step": 5119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014167884846132226,
+      "loss": 1.3924,
+      "step": 5120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001416536752575551,
+      "loss": 1.7254,
+      "step": 5121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014162849885972055,
+      "loss": 1.3662,
+      "step": 5122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014160331926974925,
+      "loss": 1.7032,
+      "step": 5123
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014157813648957196,
+      "loss": 1.1142,
+      "step": 5124
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014155295052111974,
+      "loss": 1.3628,
+      "step": 5125
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014152776136632393,
+      "loss": 1.9071,
+      "step": 5126
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014150256902711604,
+      "loss": 1.5887,
+      "step": 5127
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014147737350542783,
+      "loss": 0.9998,
+      "step": 5128
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014145217480319137,
+      "loss": 0.2717,
+      "step": 5129
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014142697292233894,
+      "loss": 1.3605,
+      "step": 5130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014140176786480303,
+      "loss": 0.9696,
+      "step": 5131
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001413765596325164,
+      "loss": 1.5894,
+      "step": 5132
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001413513482274121,
+      "loss": 1.6513,
+      "step": 5133
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014132613365142337,
+      "loss": 1.5321,
+      "step": 5134
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014130091590648362,
+      "loss": 1.4156,
+      "step": 5135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001412756949945267,
+      "loss": 1.1186,
+      "step": 5136
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014125047091748653,
+      "loss": 1.2757,
+      "step": 5137
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014122524367729734,
+      "loss": 1.5915,
+      "step": 5138
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014120001327589358,
+      "loss": 1.3794,
+      "step": 5139
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014117477971520995,
+      "loss": 1.6961,
+      "step": 5140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014114954299718143,
+      "loss": 1.1918,
+      "step": 5141
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001411243031237432,
+      "loss": 1.4277,
+      "step": 5142
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014109906009683064,
+      "loss": 1.7073,
+      "step": 5143
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014107381391837952,
+      "loss": 1.1076,
+      "step": 5144
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001410485645903257,
+      "loss": 1.2239,
+      "step": 5145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014102331211460531,
+      "loss": 1.7744,
+      "step": 5146
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001409980564931548,
+      "loss": 1.2967,
+      "step": 5147
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014097279772791078,
+      "loss": 1.529,
+      "step": 5148
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014094753582081016,
+      "loss": 1.4575,
+      "step": 5149
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014092227077379006,
+      "loss": 1.4156,
+      "step": 5150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014089700258878776,
+      "loss": 1.8002,
+      "step": 5151
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000140871731267741,
+      "loss": 1.3968,
+      "step": 5152
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001408464568125875,
+      "loss": 1.1245,
+      "step": 5153
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014082117922526542,
+      "loss": 1.3434,
+      "step": 5154
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014079589850771304,
+      "loss": 1.3481,
+      "step": 5155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014077061466186895,
+      "loss": 1.5569,
+      "step": 5156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014074532768967196,
+      "loss": 1.4543,
+      "step": 5157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001407200375930611,
+      "loss": 0.5551,
+      "step": 5158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014069474437397562,
+      "loss": 1.4039,
+      "step": 5159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014066944803435512,
+      "loss": 1.4977,
+      "step": 5160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014064414857613928,
+      "loss": 0.9109,
+      "step": 5161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014061884600126813,
+      "loss": 1.2736,
+      "step": 5162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014059354031168198,
+      "loss": 1.3744,
+      "step": 5163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001405682315093212,
+      "loss": 1.3708,
+      "step": 5164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014054291959612655,
+      "loss": 1.5039,
+      "step": 5165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000140517604574039,
+      "loss": 1.8991,
+      "step": 5166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014049228644499976,
+      "loss": 1.6103,
+      "step": 5167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001404669652109502,
+      "loss": 1.8382,
+      "step": 5168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014044164087383205,
+      "loss": 1.6856,
+      "step": 5169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001404163134355872,
+      "loss": 1.4221,
+      "step": 5170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014039098289815783,
+      "loss": 2.0179,
+      "step": 5171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014036564926348622,
+      "loss": 1.0441,
+      "step": 5172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001403403125335151,
+      "loss": 1.8829,
+      "step": 5173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014031497271018724,
+      "loss": 0.8936,
+      "step": 5174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014028962979544584,
+      "loss": 1.5333,
+      "step": 5175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001402642837912342,
+      "loss": 1.1296,
+      "step": 5176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014023893469949583,
+      "loss": 1.2259,
+      "step": 5177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014021358252217455,
+      "loss": 1.5265,
+      "step": 5178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014018822726121448,
+      "loss": 1.6359,
+      "step": 5179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014016286891855984,
+      "loss": 1.7316,
+      "step": 5180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014013750749615516,
+      "loss": 1.7244,
+      "step": 5181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014011214299594518,
+      "loss": 1.4568,
+      "step": 5182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014008677541987486,
+      "loss": 1.1556,
+      "step": 5183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014006140476988953,
+      "loss": 1.4459,
+      "step": 5184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014003603104793455,
+      "loss": 1.5721,
+      "step": 5185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014001065425595563,
+      "loss": 1.7702,
+      "step": 5186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013998527439589874,
+      "loss": 0.8648,
+      "step": 5187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013995989146971006,
+      "loss": 1.0783,
+      "step": 5188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001399345054793359,
+      "loss": 1.617,
+      "step": 5189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000139909116426723,
+      "loss": 1.514,
+      "step": 5190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013988372431381814,
+      "loss": 1.2925,
+      "step": 5191
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013985832914256848,
+      "loss": 1.2672,
+      "step": 5192
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013983293091492137,
+      "loss": 1.54,
+      "step": 5193
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013980752963282434,
+      "loss": 1.6331,
+      "step": 5194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013978212529822527,
+      "loss": 1.4959,
+      "step": 5195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013975671791307214,
+      "loss": 1.3522,
+      "step": 5196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013973130747931323,
+      "loss": 1.0011,
+      "step": 5197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013970589399889705,
+      "loss": 1.1091,
+      "step": 5198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001396804774737724,
+      "loss": 1.6857,
+      "step": 5199
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013965505790588823,
+      "loss": 1.2828,
+      "step": 5200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013962963529719368,
+      "loss": 1.2887,
+      "step": 5201
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001396042096496383,
+      "loss": 1.8806,
+      "step": 5202
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001395787809651717,
+      "loss": 1.6152,
+      "step": 5203
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001395533492457438,
+      "loss": 1.1075,
+      "step": 5204
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001395279144933048,
+      "loss": 1.6314,
+      "step": 5205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013950247670980502,
+      "loss": 1.7813,
+      "step": 5206
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013947703589719504,
+      "loss": 0.9326,
+      "step": 5207
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001394515920574258,
+      "loss": 1.4087,
+      "step": 5208
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013942614519244826,
+      "loss": 1.5332,
+      "step": 5209
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013940069530421375,
+      "loss": 1.463,
+      "step": 5210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001393752423946739,
+      "loss": 1.208,
+      "step": 5211
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013934978646578037,
+      "loss": 1.5137,
+      "step": 5212
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013932432751948524,
+      "loss": 1.2027,
+      "step": 5213
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013929886555774065,
+      "loss": 1.8614,
+      "step": 5214
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013927340058249914,
+      "loss": 1.9919,
+      "step": 5215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013924793259571336,
+      "loss": 1.5595,
+      "step": 5216
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013922246159933624,
+      "loss": 1.6385,
+      "step": 5217
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000139196987595321,
+      "loss": 0.9384,
+      "step": 5218
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013917151058562093,
+      "loss": 1.6148,
+      "step": 5219
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013914603057218968,
+      "loss": 1.723,
+      "step": 5220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013912054755698113,
+      "loss": 1.1633,
+      "step": 5221
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013909506154194927,
+      "loss": 1.695,
+      "step": 5222
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001390695725290485,
+      "loss": 1.3681,
+      "step": 5223
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013904408052023334,
+      "loss": 1.2448,
+      "step": 5224
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013901858551745852,
+      "loss": 1.2062,
+      "step": 5225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001389930875226791,
+      "loss": 1.2971,
+      "step": 5226
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013896758653785018,
+      "loss": 1.5917,
+      "step": 5227
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013894208256492733,
+      "loss": 1.5653,
+      "step": 5228
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013891657560586616,
+      "loss": 1.1697,
+      "step": 5229
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013889106566262267,
+      "loss": 1.296,
+      "step": 5230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013886555273715293,
+      "loss": 1.3675,
+      "step": 5231
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013884003683141335,
+      "loss": 2.0996,
+      "step": 5232
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001388145179473605,
+      "loss": 1.4087,
+      "step": 5233
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001387889960869512,
+      "loss": 1.6249,
+      "step": 5234
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001387634712521425,
+      "loss": 1.1834,
+      "step": 5235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013873794344489176,
+      "loss": 1.5314,
+      "step": 5236
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001387124126671564,
+      "loss": 1.5362,
+      "step": 5237
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013868687892089422,
+      "loss": 1.7753,
+      "step": 5238
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001386613422080631,
+      "loss": 1.1092,
+      "step": 5239
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013863580253062137,
+      "loss": 1.3504,
+      "step": 5240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013861025989052731,
+      "loss": 1.8615,
+      "step": 5241
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001385847142897397,
+      "loss": 1.4756,
+      "step": 5242
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001385591657302173,
+      "loss": 1.7203,
+      "step": 5243
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001385336142139193,
+      "loss": 0.9579,
+      "step": 5244
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013850805974280496,
+      "loss": 1.5128,
+      "step": 5245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001384825023188339,
+      "loss": 1.3045,
+      "step": 5246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013845694194396585,
+      "loss": 2.4584,
+      "step": 5247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013843137862016083,
+      "loss": 1.4941,
+      "step": 5248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001384058123493791,
+      "loss": 1.2891,
+      "step": 5249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001383802431335811,
+      "loss": 1.294,
+      "step": 5250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013835467097472753,
+      "loss": 1.283,
+      "step": 5251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013832909587477928,
+      "loss": 1.3102,
+      "step": 5252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001383035178356975,
+      "loss": 1.5602,
+      "step": 5253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013827793685944356,
+      "loss": 1.8568,
+      "step": 5254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013825235294797903,
+      "loss": 0.936,
+      "step": 5255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013822676610326576,
+      "loss": 1.968,
+      "step": 5256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013820117632726574,
+      "loss": 1.368,
+      "step": 5257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013817558362194129,
+      "loss": 1.3695,
+      "step": 5258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013814998798925482,
+      "loss": 1.3732,
+      "step": 5259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001381243894311691,
+      "loss": 1.3266,
+      "step": 5260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013809878794964706,
+      "loss": 0.9812,
+      "step": 5261
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013807318354665187,
+      "loss": 1.5182,
+      "step": 5262
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013804757622414688,
+      "loss": 1.3968,
+      "step": 5263
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013802196598409572,
+      "loss": 1.4114,
+      "step": 5264
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001379963528284622,
+      "loss": 1.4729,
+      "step": 5265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013797073675921043,
+      "loss": 1.8748,
+      "step": 5266
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001379451177783046,
+      "loss": 1.9769,
+      "step": 5267
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001379194958877093,
+      "loss": 1.5205,
+      "step": 5268
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013789387108938925,
+      "loss": 1.7377,
+      "step": 5269
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013786824338530936,
+      "loss": 1.6239,
+      "step": 5270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013784261277743478,
+      "loss": 1.2415,
+      "step": 5271
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013781697926773095,
+      "loss": 1.3918,
+      "step": 5272
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013779134285816347,
+      "loss": 1.5901,
+      "step": 5273
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001377657035506982,
+      "loss": 1.752,
+      "step": 5274
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013774006134730117,
+      "loss": 1.3888,
+      "step": 5275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001377144162499387,
+      "loss": 1.0202,
+      "step": 5276
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013768876826057724,
+      "loss": 1.3142,
+      "step": 5277
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013766311738118355,
+      "loss": 1.5359,
+      "step": 5278
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013763746361372458,
+      "loss": 1.8425,
+      "step": 5279
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013761180696016752,
+      "loss": 1.7212,
+      "step": 5280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001375861474224797,
+      "loss": 1.8221,
+      "step": 5281
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001375604850026288,
+      "loss": 1.0376,
+      "step": 5282
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001375348197025826,
+      "loss": 1.6065,
+      "step": 5283
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001375091515243092,
+      "loss": 1.6612,
+      "step": 5284
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013748348046977685,
+      "loss": 1.624,
+      "step": 5285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013745780654095407,
+      "loss": 1.3235,
+      "step": 5286
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013743212973980955,
+      "loss": 1.4241,
+      "step": 5287
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001374064500683122,
+      "loss": 1.1727,
+      "step": 5288
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013738076752843124,
+      "loss": 1.2226,
+      "step": 5289
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013735508212213597,
+      "loss": 1.3612,
+      "step": 5290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013732939385139608,
+      "loss": 1.0732,
+      "step": 5291
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001373037027181813,
+      "loss": 1.4612,
+      "step": 5292
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013727800872446173,
+      "loss": 1.4822,
+      "step": 5293
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001372523118722076,
+      "loss": 1.3607,
+      "step": 5294
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013722661216338937,
+      "loss": 1.1559,
+      "step": 5295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013720090959997772,
+      "loss": 1.1989,
+      "step": 5296
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013717520418394358,
+      "loss": 1.331,
+      "step": 5297
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013714949591725812,
+      "loss": 1.8195,
+      "step": 5298
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013712378480189264,
+      "loss": 1.3126,
+      "step": 5299
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001370980708398187,
+      "loss": 1.8032,
+      "step": 5300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013707235403300817,
+      "loss": 1.4864,
+      "step": 5301
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013704663438343295,
+      "loss": 1.2086,
+      "step": 5302
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001370209118930653,
+      "loss": 1.4956,
+      "step": 5303
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013699518656387765,
+      "loss": 1.6046,
+      "step": 5304
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001369694583978427,
+      "loss": 1.6637,
+      "step": 5305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013694372739693332,
+      "loss": 0.91,
+      "step": 5306
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013691799356312253,
+      "loss": 1.1693,
+      "step": 5307
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013689225689838375,
+      "loss": 1.4717,
+      "step": 5308
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013686651740469035,
+      "loss": 1.1566,
+      "step": 5309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001368407750840162,
+      "loss": 1.1383,
+      "step": 5310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013681502993833527,
+      "loss": 1.4116,
+      "step": 5311
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013678928196962165,
+      "loss": 1.3178,
+      "step": 5312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001367635311798498,
+      "loss": 1.4321,
+      "step": 5313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013673777757099436,
+      "loss": 0.77,
+      "step": 5314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013671202114503002,
+      "loss": 1.4001,
+      "step": 5315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013668626190393196,
+      "loss": 1.6081,
+      "step": 5316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013666049984967536,
+      "loss": 0.7397,
+      "step": 5317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013663473498423572,
+      "loss": 1.2888,
+      "step": 5318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013660896730958877,
+      "loss": 1.9924,
+      "step": 5319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013658319682771032,
+      "loss": 1.2844,
+      "step": 5320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013655742354057654,
+      "loss": 1.4016,
+      "step": 5321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013653164745016378,
+      "loss": 0.7781,
+      "step": 5322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001365058685584486,
+      "loss": 1.2913,
+      "step": 5323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001364800868674077,
+      "loss": 2.1311,
+      "step": 5324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013645430237901814,
+      "loss": 1.4831,
+      "step": 5325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013642851509525707,
+      "loss": 1.4697,
+      "step": 5326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013640272501810188,
+      "loss": 1.5638,
+      "step": 5327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013637693214953024,
+      "loss": 1.8941,
+      "step": 5328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013635113649151993,
+      "loss": 1.4699,
+      "step": 5329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013632533804604907,
+      "loss": 1.6914,
+      "step": 5330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001362995368150959,
+      "loss": 1.8466,
+      "step": 5331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013627373280063884,
+      "loss": 1.212,
+      "step": 5332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013624792600465665,
+      "loss": 1.2501,
+      "step": 5333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001362221164291282,
+      "loss": 1.4784,
+      "step": 5334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001361963040760326,
+      "loss": 1.6558,
+      "step": 5335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013617048894734923,
+      "loss": 1.5924,
+      "step": 5336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001361446710450576,
+      "loss": 1.18,
+      "step": 5337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013611885037113748,
+      "loss": 1.2556,
+      "step": 5338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001360930269275688,
+      "loss": 0.9195,
+      "step": 5339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013606720071633176,
+      "loss": 1.5146,
+      "step": 5340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013604137173940677,
+      "loss": 1.257,
+      "step": 5341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013601553999877442,
+      "loss": 0.9112,
+      "step": 5342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013598970549641556,
+      "loss": 1.794,
+      "step": 5343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013596386823431117,
+      "loss": 1.2495,
+      "step": 5344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013593802821444253,
+      "loss": 1.3529,
+      "step": 5345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013591218543879108,
+      "loss": 2.1447,
+      "step": 5346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013588633990933843,
+      "loss": 1.2987,
+      "step": 5347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001358604916280666,
+      "loss": 1.2975,
+      "step": 5348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013583464059695747,
+      "loss": 1.4231,
+      "step": 5349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001358087868179935,
+      "loss": 2.1568,
+      "step": 5350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013578293029315716,
+      "loss": 1.1631,
+      "step": 5351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013575707102443112,
+      "loss": 1.8443,
+      "step": 5352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013573120901379834,
+      "loss": 1.5767,
+      "step": 5353
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013570534426324198,
+      "loss": 1.4931,
+      "step": 5354
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013567947677474532,
+      "loss": 1.4412,
+      "step": 5355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013565360655029202,
+      "loss": 0.8919,
+      "step": 5356
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013562773359186576,
+      "loss": 1.3208,
+      "step": 5357
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013560185790145054,
+      "loss": 1.7515,
+      "step": 5358
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013557597948103057,
+      "loss": 1.3308,
+      "step": 5359
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013555009833259024,
+      "loss": 1.4992,
+      "step": 5360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013552421445811418,
+      "loss": 1.2068,
+      "step": 5361
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013549832785958714,
+      "loss": 1.1521,
+      "step": 5362
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001354724385389942,
+      "loss": 1.0264,
+      "step": 5363
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013544654649832056,
+      "loss": 1.472,
+      "step": 5364
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013542065173955162,
+      "loss": 1.8685,
+      "step": 5365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013539475426467315,
+      "loss": 1.7995,
+      "step": 5366
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001353688540756709,
+      "loss": 1.5883,
+      "step": 5367
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013534295117453103,
+      "loss": 1.1931,
+      "step": 5368
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013531704556323974,
+      "loss": 1.0183,
+      "step": 5369
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001352911372437835,
+      "loss": 1.1806,
+      "step": 5370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013526522621814909,
+      "loss": 1.0648,
+      "step": 5371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001352393124883233,
+      "loss": 1.9281,
+      "step": 5372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013521339605629325,
+      "loss": 1.4856,
+      "step": 5373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013518747692404634,
+      "loss": 1.0017,
+      "step": 5374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013516155509357006,
+      "loss": 1.408,
+      "step": 5375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013513563056685206,
+      "loss": 1.5583,
+      "step": 5376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013510970334588033,
+      "loss": 1.2411,
+      "step": 5377
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000135083773432643,
+      "loss": 1.27,
+      "step": 5378
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013505784082912841,
+      "loss": 1.1949,
+      "step": 5379
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013503190553732514,
+      "loss": 1.6844,
+      "step": 5380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001350059675592219,
+      "loss": 1.3165,
+      "step": 5381
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013498002689680768,
+      "loss": 1.2267,
+      "step": 5382
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013495408355207168,
+      "loss": 1.3798,
+      "step": 5383
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001349281375270032,
+      "loss": 1.5245,
+      "step": 5384
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013490218882359188,
+      "loss": 2.0332,
+      "step": 5385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013487623744382749,
+      "loss": 1.2623,
+      "step": 5386
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013485028338970002,
+      "loss": 1.611,
+      "step": 5387
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013482432666319969,
+      "loss": 1.2404,
+      "step": 5388
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013479836726631685,
+      "loss": 1.4915,
+      "step": 5389
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013477240520104212,
+      "loss": 1.6964,
+      "step": 5390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013474644046936634,
+      "loss": 1.5412,
+      "step": 5391
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013472047307328054,
+      "loss": 1.1648,
+      "step": 5392
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013469450301477592,
+      "loss": 1.5308,
+      "step": 5393
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013466853029584386,
+      "loss": 1.9243,
+      "step": 5394
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013464255491847603,
+      "loss": 1.0266,
+      "step": 5395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013461657688466422,
+      "loss": 1.228,
+      "step": 5396
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013459059619640054,
+      "loss": 1.1529,
+      "step": 5397
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001345646128556772,
+      "loss": 1.1965,
+      "step": 5398
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001345386268644866,
+      "loss": 1.5642,
+      "step": 5399
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001345126382248214,
+      "loss": 1.0464,
+      "step": 5400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013448664693867452,
+      "loss": 1.488,
+      "step": 5401
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001344606530080389,
+      "loss": 1.6834,
+      "step": 5402
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013443465643490788,
+      "loss": 1.7276,
+      "step": 5403
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013440865722127486,
+      "loss": 1.4373,
+      "step": 5404
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013438265536913356,
+      "loss": 1.2474,
+      "step": 5405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001343566508804778,
+      "loss": 1.5379,
+      "step": 5406
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013433064375730164,
+      "loss": 1.3997,
+      "step": 5407
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013430463400159937,
+      "loss": 1.2993,
+      "step": 5408
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013427862161536543,
+      "loss": 1.5273,
+      "step": 5409
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001342526066005945,
+      "loss": 1.1538,
+      "step": 5410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013422658895928148,
+      "loss": 2.0437,
+      "step": 5411
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013420056869342138,
+      "loss": 1.724,
+      "step": 5412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013417454580500954,
+      "loss": 1.3238,
+      "step": 5413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013414852029604142,
+      "loss": 2.4156,
+      "step": 5414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013412249216851265,
+      "loss": 1.3427,
+      "step": 5415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013409646142441915,
+      "loss": 1.5986,
+      "step": 5416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000134070428065757,
+      "loss": 1.4596,
+      "step": 5417
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013404439209452245,
+      "loss": 1.567,
+      "step": 5418
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013401835351271196,
+      "loss": 1.5685,
+      "step": 5419
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001339923123223223,
+      "loss": 1.4771,
+      "step": 5420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013396626852535024,
+      "loss": 1.1652,
+      "step": 5421
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013394022212379295,
+      "loss": 1.2693,
+      "step": 5422
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001339141731196476,
+      "loss": 1.7086,
+      "step": 5423
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001338881215149118,
+      "loss": 1.5818,
+      "step": 5424
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013386206731158312,
+      "loss": 1.0193,
+      "step": 5425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001338360105116595,
+      "loss": 1.8734,
+      "step": 5426
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013380995111713894,
+      "loss": 1.9599,
+      "step": 5427
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013378388913001978,
+      "loss": 1.5489,
+      "step": 5428
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013375782455230052,
+      "loss": 1.8269,
+      "step": 5429
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013373175738597975,
+      "loss": 3.3254,
+      "step": 5430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013370568763305637,
+      "loss": 1.3945,
+      "step": 5431
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013367961529552947,
+      "loss": 1.6536,
+      "step": 5432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001336535403753983,
+      "loss": 1.1533,
+      "step": 5433
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013362746287466233,
+      "loss": 0.8765,
+      "step": 5434
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013360138279532122,
+      "loss": 1.6095,
+      "step": 5435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013357530013937481,
+      "loss": 1.1635,
+      "step": 5436
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013354921490882322,
+      "loss": 1.0135,
+      "step": 5437
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013352312710566662,
+      "loss": 1.6913,
+      "step": 5438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013349703673190555,
+      "loss": 1.791,
+      "step": 5439
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013347094378954056,
+      "loss": 1.3153,
+      "step": 5440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001334448482805726,
+      "loss": 1.8257,
+      "step": 5441
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013341875020700264,
+      "loss": 0.9711,
+      "step": 5442
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013339264957083196,
+      "loss": 1.2598,
+      "step": 5443
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013336654637406198,
+      "loss": 1.7434,
+      "step": 5444
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013334044061869433,
+      "loss": 1.8005,
+      "step": 5445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013331433230673082,
+      "loss": 1.6024,
+      "step": 5446
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013328822144017353,
+      "loss": 2.0566,
+      "step": 5447
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013326210802102465,
+      "loss": 1.9918,
+      "step": 5448
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013323599205128657,
+      "loss": 1.4331,
+      "step": 5449
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013320987353296197,
+      "loss": 1.3959,
+      "step": 5450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013318375246805362,
+      "loss": 1.7761,
+      "step": 5451
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013315762885856446,
+      "loss": 1.0134,
+      "step": 5452
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013313150270649778,
+      "loss": 1.5569,
+      "step": 5453
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013310537401385697,
+      "loss": 2.0174,
+      "step": 5454
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013307924278264555,
+      "loss": 1.5795,
+      "step": 5455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013305310901486737,
+      "loss": 1.8581,
+      "step": 5456
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001330269727125264,
+      "loss": 1.4551,
+      "step": 5457
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013300083387762673,
+      "loss": 1.9523,
+      "step": 5458
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013297469251217283,
+      "loss": 2.1076,
+      "step": 5459
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013294854861816918,
+      "loss": 1.0407,
+      "step": 5460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001329224021976206,
+      "loss": 1.7665,
+      "step": 5461
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013289625325253203,
+      "loss": 1.1486,
+      "step": 5462
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013287010178490854,
+      "loss": 1.4478,
+      "step": 5463
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013284394779675552,
+      "loss": 0.9692,
+      "step": 5464
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013281779129007852,
+      "loss": 1.2307,
+      "step": 5465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013279163226688321,
+      "loss": 1.3087,
+      "step": 5466
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013276547072917555,
+      "loss": 1.4167,
+      "step": 5467
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001327393066789616,
+      "loss": 1.2836,
+      "step": 5468
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001327131401182477,
+      "loss": 1.4892,
+      "step": 5469
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001326869710490403,
+      "loss": 1.6363,
+      "step": 5470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001326607994733461,
+      "loss": 1.5618,
+      "step": 5471
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000132634625393172,
+      "loss": 1.3936,
+      "step": 5472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013260844881052508,
+      "loss": 1.5656,
+      "step": 5473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001325822697274125,
+      "loss": 1.6255,
+      "step": 5474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013255608814584187,
+      "loss": 1.157,
+      "step": 5475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001325299040678207,
+      "loss": 1.4109,
+      "step": 5476
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013250371749535687,
+      "loss": 1.4913,
+      "step": 5477
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013247752843045842,
+      "loss": 1.3729,
+      "step": 5478
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013245133687513357,
+      "loss": 1.5801,
+      "step": 5479
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013242514283139068,
+      "loss": 1.4306,
+      "step": 5480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013239894630123845,
+      "loss": 1.7177,
+      "step": 5481
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013237274728668553,
+      "loss": 1.1876,
+      "step": 5482
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000132346545789741,
+      "loss": 1.5004,
+      "step": 5483
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013232034181241402,
+      "loss": 0.8459,
+      "step": 5484
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001322941353567139,
+      "loss": 2.1803,
+      "step": 5485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013226792642465028,
+      "loss": 1.2376,
+      "step": 5486
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001322417150182328,
+      "loss": 1.4863,
+      "step": 5487
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001322155011394715,
+      "loss": 1.2317,
+      "step": 5488
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013218928479037637,
+      "loss": 1.553,
+      "step": 5489
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001321630659729578,
+      "loss": 1.1954,
+      "step": 5490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013213684468922634,
+      "loss": 1.4557,
+      "step": 5491
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013211062094119254,
+      "loss": 2.1058,
+      "step": 5492
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001320843947308674,
+      "loss": 1.4327,
+      "step": 5493
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013205816606026192,
+      "loss": 1.6316,
+      "step": 5494
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001320319349313874,
+      "loss": 1.154,
+      "step": 5495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001320057013462552,
+      "loss": 1.1236,
+      "step": 5496
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013197946530687705,
+      "loss": 1.7834,
+      "step": 5497
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001319532268152647,
+      "loss": 1.6631,
+      "step": 5498
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001319269858734302,
+      "loss": 1.4793,
+      "step": 5499
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013190074248338576,
+      "loss": 1.3851,
+      "step": 5500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001318744966471437,
+      "loss": 1.0712,
+      "step": 5501
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001318482483667166,
+      "loss": 2.1103,
+      "step": 5502
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013182199764411726,
+      "loss": 1.6275,
+      "step": 5503
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013179574448135862,
+      "loss": 1.6705,
+      "step": 5504
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013176948888045378,
+      "loss": 1.5659,
+      "step": 5505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001317432308434161,
+      "loss": 1.4606,
+      "step": 5506
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001317169703722591,
+      "loss": 1.0507,
+      "step": 5507
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013169070746899637,
+      "loss": 1.1309,
+      "step": 5508
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013166444213564188,
+      "loss": 1.0458,
+      "step": 5509
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013163817437420968,
+      "loss": 1.4731,
+      "step": 5510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013161190418671404,
+      "loss": 1.4036,
+      "step": 5511
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013158563157516934,
+      "loss": 1.4493,
+      "step": 5512
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013155935654159023,
+      "loss": 1.6561,
+      "step": 5513
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013153307908799158,
+      "loss": 1.3487,
+      "step": 5514
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013150679921638826,
+      "loss": 1.3777,
+      "step": 5515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013148051692879557,
+      "loss": 2.0976,
+      "step": 5516
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013145423222722883,
+      "loss": 1.356,
+      "step": 5517
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013142794511370357,
+      "loss": 1.6168,
+      "step": 5518
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013140165559023557,
+      "loss": 1.4152,
+      "step": 5519
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001313753636588407,
+      "loss": 1.8407,
+      "step": 5520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013134906932153512,
+      "loss": 1.5183,
+      "step": 5521
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013132277258033506,
+      "loss": 1.3216,
+      "step": 5522
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013129647343725707,
+      "loss": 1.1257,
+      "step": 5523
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013127017189431772,
+      "loss": 1.7254,
+      "step": 5524
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001312438679535339,
+      "loss": 0.965,
+      "step": 5525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001312175616169227,
+      "loss": 1.3419,
+      "step": 5526
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001311912528865012,
+      "loss": 1.5073,
+      "step": 5527
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013116494176428685,
+      "loss": 1.6618,
+      "step": 5528
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001311386282522973,
+      "loss": 1.3615,
+      "step": 5529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013111231235255017,
+      "loss": 1.7459,
+      "step": 5530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013108599406706352,
+      "loss": 1.5653,
+      "step": 5531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013105967339785543,
+      "loss": 1.032,
+      "step": 5532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013103335034694418,
+      "loss": 1.372,
+      "step": 5533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013100702491634832,
+      "loss": 1.1826,
+      "step": 5534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001309806971080865,
+      "loss": 1.4226,
+      "step": 5535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013095436692417759,
+      "loss": 1.5657,
+      "step": 5536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013092803436664057,
+      "loss": 1.0472,
+      "step": 5537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013090169943749476,
+      "loss": 1.056,
+      "step": 5538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013087536213875946,
+      "loss": 1.3829,
+      "step": 5539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001308490224724543,
+      "loss": 1.5152,
+      "step": 5540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001308226804405991,
+      "loss": 1.1279,
+      "step": 5541
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013079633604521374,
+      "loss": 1.5771,
+      "step": 5542
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013076998928831835,
+      "loss": 1.3762,
+      "step": 5543
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013074364017193326,
+      "loss": 1.682,
+      "step": 5544
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001307172886980789,
+      "loss": 1.5116,
+      "step": 5545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013069093486877604,
+      "loss": 1.768,
+      "step": 5546
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001306645786860455,
+      "loss": 1.3332,
+      "step": 5547
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013063822015190826,
+      "loss": 1.2296,
+      "step": 5548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013061185926838562,
+      "loss": 1.1268,
+      "step": 5549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001305854960374989,
+      "loss": 1.4204,
+      "step": 5550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001305591304612697,
+      "loss": 1.1594,
+      "step": 5551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013053276254171973,
+      "loss": 1.3165,
+      "step": 5552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000130506392280871,
+      "loss": 1.3939,
+      "step": 5553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013048001968074554,
+      "loss": 1.757,
+      "step": 5554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013045364474336573,
+      "loss": 1.3785,
+      "step": 5555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000130427267470754,
+      "loss": 1.3469,
+      "step": 5556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000130400887864933,
+      "loss": 0.795,
+      "step": 5557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001303745059279255,
+      "loss": 1.4568,
+      "step": 5558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013034812166175458,
+      "loss": 1.7645,
+      "step": 5559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013032173506844343,
+      "loss": 1.4139,
+      "step": 5560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013029534615001536,
+      "loss": 1.6261,
+      "step": 5561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013026895490849395,
+      "loss": 1.4986,
+      "step": 5562
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013024256134590292,
+      "loss": 1.7336,
+      "step": 5563
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013021616546426615,
+      "loss": 1.1286,
+      "step": 5564
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013018976726560778,
+      "loss": 1.6007,
+      "step": 5565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013016336675195195,
+      "loss": 1.5087,
+      "step": 5566
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013013696392532318,
+      "loss": 1.3809,
+      "step": 5567
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013011055878774604,
+      "loss": 1.3289,
+      "step": 5568
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013008415134124535,
+      "loss": 1.0807,
+      "step": 5569
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013005774158784603,
+      "loss": 1.2588,
+      "step": 5570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013003132952957324,
+      "loss": 1.1369,
+      "step": 5571
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013000491516845233,
+      "loss": 1.5596,
+      "step": 5572
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012997849850650875,
+      "loss": 1.7128,
+      "step": 5573
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012995207954576816,
+      "loss": 1.0438,
+      "step": 5574
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012992565828825645,
+      "loss": 1.9582,
+      "step": 5575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001298992347359996,
+      "loss": 1.5074,
+      "step": 5576
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012987280889102386,
+      "loss": 1.5118,
+      "step": 5577
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012984638075535555,
+      "loss": 1.3224,
+      "step": 5578
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012981995033102125,
+      "loss": 1.3036,
+      "step": 5579
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001297935176200477,
+      "loss": 1.0314,
+      "step": 5580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012976708262446175,
+      "loss": 0.9899,
+      "step": 5581
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001297406453462905,
+      "loss": 1.8428,
+      "step": 5582
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001297142057875612,
+      "loss": 1.4437,
+      "step": 5583
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012968776395030134,
+      "loss": 1.4054,
+      "step": 5584
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001296613198365384,
+      "loss": 1.2094,
+      "step": 5585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012963487344830026,
+      "loss": 2.1963,
+      "step": 5586
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001296084247876148,
+      "loss": 1.4899,
+      "step": 5587
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001295819738565102,
+      "loss": 1.4052,
+      "step": 5588
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012955552065701468,
+      "loss": 1.4422,
+      "step": 5589
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012952906519115678,
+      "loss": 1.0209,
+      "step": 5590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001295026074609651,
+      "loss": 1.7809,
+      "step": 5591
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012947614746846854,
+      "loss": 1.2687,
+      "step": 5592
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012944968521569602,
+      "loss": 1.5822,
+      "step": 5593
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012942322070467669,
+      "loss": 1.5825,
+      "step": 5594
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012939675393743995,
+      "loss": 1.1486,
+      "step": 5595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012937028491601525,
+      "loss": 1.4664,
+      "step": 5596
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012934381364243237,
+      "loss": 1.8533,
+      "step": 5597
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012931734011872105,
+      "loss": 1.3445,
+      "step": 5598
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001292908643469114,
+      "loss": 1.5175,
+      "step": 5599
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012926438632903362,
+      "loss": 1.706,
+      "step": 5600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012923790606711802,
+      "loss": 1.4857,
+      "step": 5601
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001292114235631952,
+      "loss": 1.368,
+      "step": 5602
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001291849388192959,
+      "loss": 1.5654,
+      "step": 5603
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012915845183745096,
+      "loss": 1.7108,
+      "step": 5604
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012913196261969147,
+      "loss": 2.1568,
+      "step": 5605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012910547116804863,
+      "loss": 1.594,
+      "step": 5606
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012907897748455388,
+      "loss": 1.3953,
+      "step": 5607
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001290524815712388,
+      "loss": 1.4039,
+      "step": 5608
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001290259834301351,
+      "loss": 1.3434,
+      "step": 5609
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012899948306327477,
+      "loss": 1.8236,
+      "step": 5610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012897298047268978,
+      "loss": 1.8117,
+      "step": 5611
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012894647566041254,
+      "loss": 0.8702,
+      "step": 5612
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012891996862847532,
+      "loss": 1.3253,
+      "step": 5613
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012889345937891085,
+      "loss": 1.7008,
+      "step": 5614
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012886694791375183,
+      "loss": 1.3116,
+      "step": 5615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012884043423503123,
+      "loss": 1.5948,
+      "step": 5616
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012881391834478214,
+      "loss": 1.3871,
+      "step": 5617
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012878740024503785,
+      "loss": 1.3051,
+      "step": 5618
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012876087993783178,
+      "loss": 1.1522,
+      "step": 5619
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012873435742519757,
+      "loss": 1.4646,
+      "step": 5620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012870783270916905,
+      "loss": 1.5606,
+      "step": 5621
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001286813057917801,
+      "loss": 1.6487,
+      "step": 5622
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001286547766750649,
+      "loss": 1.2015,
+      "step": 5623
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001286282453610577,
+      "loss": 1.4666,
+      "step": 5624
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000128601711851793,
+      "loss": 1.419,
+      "step": 5625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012857517614930542,
+      "loss": 1.4011,
+      "step": 5626
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001285486382556297,
+      "loss": 1.2157,
+      "step": 5627
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001285220981728009,
+      "loss": 1.6328,
+      "step": 5628
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012849555590285414,
+      "loss": 0.9581,
+      "step": 5629
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012846901144782463,
+      "loss": 1.1365,
+      "step": 5630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012844246480974797,
+      "loss": 1.2285,
+      "step": 5631
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012841591599065967,
+      "loss": 1.6216,
+      "step": 5632
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012838936499259563,
+      "loss": 0.876,
+      "step": 5633
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012836281181759175,
+      "loss": 1.4253,
+      "step": 5634
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012833625646768425,
+      "loss": 1.0208,
+      "step": 5635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012830969894490934,
+      "loss": 1.3611,
+      "step": 5636
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012828313925130358,
+      "loss": 1.3406,
+      "step": 5637
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012825657738890352,
+      "loss": 1.5636,
+      "step": 5638
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012823001335974602,
+      "loss": 1.5335,
+      "step": 5639
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012820344716586805,
+      "loss": 1.8299,
+      "step": 5640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012817687880930677,
+      "loss": 1.5272,
+      "step": 5641
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012815030829209942,
+      "loss": 1.3107,
+      "step": 5642
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012812373561628347,
+      "loss": 1.5374,
+      "step": 5643
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012809716078389662,
+      "loss": 1.3175,
+      "step": 5644
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001280705837969766,
+      "loss": 1.2812,
+      "step": 5645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001280440046575614,
+      "loss": 1.0337,
+      "step": 5646
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001280174233676892,
+      "loss": 1.4861,
+      "step": 5647
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012799083992939822,
+      "loss": 1.4224,
+      "step": 5648
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012796425434472694,
+      "loss": 1.6381,
+      "step": 5649
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012793766661571397,
+      "loss": 1.0027,
+      "step": 5650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012791107674439814,
+      "loss": 1.1029,
+      "step": 5651
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012788448473281835,
+      "loss": 1.6647,
+      "step": 5652
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012785789058301377,
+      "loss": 1.2576,
+      "step": 5653
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012783129429702362,
+      "loss": 1.7524,
+      "step": 5654
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012780469587688744,
+      "loss": 1.3771,
+      "step": 5655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001277780953246447,
+      "loss": 1.1761,
+      "step": 5656
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012775149264233527,
+      "loss": 1.0964,
+      "step": 5657
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012772488783199907,
+      "loss": 2.7106,
+      "step": 5658
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001276982808956762,
+      "loss": 2.007,
+      "step": 5659
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012767167183540688,
+      "loss": 1.6114,
+      "step": 5660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012764506065323156,
+      "loss": 1.8488,
+      "step": 5661
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012761844735119084,
+      "loss": 1.1795,
+      "step": 5662
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012759183193132546,
+      "loss": 0.9404,
+      "step": 5663
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001275652143956763,
+      "loss": 1.0316,
+      "step": 5664
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012753859474628446,
+      "loss": 1.5314,
+      "step": 5665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001275119729851912,
+      "loss": 0.9335,
+      "step": 5666
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012748534911443787,
+      "loss": 1.3252,
+      "step": 5667
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012745872313606608,
+      "loss": 1.4282,
+      "step": 5668
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012743209505211749,
+      "loss": 1.7898,
+      "step": 5669
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012740546486463404,
+      "loss": 1.4833,
+      "step": 5670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012737883257565773,
+      "loss": 1.3057,
+      "step": 5671
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012735219818723077,
+      "loss": 0.9438,
+      "step": 5672
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012732556170139554,
+      "loss": 1.2207,
+      "step": 5673
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001272989231201946,
+      "loss": 1.172,
+      "step": 5674
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012727228244567055,
+      "loss": 1.5093,
+      "step": 5675
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012724563967986632,
+      "loss": 1.4878,
+      "step": 5676
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012721899482482485,
+      "loss": 1.7375,
+      "step": 5677
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001271923478825894,
+      "loss": 1.6713,
+      "step": 5678
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001271656988552032,
+      "loss": 1.1158,
+      "step": 5679
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001271390477447098,
+      "loss": 1.4194,
+      "step": 5680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012711239455315282,
+      "loss": 1.3101,
+      "step": 5681
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012708573928257605,
+      "loss": 1.4929,
+      "step": 5682
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012705908193502352,
+      "loss": 1.313,
+      "step": 5683
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012703242251253932,
+      "loss": 1.8143,
+      "step": 5684
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001270057610171677,
+      "loss": 1.8372,
+      "step": 5685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012697909745095314,
+      "loss": 1.5314,
+      "step": 5686
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012695243181594026,
+      "loss": 1.5834,
+      "step": 5687
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012692576411417377,
+      "loss": 2.4548,
+      "step": 5688
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012689909434769866,
+      "loss": 1.5288,
+      "step": 5689
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012687242251855998,
+      "loss": 1.4746,
+      "step": 5690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001268457486288029,
+      "loss": 1.3973,
+      "step": 5691
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012681907268047296,
+      "loss": 1.1927,
+      "step": 5692
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012679239467561554,
+      "loss": 1.6085,
+      "step": 5693
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012676571461627646,
+      "loss": 0.8985,
+      "step": 5694
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012673903250450157,
+      "loss": 1.5173,
+      "step": 5695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012671234834233687,
+      "loss": 1.318,
+      "step": 5696
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012668566213182858,
+      "loss": 1.4135,
+      "step": 5697
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000126658973875023,
+      "loss": 1.1096,
+      "step": 5698
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012663228357396667,
+      "loss": 1.7991,
+      "step": 5699
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001266055912307062,
+      "loss": 1.1964,
+      "step": 5700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012657889684728838,
+      "loss": 1.347,
+      "step": 5701
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012655220042576026,
+      "loss": 1.5171,
+      "step": 5702
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012652550196816893,
+      "loss": 0.9818,
+      "step": 5703
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001264988014765616,
+      "loss": 1.8586,
+      "step": 5704
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012647209895298578,
+      "loss": 1.5713,
+      "step": 5705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012644539439948907,
+      "loss": 1.4593,
+      "step": 5706
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012641868781811913,
+      "loss": 1.3947,
+      "step": 5707
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012639197921092396,
+      "loss": 1.3421,
+      "step": 5708
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012636526857995155,
+      "loss": 1.4094,
+      "step": 5709
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012633855592725015,
+      "loss": 1.5889,
+      "step": 5710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001263118412548681,
+      "loss": 1.4857,
+      "step": 5711
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012628512456485392,
+      "loss": 1.4402,
+      "step": 5712
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012625840585925633,
+      "loss": 1.293,
+      "step": 5713
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001262316851401241,
+      "loss": 1.4794,
+      "step": 5714
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012620496240950627,
+      "loss": 1.5102,
+      "step": 5715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012617823766945195,
+      "loss": 1.5226,
+      "step": 5716
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012615151092201042,
+      "loss": 1.4288,
+      "step": 5717
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012612478216923119,
+      "loss": 1.3102,
+      "step": 5718
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012609805141316376,
+      "loss": 1.0447,
+      "step": 5719
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012607131865585797,
+      "loss": 1.1228,
+      "step": 5720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012604458389936374,
+      "loss": 1.4973,
+      "step": 5721
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012601784714573106,
+      "loss": 1.3206,
+      "step": 5722
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001259911083970102,
+      "loss": 1.7864,
+      "step": 5723
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001259643676552515,
+      "loss": 1.2413,
+      "step": 5724
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012593762492250544,
+      "loss": 1.7227,
+      "step": 5725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012591088020082278,
+      "loss": 1.8916,
+      "step": 5726
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012588413349225432,
+      "loss": 1.4507,
+      "step": 5727
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000125857384798851,
+      "loss": 1.5262,
+      "step": 5728
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012583063412266403,
+      "loss": 1.373,
+      "step": 5729
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001258038814657446,
+      "loss": 1.1975,
+      "step": 5730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012577712683014416,
+      "loss": 1.6978,
+      "step": 5731
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012575037021791437,
+      "loss": 1.3195,
+      "step": 5732
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012572361163110687,
+      "loss": 1.3154,
+      "step": 5733
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001256968510717736,
+      "loss": 1.4892,
+      "step": 5734
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012567008854196668,
+      "loss": 1.459,
+      "step": 5735
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012564332404373815,
+      "loss": 1.2949,
+      "step": 5736
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001256165575791404,
+      "loss": 1.7559,
+      "step": 5737
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012558978915022602,
+      "loss": 1.546,
+      "step": 5738
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012556301875904751,
+      "loss": 1.2922,
+      "step": 5739
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001255362464076578,
+      "loss": 1.4612,
+      "step": 5740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012550947209810973,
+      "loss": 1.6126,
+      "step": 5741
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001254826958324564,
+      "loss": 1.0572,
+      "step": 5742
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012545591761275115,
+      "loss": 1.0647,
+      "step": 5743
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012542913744104727,
+      "loss": 1.7462,
+      "step": 5744
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012540235531939838,
+      "loss": 1.6831,
+      "step": 5745
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012537557124985812,
+      "loss": 1.4531,
+      "step": 5746
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012534878523448036,
+      "loss": 1.4685,
+      "step": 5747
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001253219972753191,
+      "loss": 2.1084,
+      "step": 5748
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012529520737442846,
+      "loss": 1.7205,
+      "step": 5749
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012526841553386274,
+      "loss": 1.6148,
+      "step": 5750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012524162175567638,
+      "loss": 1.2529,
+      "step": 5751
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012521482604192395,
+      "loss": 1.7963,
+      "step": 5752
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012518802839466022,
+      "loss": 1.6665,
+      "step": 5753
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012516122881594006,
+      "loss": 1.5684,
+      "step": 5754
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012513442730781847,
+      "loss": 1.618,
+      "step": 5755
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012510762387235068,
+      "loss": 1.7393,
+      "step": 5756
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012508081851159202,
+      "loss": 1.2475,
+      "step": 5757
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001250540112275979,
+      "loss": 0.8644,
+      "step": 5758
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012502720202242404,
+      "loss": 1.4092,
+      "step": 5759
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012500039089812615,
+      "loss": 1.3228,
+      "step": 5760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012497357785676015,
+      "loss": 1.2652,
+      "step": 5761
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012494676290038207,
+      "loss": 1.0554,
+      "step": 5762
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012491994603104818,
+      "loss": 1.4162,
+      "step": 5763
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012489312725081486,
+      "loss": 1.363,
+      "step": 5764
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012486630656173854,
+      "loss": 0.9514,
+      "step": 5765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012483948396587594,
+      "loss": 1.2455,
+      "step": 5766
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001248126594652838,
+      "loss": 1.6247,
+      "step": 5767
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012478583306201909,
+      "loss": 1.3399,
+      "step": 5768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001247590047581389,
+      "loss": 1.4832,
+      "step": 5769
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012473217455570044,
+      "loss": 0.955,
+      "step": 5770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012470534245676116,
+      "loss": 2.8305,
+      "step": 5771
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012467850846337853,
+      "loss": 1.2681,
+      "step": 5772
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001246516725776102,
+      "loss": 1.3488,
+      "step": 5773
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012462483480151407,
+      "loss": 1.7631,
+      "step": 5774
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000124597995137148,
+      "loss": 1.536,
+      "step": 5775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012457115358657016,
+      "loss": 1.8622,
+      "step": 5776
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012454431015183881,
+      "loss": 1.4765,
+      "step": 5777
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012451746483501232,
+      "loss": 1.4369,
+      "step": 5778
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012449061763814923,
+      "loss": 1.0441,
+      "step": 5779
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012446376856330824,
+      "loss": 1.1584,
+      "step": 5780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012443691761254813,
+      "loss": 1.4294,
+      "step": 5781
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012441006478792794,
+      "loss": 1.2182,
+      "step": 5782
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012438321009150674,
+      "loss": 1.1286,
+      "step": 5783
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001243563535253438,
+      "loss": 1.4116,
+      "step": 5784
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012432949509149854,
+      "loss": 1.4522,
+      "step": 5785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012430263479203045,
+      "loss": 1.3492,
+      "step": 5786
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001242757726289993,
+      "loss": 1.2906,
+      "step": 5787
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012424890860446484,
+      "loss": 1.7964,
+      "step": 5788
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012422204272048712,
+      "loss": 0.9297,
+      "step": 5789
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012419517497912623,
+      "loss": 1.3332,
+      "step": 5790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012416830538244242,
+      "loss": 1.6281,
+      "step": 5791
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012414143393249606,
+      "loss": 1.2178,
+      "step": 5792
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012411456063134775,
+      "loss": 1.7866,
+      "step": 5793
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012408768548105814,
+      "loss": 1.6337,
+      "step": 5794
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012406080848368807,
+      "loss": 0.9382,
+      "step": 5795
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012403392964129852,
+      "loss": 1.577,
+      "step": 5796
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001240070489559506,
+      "loss": 3.0128,
+      "step": 5797
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012398016642970554,
+      "loss": 1.5423,
+      "step": 5798
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012395328206462472,
+      "loss": 1.7086,
+      "step": 5799
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012392639586276973,
+      "loss": 1.5687,
+      "step": 5800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012389950782620222,
+      "loss": 1.2637,
+      "step": 5801
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012387261795698397,
+      "loss": 1.4341,
+      "step": 5802
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000123845726257177,
+      "loss": 1.7861,
+      "step": 5803
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012381883272884333,
+      "loss": 1.1817,
+      "step": 5804
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012379193737404527,
+      "loss": 1.7876,
+      "step": 5805
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012376504019484514,
+      "loss": 1.3039,
+      "step": 5806
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001237381411933055,
+      "loss": 1.2675,
+      "step": 5807
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012371124037148897,
+      "loss": 1.3774,
+      "step": 5808
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001236843377314584,
+      "loss": 1.6299,
+      "step": 5809
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012365743327527669,
+      "loss": 1.113,
+      "step": 5810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012363052700500692,
+      "loss": 1.5461,
+      "step": 5811
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012360361892271225,
+      "loss": 1.1132,
+      "step": 5812
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001235767090304561,
+      "loss": 1.2376,
+      "step": 5813
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012354979733030197,
+      "loss": 1.554,
+      "step": 5814
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012352288382431342,
+      "loss": 1.4235,
+      "step": 5815
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001234959685145543,
+      "loss": 1.3893,
+      "step": 5816
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012346905140308848,
+      "loss": 1.4075,
+      "step": 5817
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012344213249197998,
+      "loss": 2.1303,
+      "step": 5818
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012341521178329303,
+      "loss": 1.4649,
+      "step": 5819
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001233882892790919,
+      "loss": 1.3285,
+      "step": 5820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012336136498144112,
+      "loss": 2.3988,
+      "step": 5821
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001233344388924052,
+      "loss": 1.3046,
+      "step": 5822
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012330751101404896,
+      "loss": 1.5053,
+      "step": 5823
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001232805813484372,
+      "loss": 1.3678,
+      "step": 5824
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012325364989763496,
+      "loss": 1.684,
+      "step": 5825
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012322671666370734,
+      "loss": 1.1968,
+      "step": 5826
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012319978164871976,
+      "loss": 1.8965,
+      "step": 5827
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012317284485473744,
+      "loss": 1.2056,
+      "step": 5828
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001231459062838261,
+      "loss": 0.9477,
+      "step": 5829
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012311896593805136,
+      "loss": 1.6182,
+      "step": 5830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000123092023819479,
+      "loss": 1.709,
+      "step": 5831
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012306507993017506,
+      "loss": 1.3639,
+      "step": 5832
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012303813427220564,
+      "loss": 1.5325,
+      "step": 5833
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012301118684763694,
+      "loss": 1.0042,
+      "step": 5834
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012298423765853532,
+      "loss": 1.6357,
+      "step": 5835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001229572867069673,
+      "loss": 0.782,
+      "step": 5836
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012293033399499953,
+      "loss": 1.3663,
+      "step": 5837
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001229033795246988,
+      "loss": 1.6087,
+      "step": 5838
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012287642329813196,
+      "loss": 1.0648,
+      "step": 5839
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001228494653173661,
+      "loss": 1.0702,
+      "step": 5840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001228225055844684,
+      "loss": 0.8906,
+      "step": 5841
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012279554410150615,
+      "loss": 1.4348,
+      "step": 5842
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012276858087054683,
+      "loss": 1.8311,
+      "step": 5843
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000122741615893658,
+      "loss": 1.3607,
+      "step": 5844
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012271464917290736,
+      "loss": 1.635,
+      "step": 5845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012268768071036282,
+      "loss": 1.4664,
+      "step": 5846
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012266071050809227,
+      "loss": 1.1119,
+      "step": 5847
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001226337385681639,
+      "loss": 1.5382,
+      "step": 5848
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012260676489264595,
+      "loss": 1.4637,
+      "step": 5849
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012257978948360676,
+      "loss": 1.2004,
+      "step": 5850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001225528123431149,
+      "loss": 1.844,
+      "step": 5851
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012252583347323898,
+      "loss": 1.1698,
+      "step": 5852
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001224988528760478,
+      "loss": 0.8172,
+      "step": 5853
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012247187055361027,
+      "loss": 1.2016,
+      "step": 5854
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012244488650799544,
+      "loss": 1.2964,
+      "step": 5855
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012241790074127247,
+      "loss": 1.4755,
+      "step": 5856
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012239091325551066,
+      "loss": 0.9767,
+      "step": 5857
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012236392405277948,
+      "loss": 1.4204,
+      "step": 5858
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012233693313514854,
+      "loss": 1.539,
+      "step": 5859
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012230994050468745,
+      "loss": 1.8595,
+      "step": 5860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001222829461634661,
+      "loss": 1.2966,
+      "step": 5861
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012225595011355447,
+      "loss": 1.6929,
+      "step": 5862
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012222895235702264,
+      "loss": 1.406,
+      "step": 5863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012220195289594084,
+      "loss": 2.214,
+      "step": 5864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012217495173237942,
+      "loss": 1.7366,
+      "step": 5865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001221479488684089,
+      "loss": 1.3219,
+      "step": 5866
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012212094430609987,
+      "loss": 1.2615,
+      "step": 5867
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001220939380475231,
+      "loss": 0.9697,
+      "step": 5868
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012206693009474943,
+      "loss": 1.7341,
+      "step": 5869
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012203992044984995,
+      "loss": 0.9756,
+      "step": 5870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012201290911489573,
+      "loss": 1.4717,
+      "step": 5871
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001219858960919581,
+      "loss": 0.8984,
+      "step": 5872
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012195888138310838,
+      "loss": 1.5507,
+      "step": 5873
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012193186499041818,
+      "loss": 1.5866,
+      "step": 5874
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012190484691595909,
+      "loss": 1.2209,
+      "step": 5875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012187782716180293,
+      "loss": 0.9842,
+      "step": 5876
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012185080573002165,
+      "loss": 1.305,
+      "step": 5877
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012182378262268725,
+      "loss": 0.9419,
+      "step": 5878
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001217967578418719,
+      "loss": 1.3356,
+      "step": 5879
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012176973138964793,
+      "loss": 1.1289,
+      "step": 5880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012174270326808774,
+      "loss": 1.6275,
+      "step": 5881
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012171567347926392,
+      "loss": 1.71,
+      "step": 5882
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012168864202524914,
+      "loss": 1.7496,
+      "step": 5883
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001216616089081162,
+      "loss": 1.4321,
+      "step": 5884
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012163457412993804,
+      "loss": 1.5967,
+      "step": 5885
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012160753769278777,
+      "loss": 1.4597,
+      "step": 5886
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012158049959873852,
+      "loss": 1.023,
+      "step": 5887
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012155345984986368,
+      "loss": 1.7963,
+      "step": 5888
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012152641844823665,
+      "loss": 0.8515,
+      "step": 5889
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000121499375395931,
+      "loss": 1.3192,
+      "step": 5890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012147233069502051,
+      "loss": 0.9671,
+      "step": 5891
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012144528434757892,
+      "loss": 1.2519,
+      "step": 5892
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001214182363556802,
+      "loss": 1.6221,
+      "step": 5893
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012139118672139847,
+      "loss": 1.2962,
+      "step": 5894
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012136413544680792,
+      "loss": 1.559,
+      "step": 5895
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012133708253398288,
+      "loss": 1.3286,
+      "step": 5896
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001213100279849978,
+      "loss": 1.5869,
+      "step": 5897
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012128297180192727,
+      "loss": 2.1598,
+      "step": 5898
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012125591398684601,
+      "loss": 1.3846,
+      "step": 5899
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012122885454182884,
+      "loss": 1.0869,
+      "step": 5900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012120179346895075,
+      "loss": 1.6066,
+      "step": 5901
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001211747307702868,
+      "loss": 0.9985,
+      "step": 5902
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012114766644791218,
+      "loss": 1.7714,
+      "step": 5903
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012112060050390227,
+      "loss": 1.6042,
+      "step": 5904
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012109353294033248,
+      "loss": 1.7273,
+      "step": 5905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012106646375927837,
+      "loss": 2.1506,
+      "step": 5906
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012103939296281575,
+      "loss": 1.7453,
+      "step": 5907
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012101232055302039,
+      "loss": 1.1577,
+      "step": 5908
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012098524653196823,
+      "loss": 0.9199,
+      "step": 5909
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012095817090173535,
+      "loss": 0.6375,
+      "step": 5910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012093109366439796,
+      "loss": 1.2039,
+      "step": 5911
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001209040148220324,
+      "loss": 1.2447,
+      "step": 5912
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012087693437671508,
+      "loss": 1.3094,
+      "step": 5913
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012084985233052259,
+      "loss": 1.1998,
+      "step": 5914
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012082276868553167,
+      "loss": 1.2867,
+      "step": 5915
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012079568344381905,
+      "loss": 1.3993,
+      "step": 5916
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012076859660746174,
+      "loss": 1.5912,
+      "step": 5917
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012074150817853671,
+      "loss": 2.1303,
+      "step": 5918
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012071441815912122,
+      "loss": 1.2111,
+      "step": 5919
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012068732655129258,
+      "loss": 1.1484,
+      "step": 5920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012066023335712818,
+      "loss": 1.2123,
+      "step": 5921
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012063313857870555,
+      "loss": 1.7312,
+      "step": 5922
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012060604221810242,
+      "loss": 1.3653,
+      "step": 5923
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012057894427739653,
+      "loss": 1.0968,
+      "step": 5924
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012055184475866579,
+      "loss": 1.6324,
+      "step": 5925
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012052474366398825,
+      "loss": 1.6475,
+      "step": 5926
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012049764099544208,
+      "loss": 1.6158,
+      "step": 5927
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012047053675510556,
+      "loss": 1.4122,
+      "step": 5928
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012044343094505701,
+      "loss": 1.4296,
+      "step": 5929
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012041632356737502,
+      "loss": 1.4921,
+      "step": 5930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001203892146241382,
+      "loss": 1.5071,
+      "step": 5931
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001203621041174253,
+      "loss": 1.3897,
+      "step": 5932
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012033499204931522,
+      "loss": 1.3471,
+      "step": 5933
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012030787842188693,
+      "loss": 1.3536,
+      "step": 5934
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012028076323721953,
+      "loss": 1.497,
+      "step": 5935
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012025364649739228,
+      "loss": 0.9319,
+      "step": 5936
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012022652820448454,
+      "loss": 1.6292,
+      "step": 5937
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012019940836057577,
+      "loss": 1.9421,
+      "step": 5938
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012017228696774556,
+      "loss": 1.0731,
+      "step": 5939
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001201451640280736,
+      "loss": 1.6246,
+      "step": 5940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012011803954363978,
+      "loss": 1.6483,
+      "step": 5941
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012009091351652395,
+      "loss": 2.093,
+      "step": 5942
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012006378594880625,
+      "loss": 1.0099,
+      "step": 5943
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012003665684256684,
+      "loss": 1.257,
+      "step": 5944
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012000952619988604,
+      "loss": 1.2493,
+      "step": 5945
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011998239402284426,
+      "loss": 1.6795,
+      "step": 5946
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011995526031352203,
+      "loss": 1.3938,
+      "step": 5947
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011992812507399995,
+      "loss": 1.2972,
+      "step": 5948
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001199009883063589,
+      "loss": 1.577,
+      "step": 5949
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001198738500126797,
+      "loss": 1.1597,
+      "step": 5950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011984671019504335,
+      "loss": 1.1649,
+      "step": 5951
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011981956885553103,
+      "loss": 2.0474,
+      "step": 5952
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011979242599622394,
+      "loss": 1.729,
+      "step": 5953
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011976528161920342,
+      "loss": 1.3402,
+      "step": 5954
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011973813572655095,
+      "loss": 1.4638,
+      "step": 5955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011971098832034813,
+      "loss": 1.2802,
+      "step": 5956
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011968383940267669,
+      "loss": 1.3591,
+      "step": 5957
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011965668897561841,
+      "loss": 1.6289,
+      "step": 5958
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011962953704125523,
+      "loss": 1.3533,
+      "step": 5959
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011960238360166924,
+      "loss": 1.6715,
+      "step": 5960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011957522865894255,
+      "loss": 1.2224,
+      "step": 5961
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001195480722151575,
+      "loss": 1.2031,
+      "step": 5962
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011952091427239642,
+      "loss": 1.256,
+      "step": 5963
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011949375483274187,
+      "loss": 1.3005,
+      "step": 5964
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001194665938982765,
+      "loss": 1.4676,
+      "step": 5965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011943943147108299,
+      "loss": 1.2097,
+      "step": 5966
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011941226755324424,
+      "loss": 1.6504,
+      "step": 5967
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001193851021468432,
+      "loss": 1.2525,
+      "step": 5968
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011935793525396297,
+      "loss": 1.7476,
+      "step": 5969
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011933076687668674,
+      "loss": 1.593,
+      "step": 5970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001193035970170978,
+      "loss": 1.1716,
+      "step": 5971
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011927642567727964,
+      "loss": 1.2735,
+      "step": 5972
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011924925285931574,
+      "loss": 1.6732,
+      "step": 5973
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011922207856528975,
+      "loss": 1.4019,
+      "step": 5974
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001191949027972855,
+      "loss": 2.0427,
+      "step": 5975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011916772555738683,
+      "loss": 1.5945,
+      "step": 5976
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011914054684767768,
+      "loss": 1.2259,
+      "step": 5977
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011911336667024226,
+      "loss": 1.065,
+      "step": 5978
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011908618502716471,
+      "loss": 1.4503,
+      "step": 5979
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011905900192052937,
+      "loss": 1.6129,
+      "step": 5980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011903181735242071,
+      "loss": 1.1867,
+      "step": 5981
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011900463132492325,
+      "loss": 1.5135,
+      "step": 5982
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001189774438401217,
+      "loss": 1.3718,
+      "step": 5983
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011895025490010081,
+      "loss": 1.6189,
+      "step": 5984
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011892306450694545,
+      "loss": 1.0208,
+      "step": 5985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011889587266274067,
+      "loss": 1.4431,
+      "step": 5986
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011886867936957153,
+      "loss": 1.6881,
+      "step": 5987
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011884148462952329,
+      "loss": 1.3681,
+      "step": 5988
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011881428844468127,
+      "loss": 1.2164,
+      "step": 5989
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011878709081713093,
+      "loss": 1.3723,
+      "step": 5990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011875989174895783,
+      "loss": 1.8817,
+      "step": 5991
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011873269124224759,
+      "loss": 1.1874,
+      "step": 5992
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011870548929908599,
+      "loss": 1.1367,
+      "step": 5993
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011867828592155899,
+      "loss": 1.2707,
+      "step": 5994
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011865108111175251,
+      "loss": 1.484,
+      "step": 5995
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011862387487175269,
+      "loss": 1.0533,
+      "step": 5996
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011859666720364575,
+      "loss": 1.12,
+      "step": 5997
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011856945810951801,
+      "loss": 1.7102,
+      "step": 5998
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011854224759145589,
+      "loss": 1.7534,
+      "step": 5999
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011851503565154594,
+      "loss": 1.9588,
+      "step": 6000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001184878222918748,
+      "loss": 1.2855,
+      "step": 6001
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001184606075145293,
+      "loss": 1.5902,
+      "step": 6002
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011843339132159623,
+      "loss": 1.247,
+      "step": 6003
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011840617371516261,
+      "loss": 1.0527,
+      "step": 6004
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001183789546973155,
+      "loss": 1.252,
+      "step": 6005
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011835173427014212,
+      "loss": 1.2209,
+      "step": 6006
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011832451243572977,
+      "loss": 1.4953,
+      "step": 6007
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011829728919616587,
+      "loss": 1.2385,
+      "step": 6008
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011827006455353791,
+      "loss": 1.3532,
+      "step": 6009
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011824283850993356,
+      "loss": 1.0308,
+      "step": 6010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001182156110674405,
+      "loss": 1.7099,
+      "step": 6011
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011818838222814661,
+      "loss": 1.2762,
+      "step": 6012
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011816115199413984,
+      "loss": 1.6481,
+      "step": 6013
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011813392036750823,
+      "loss": 1.4912,
+      "step": 6014
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011810668735033999,
+      "loss": 1.1446,
+      "step": 6015
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011807945294472329,
+      "loss": 1.5014,
+      "step": 6016
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011805221715274656,
+      "loss": 2.1983,
+      "step": 6017
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011802497997649833,
+      "loss": 0.9538,
+      "step": 6018
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011799774141806712,
+      "loss": 1.3737,
+      "step": 6019
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011797050147954165,
+      "loss": 1.1781,
+      "step": 6020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011794326016301074,
+      "loss": 1.4345,
+      "step": 6021
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011791601747056325,
+      "loss": 2.1494,
+      "step": 6022
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011788877340428821,
+      "loss": 1.2632,
+      "step": 6023
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011786152796627476,
+      "loss": 1.5986,
+      "step": 6024
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011783428115861208,
+      "loss": 1.6693,
+      "step": 6025
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011780703298338957,
+      "loss": 1.8662,
+      "step": 6026
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011777978344269658,
+      "loss": 1.4094,
+      "step": 6027
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011775253253862269,
+      "loss": 1.4007,
+      "step": 6028
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011772528027325751,
+      "loss": 1.0201,
+      "step": 6029
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001176980266486908,
+      "loss": 1.263,
+      "step": 6030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011767077166701244,
+      "loss": 1.4301,
+      "step": 6031
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011764351533031234,
+      "loss": 1.8894,
+      "step": 6032
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011761625764068058,
+      "loss": 1.1358,
+      "step": 6033
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011758899860020734,
+      "loss": 1.5765,
+      "step": 6034
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011756173821098284,
+      "loss": 1.5859,
+      "step": 6035
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011753447647509746,
+      "loss": 1.1119,
+      "step": 6036
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001175072133946417,
+      "loss": 1.5312,
+      "step": 6037
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011747994897170608,
+      "loss": 1.857,
+      "step": 6038
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011745268320838136,
+      "loss": 1.6883,
+      "step": 6039
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011742541610675826,
+      "loss": 1.2012,
+      "step": 6040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011739814766892767,
+      "loss": 1.7113,
+      "step": 6041
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011737087789698058,
+      "loss": 0.9811,
+      "step": 6042
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011734360679300808,
+      "loss": 1.4342,
+      "step": 6043
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011731633435910136,
+      "loss": 0.977,
+      "step": 6044
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011728906059735172,
+      "loss": 0.8305,
+      "step": 6045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011726178550985054,
+      "loss": 1.9439,
+      "step": 6046
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011723450909868934,
+      "loss": 1.6603,
+      "step": 6047
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011720723136595965,
+      "loss": 0.8775,
+      "step": 6048
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011717995231375323,
+      "loss": 1.3275,
+      "step": 6049
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011715267194416188,
+      "loss": 0.9859,
+      "step": 6050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011712539025927749,
+      "loss": 1.4971,
+      "step": 6051
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011709810726119206,
+      "loss": 1.3291,
+      "step": 6052
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011707082295199767,
+      "loss": 0.794,
+      "step": 6053
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011704353733378654,
+      "loss": 1.3222,
+      "step": 6054
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011701625040865096,
+      "loss": 1.3241,
+      "step": 6055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011698896217868337,
+      "loss": 0.9905,
+      "step": 6056
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011696167264597624,
+      "loss": 1.1823,
+      "step": 6057
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001169343818126222,
+      "loss": 1.9556,
+      "step": 6058
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011690708968071392,
+      "loss": 1.6562,
+      "step": 6059
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011687979625234421,
+      "loss": 1.0569,
+      "step": 6060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011685250152960598,
+      "loss": 0.6156,
+      "step": 6061
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011682520551459224,
+      "loss": 1.2501,
+      "step": 6062
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011679790820939609,
+      "loss": 1.2979,
+      "step": 6063
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001167706096161107,
+      "loss": 1.358,
+      "step": 6064
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011674330973682938,
+      "loss": 1.2506,
+      "step": 6065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011671600857364558,
+      "loss": 1.358,
+      "step": 6066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001166887061286527,
+      "loss": 1.1461,
+      "step": 6067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011666140240394438,
+      "loss": 0.8698,
+      "step": 6068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011663409740161436,
+      "loss": 1.1384,
+      "step": 6069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011660679112375632,
+      "loss": 1.2095,
+      "step": 6070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011657948357246427,
+      "loss": 0.9871,
+      "step": 6071
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011655217474983208,
+      "loss": 1.6121,
+      "step": 6072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011652486465795392,
+      "loss": 1.7094,
+      "step": 6073
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011649755329892391,
+      "loss": 1.7334,
+      "step": 6074
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011647024067483636,
+      "loss": 1.8703,
+      "step": 6075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011644292678778566,
+      "loss": 1.5361,
+      "step": 6076
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011641561163986623,
+      "loss": 1.4943,
+      "step": 6077
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011638829523317268,
+      "loss": 1.7942,
+      "step": 6078
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011636097756979962,
+      "loss": 1.8663,
+      "step": 6079
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011633365865184187,
+      "loss": 1.3803,
+      "step": 6080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001163063384813943,
+      "loss": 1.5169,
+      "step": 6081
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011627901706055177,
+      "loss": 1.1329,
+      "step": 6082
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011625169439140941,
+      "loss": 1.059,
+      "step": 6083
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011622437047606233,
+      "loss": 1.1065,
+      "step": 6084
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001161970453166058,
+      "loss": 1.1322,
+      "step": 6085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011616971891513508,
+      "loss": 1.7964,
+      "step": 6086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001161423912737457,
+      "loss": 1.3745,
+      "step": 6087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011611506239453311,
+      "loss": 1.4121,
+      "step": 6088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000116087732279593,
+      "loss": 1.3082,
+      "step": 6089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011606040093102099,
+      "loss": 1.1661,
+      "step": 6090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011603306835091296,
+      "loss": 1.4484,
+      "step": 6091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011600573454136481,
+      "loss": 1.3006,
+      "step": 6092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001159783995044725,
+      "loss": 1.5237,
+      "step": 6093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011595106324233216,
+      "loss": 1.5132,
+      "step": 6094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011592372575703998,
+      "loss": 1.2145,
+      "step": 6095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011589638705069221,
+      "loss": 1.0692,
+      "step": 6096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011586904712538524,
+      "loss": 1.6106,
+      "step": 6097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011584170598321551,
+      "loss": 1.3209,
+      "step": 6098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011581436362627963,
+      "loss": 1.3099,
+      "step": 6099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011578702005667424,
+      "loss": 2.1127,
+      "step": 6100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011575967527649603,
+      "loss": 1.6485,
+      "step": 6101
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001157323292878419,
+      "loss": 1.1895,
+      "step": 6102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001157049820928088,
+      "loss": 1.5882,
+      "step": 6103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011567763369349369,
+      "loss": 1.5076,
+      "step": 6104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001156502840919937,
+      "loss": 1.2104,
+      "step": 6105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011562293329040608,
+      "loss": 1.3516,
+      "step": 6106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001155955812908281,
+      "loss": 1.5121,
+      "step": 6107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011556822809535719,
+      "loss": 1.5203,
+      "step": 6108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011554087370609079,
+      "loss": 1.4798,
+      "step": 6109
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011551351812512648,
+      "loss": 0.6179,
+      "step": 6110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011548616135456191,
+      "loss": 1.6763,
+      "step": 6111
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011545880339649491,
+      "loss": 1.4426,
+      "step": 6112
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001154314442530233,
+      "loss": 1.4063,
+      "step": 6113
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000115404083926245,
+      "loss": 1.231,
+      "step": 6114
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011537672241825803,
+      "loss": 1.5192,
+      "step": 6115
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011534935973116057,
+      "loss": 1.3157,
+      "step": 6116
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011532199586705075,
+      "loss": 1.4778,
+      "step": 6117
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011529463082802694,
+      "loss": 1.4991,
+      "step": 6118
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011526726461618752,
+      "loss": 1.7058,
+      "step": 6119
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011523989723363094,
+      "loss": 1.2415,
+      "step": 6120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011521252868245583,
+      "loss": 1.7901,
+      "step": 6121
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011518515896476079,
+      "loss": 1.3007,
+      "step": 6122
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011515778808264462,
+      "loss": 1.252,
+      "step": 6123
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011513041603820611,
+      "loss": 1.3343,
+      "step": 6124
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011510304283354425,
+      "loss": 1.2655,
+      "step": 6125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011507566847075804,
+      "loss": 1.317,
+      "step": 6126
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011504829295194657,
+      "loss": 1.8729,
+      "step": 6127
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011502091627920903,
+      "loss": 1.7155,
+      "step": 6128
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011499353845464474,
+      "loss": 1.5544,
+      "step": 6129
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011496615948035303,
+      "loss": 1.4103,
+      "step": 6130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001149387793584334,
+      "loss": 1.9722,
+      "step": 6131
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011491139809098541,
+      "loss": 2.3994,
+      "step": 6132
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011488401568010863,
+      "loss": 0.9717,
+      "step": 6133
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011485663212790288,
+      "loss": 1.5527,
+      "step": 6134
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011482924743646783,
+      "loss": 1.6889,
+      "step": 6135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011480186160790354,
+      "loss": 1.1104,
+      "step": 6136
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011477447464430992,
+      "loss": 1.2645,
+      "step": 6137
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011474708654778705,
+      "loss": 1.0422,
+      "step": 6138
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011471969732043509,
+      "loss": 1.2471,
+      "step": 6139
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011469230696435429,
+      "loss": 1.1899,
+      "step": 6140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011466491548164497,
+      "loss": 1.2558,
+      "step": 6141
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011463752287440757,
+      "loss": 0.8306,
+      "step": 6142
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011461012914474258,
+      "loss": 1.5837,
+      "step": 6143
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011458273429475061,
+      "loss": 1.1724,
+      "step": 6144
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011455533832653236,
+      "loss": 0.8933,
+      "step": 6145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011452794124218856,
+      "loss": 1.6446,
+      "step": 6146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011450054304382005,
+      "loss": 1.5022,
+      "step": 6147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011447314373352778,
+      "loss": 0.9971,
+      "step": 6148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011444574331341278,
+      "loss": 2.345,
+      "step": 6149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011441834178557617,
+      "loss": 1.5505,
+      "step": 6150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011439093915211912,
+      "loss": 1.5084,
+      "step": 6151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011436353541514289,
+      "loss": 1.2836,
+      "step": 6152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011433613057674887,
+      "loss": 1.8871,
+      "step": 6153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011430872463903848,
+      "loss": 1.2674,
+      "step": 6154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011428131760411324,
+      "loss": 1.4478,
+      "step": 6155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011425390947407483,
+      "loss": 1.4347,
+      "step": 6156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001142265002510249,
+      "loss": 0.8451,
+      "step": 6157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011419908993706522,
+      "loss": 1.3706,
+      "step": 6158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011417167853429767,
+      "loss": 1.5028,
+      "step": 6159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001141442660448242,
+      "loss": 1.1064,
+      "step": 6160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011411685247074681,
+      "loss": 1.0719,
+      "step": 6161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011408943781416768,
+      "loss": 1.2953,
+      "step": 6162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011406202207718895,
+      "loss": 0.9952,
+      "step": 6163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011403460526191292,
+      "loss": 1.2862,
+      "step": 6164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011400718737044196,
+      "loss": 1.699,
+      "step": 6165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011397976840487849,
+      "loss": 1.9327,
+      "step": 6166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011395234836732507,
+      "loss": 1.7504,
+      "step": 6167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001139249272598843,
+      "loss": 1.1604,
+      "step": 6168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011389750508465888,
+      "loss": 1.0173,
+      "step": 6169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011387008184375156,
+      "loss": 1.6708,
+      "step": 6170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011384265753926524,
+      "loss": 1.1704,
+      "step": 6171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011381523217330281,
+      "loss": 1.0789,
+      "step": 6172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011378780574796726,
+      "loss": 1.4644,
+      "step": 6173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011376037826536179,
+      "loss": 1.4746,
+      "step": 6174
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 0.9425627589225769,
+      "eval_runtime": 20.537,
+      "eval_samples_per_second": 138.92,
+      "eval_steps_per_second": 138.92,
+      "step": 6174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011373294972758952,
+      "loss": 1.0934,
+      "step": 6175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001137055201367537,
+      "loss": 1.2517,
+      "step": 6176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011367808949495773,
+      "loss": 1.2539,
+      "step": 6177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011365065780430498,
+      "loss": 1.2764,
+      "step": 6178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011362322506689895,
+      "loss": 1.2389,
+      "step": 6179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011359579128484326,
+      "loss": 1.2699,
+      "step": 6180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011356835646024155,
+      "loss": 1.0052,
+      "step": 6181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011354092059519758,
+      "loss": 1.2663,
+      "step": 6182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011351348369181518,
+      "loss": 1.5325,
+      "step": 6183
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011348604575219822,
+      "loss": 1.1112,
+      "step": 6184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011345860677845071,
+      "loss": 0.9552,
+      "step": 6185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011343116677267669,
+      "loss": 1.4472,
+      "step": 6186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011340372573698034,
+      "loss": 1.5212,
+      "step": 6187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011337628367346587,
+      "loss": 1.8965,
+      "step": 6188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011334884058423754,
+      "loss": 1.9811,
+      "step": 6189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011332139647139976,
+      "loss": 1.1497,
+      "step": 6190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000113293951337057,
+      "loss": 1.9164,
+      "step": 6191
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011326650518331372,
+      "loss": 1.0371,
+      "step": 6192
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011323905801227466,
+      "loss": 0.9536,
+      "step": 6193
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001132116098260444,
+      "loss": 1.4934,
+      "step": 6194
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011318416062672779,
+      "loss": 1.3711,
+      "step": 6195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001131567104164296,
+      "loss": 1.2899,
+      "step": 6196
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011312925919725477,
+      "loss": 0.6573,
+      "step": 6197
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011310180697130834,
+      "loss": 1.4662,
+      "step": 6198
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011307435374069537,
+      "loss": 1.49,
+      "step": 6199
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011304689950752101,
+      "loss": 1.5869,
+      "step": 6200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011301944427389053,
+      "loss": 1.465,
+      "step": 6201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011299198804190918,
+      "loss": 1.349,
+      "step": 6202
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011296453081368237,
+      "loss": 1.6275,
+      "step": 6203
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011293707259131556,
+      "loss": 1.7654,
+      "step": 6204
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011290961337691428,
+      "loss": 2.0007,
+      "step": 6205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011288215317258419,
+      "loss": 1.53,
+      "step": 6206
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011285469198043095,
+      "loss": 1.5093,
+      "step": 6207
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011282722980256033,
+      "loss": 1.4234,
+      "step": 6208
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011279976664107814,
+      "loss": 1.6152,
+      "step": 6209
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011277230249809028,
+      "loss": 1.0552,
+      "step": 6210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011274483737570285,
+      "loss": 1.7496,
+      "step": 6211
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011271737127602185,
+      "loss": 1.5926,
+      "step": 6212
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001126899042011534,
+      "loss": 1.4109,
+      "step": 6213
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011266243615320375,
+      "loss": 1.4606,
+      "step": 6214
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011263496713427919,
+      "loss": 1.4097,
+      "step": 6215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011260749714648606,
+      "loss": 1.4017,
+      "step": 6216
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011258002619193082,
+      "loss": 1.3809,
+      "step": 6217
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011255255427272,
+      "loss": 1.4135,
+      "step": 6218
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011252508139096019,
+      "loss": 1.366,
+      "step": 6219
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011249760754875802,
+      "loss": 1.7227,
+      "step": 6220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011247013274822022,
+      "loss": 1.1948,
+      "step": 6221
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011244265699145364,
+      "loss": 1.6057,
+      "step": 6222
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011241518028056515,
+      "loss": 1.0488,
+      "step": 6223
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011238770261766172,
+      "loss": 1.5657,
+      "step": 6224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011236022400485033,
+      "loss": 1.5239,
+      "step": 6225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011233274444423811,
+      "loss": 1.2178,
+      "step": 6226
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011230526393793228,
+      "loss": 1.1218,
+      "step": 6227
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011227778248803999,
+      "loss": 1.0515,
+      "step": 6228
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011225030009666862,
+      "loss": 1.4108,
+      "step": 6229
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011222281676592559,
+      "loss": 1.0081,
+      "step": 6230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011219533249791831,
+      "loss": 1.3699,
+      "step": 6231
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011216784729475434,
+      "loss": 1.4078,
+      "step": 6232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011214036115854127,
+      "loss": 0.8874,
+      "step": 6233
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001121128740913868,
+      "loss": 1.2967,
+      "step": 6234
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011208538609539865,
+      "loss": 1.1173,
+      "step": 6235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011205789717268468,
+      "loss": 1.8864,
+      "step": 6236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011203040732535275,
+      "loss": 1.4822,
+      "step": 6237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011200291655551087,
+      "loss": 1.7267,
+      "step": 6238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011197542486526702,
+      "loss": 1.2812,
+      "step": 6239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011194793225672935,
+      "loss": 1.6385,
+      "step": 6240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011192043873200598,
+      "loss": 0.9716,
+      "step": 6241
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011189294429320519,
+      "loss": 1.0058,
+      "step": 6242
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011186544894243531,
+      "loss": 0.8191,
+      "step": 6243
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011183795268180471,
+      "loss": 1.1487,
+      "step": 6244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011181045551342182,
+      "loss": 1.0224,
+      "step": 6245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011178295743939522,
+      "loss": 1.8542,
+      "step": 6246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011175545846183344,
+      "loss": 0.8101,
+      "step": 6247
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011172795858284519,
+      "loss": 1.1194,
+      "step": 6248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001117004578045392,
+      "loss": 1.3588,
+      "step": 6249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011167295612902425,
+      "loss": 1.6142,
+      "step": 6250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011164545355840923,
+      "loss": 1.8228,
+      "step": 6251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011161795009480304,
+      "loss": 1.4083,
+      "step": 6252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011159044574031474,
+      "loss": 0.7852,
+      "step": 6253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011156294049705338,
+      "loss": 1.4486,
+      "step": 6254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011153543436712812,
+      "loss": 1.272,
+      "step": 6255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011150792735264817,
+      "loss": 1.4537,
+      "step": 6256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011148041945572276,
+      "loss": 0.9635,
+      "step": 6257
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011145291067846131,
+      "loss": 1.3041,
+      "step": 6258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011142540102297318,
+      "loss": 1.6369,
+      "step": 6259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011139789049136786,
+      "loss": 1.7597,
+      "step": 6260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011137037908575496,
+      "loss": 1.5953,
+      "step": 6261
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011134286680824402,
+      "loss": 2.0258,
+      "step": 6262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011131535366094477,
+      "loss": 1.0786,
+      "step": 6263
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011128783964596694,
+      "loss": 1.1634,
+      "step": 6264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011126032476542034,
+      "loss": 1.2209,
+      "step": 6265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011123280902141486,
+      "loss": 0.8987,
+      "step": 6266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011120529241606046,
+      "loss": 0.4649,
+      "step": 6267
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011117777495146713,
+      "loss": 1.2048,
+      "step": 6268
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000111150256629745,
+      "loss": 1.2916,
+      "step": 6269
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011112273745300416,
+      "loss": 1.1071,
+      "step": 6270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011109521742335487,
+      "loss": 0.9842,
+      "step": 6271
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011106769654290739,
+      "loss": 1.0746,
+      "step": 6272
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011104017481377203,
+      "loss": 1.311,
+      "step": 6273
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011101265223805923,
+      "loss": 1.1702,
+      "step": 6274
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011098512881787951,
+      "loss": 1.068,
+      "step": 6275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001109576045553433,
+      "loss": 1.2264,
+      "step": 6276
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011093007945256131,
+      "loss": 1.7367,
+      "step": 6277
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011090255351164409,
+      "loss": 1.6694,
+      "step": 6278
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011087502673470248,
+      "loss": 0.9498,
+      "step": 6279
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011084749912384724,
+      "loss": 1.0274,
+      "step": 6280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001108199706811892,
+      "loss": 1.4352,
+      "step": 6281
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011079244140883934,
+      "loss": 1.0523,
+      "step": 6282
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001107649113089086,
+      "loss": 1.6462,
+      "step": 6283
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011073738038350802,
+      "loss": 1.6618,
+      "step": 6284
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011070984863474876,
+      "loss": 1.4328,
+      "step": 6285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011068231606474196,
+      "loss": 2.1458,
+      "step": 6286
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011065478267559888,
+      "loss": 1.4166,
+      "step": 6287
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011062724846943082,
+      "loss": 1.4203,
+      "step": 6288
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011059971344834912,
+      "loss": 0.9954,
+      "step": 6289
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011057217761446523,
+      "loss": 1.0418,
+      "step": 6290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011054464096989064,
+      "loss": 1.1926,
+      "step": 6291
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001105171035167369,
+      "loss": 1.914,
+      "step": 6292
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011048956525711564,
+      "loss": 1.0738,
+      "step": 6293
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001104620261931385,
+      "loss": 1.4987,
+      "step": 6294
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011043448632691722,
+      "loss": 1.2381,
+      "step": 6295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011040694566056366,
+      "loss": 0.935,
+      "step": 6296
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011037940419618957,
+      "loss": 1.6488,
+      "step": 6297
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011035186193590698,
+      "loss": 0.9624,
+      "step": 6298
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011032431888182784,
+      "loss": 1.5313,
+      "step": 6299
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011029677503606415,
+      "loss": 1.864,
+      "step": 6300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011026923040072808,
+      "loss": 1.4087,
+      "step": 6301
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011024168497793172,
+      "loss": 1.3082,
+      "step": 6302
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011021413876978734,
+      "loss": 1.4695,
+      "step": 6303
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011018659177840725,
+      "loss": 1.6594,
+      "step": 6304
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011015904400590375,
+      "loss": 1.0737,
+      "step": 6305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011013149545438927,
+      "loss": 1.7772,
+      "step": 6306
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011010394612597626,
+      "loss": 1.5375,
+      "step": 6307
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011007639602277724,
+      "loss": 1.3986,
+      "step": 6308
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011004884514690481,
+      "loss": 2.0655,
+      "step": 6309
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011002129350047162,
+      "loss": 1.4049,
+      "step": 6310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010999374108559036,
+      "loss": 0.9345,
+      "step": 6311
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010996618790437378,
+      "loss": 1.7133,
+      "step": 6312
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010993863395893471,
+      "loss": 1.5341,
+      "step": 6313
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010991107925138606,
+      "loss": 1.7819,
+      "step": 6314
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010988352378384073,
+      "loss": 1.2291,
+      "step": 6315
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010985596755841167,
+      "loss": 0.9248,
+      "step": 6316
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010982841057721203,
+      "loss": 0.9357,
+      "step": 6317
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010980085284235487,
+      "loss": 0.9665,
+      "step": 6318
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010977329435595336,
+      "loss": 1.0137,
+      "step": 6319
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010974573512012077,
+      "loss": 1.4815,
+      "step": 6320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001097181751369703,
+      "loss": 1.2363,
+      "step": 6321
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010969061440861535,
+      "loss": 1.5455,
+      "step": 6322
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001096630529371693,
+      "loss": 1.2362,
+      "step": 6323
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010963549072474563,
+      "loss": 1.1662,
+      "step": 6324
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010960792777345785,
+      "loss": 1.1523,
+      "step": 6325
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010958036408541949,
+      "loss": 1.488,
+      "step": 6326
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010955279966274421,
+      "loss": 1.8096,
+      "step": 6327
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010952523450754566,
+      "loss": 0.7936,
+      "step": 6328
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010949766862193763,
+      "loss": 1.6721,
+      "step": 6329
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010947010200803389,
+      "loss": 1.5761,
+      "step": 6330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010944253466794825,
+      "loss": 1.3039,
+      "step": 6331
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010941496660379467,
+      "loss": 1.4263,
+      "step": 6332
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010938739781768709,
+      "loss": 1.3393,
+      "step": 6333
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010935982831173955,
+      "loss": 1.281,
+      "step": 6334
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010933225808806604,
+      "loss": 1.2759,
+      "step": 6335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010930468714878081,
+      "loss": 1.0989,
+      "step": 6336
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010927711549599796,
+      "loss": 1.303,
+      "step": 6337
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010924954313183173,
+      "loss": 1.5161,
+      "step": 6338
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010922197005839643,
+      "loss": 1.0239,
+      "step": 6339
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010919439627780641,
+      "loss": 1.5224,
+      "step": 6340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010916682179217605,
+      "loss": 1.4023,
+      "step": 6341
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010913924660361981,
+      "loss": 1.0187,
+      "step": 6342
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010911167071425224,
+      "loss": 1.5036,
+      "step": 6343
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010908409412618783,
+      "loss": 1.6594,
+      "step": 6344
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010905651684154121,
+      "loss": 1.1832,
+      "step": 6345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010902893886242709,
+      "loss": 1.6606,
+      "step": 6346
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010900136019096017,
+      "loss": 1.6126,
+      "step": 6347
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001089737808292552,
+      "loss": 1.3866,
+      "step": 6348
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010894620077942707,
+      "loss": 1.698,
+      "step": 6349
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010891862004359058,
+      "loss": 1.9766,
+      "step": 6350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010889103862386077,
+      "loss": 1.0517,
+      "step": 6351
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010886345652235248,
+      "loss": 0.9329,
+      "step": 6352
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010883587374118083,
+      "loss": 1.3808,
+      "step": 6353
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010880829028246096,
+      "loss": 0.9102,
+      "step": 6354
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001087807061483079,
+      "loss": 1.5393,
+      "step": 6355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010875312134083692,
+      "loss": 1.6472,
+      "step": 6356
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010872553586216328,
+      "loss": 0.9653,
+      "step": 6357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010869794971440222,
+      "loss": 1.6726,
+      "step": 6358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001086703628996691,
+      "loss": 1.9936,
+      "step": 6359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010864277542007932,
+      "loss": 1.0331,
+      "step": 6360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010861518727774835,
+      "loss": 1.3658,
+      "step": 6361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010858759847479169,
+      "loss": 1.2799,
+      "step": 6362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010856000901332488,
+      "loss": 1.1205,
+      "step": 6363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010853241889546353,
+      "loss": 1.2277,
+      "step": 6364
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010850482812332326,
+      "loss": 1.339,
+      "step": 6365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010847723669901981,
+      "loss": 1.4535,
+      "step": 6366
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010844964462466893,
+      "loss": 1.0217,
+      "step": 6367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010842205190238641,
+      "loss": 1.1435,
+      "step": 6368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010839445853428811,
+      "loss": 2.0635,
+      "step": 6369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010836686452248992,
+      "loss": 1.2645,
+      "step": 6370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010833926986910781,
+      "loss": 1.5144,
+      "step": 6371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010831167457625773,
+      "loss": 1.1703,
+      "step": 6372
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010828407864605578,
+      "loss": 1.5588,
+      "step": 6373
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010825648208061808,
+      "loss": 1.8921,
+      "step": 6374
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010822888488206073,
+      "loss": 1.6848,
+      "step": 6375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010820128705249992,
+      "loss": 1.2585,
+      "step": 6376
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001081736885940519,
+      "loss": 1.7098,
+      "step": 6377
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010814608950883298,
+      "loss": 1.4799,
+      "step": 6378
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010811848979895951,
+      "loss": 1.8084,
+      "step": 6379
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010809088946654784,
+      "loss": 1.2927,
+      "step": 6380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010806328851371445,
+      "loss": 1.4956,
+      "step": 6381
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010803568694257577,
+      "loss": 1.5337,
+      "step": 6382
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010800808475524839,
+      "loss": 0.8876,
+      "step": 6383
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010798048195384884,
+      "loss": 1.2518,
+      "step": 6384
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010795287854049376,
+      "loss": 0.9479,
+      "step": 6385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010792527451729984,
+      "loss": 1.1101,
+      "step": 6386
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010789766988638376,
+      "loss": 1.614,
+      "step": 6387
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010787006464986234,
+      "loss": 1.4625,
+      "step": 6388
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010784245880985234,
+      "loss": 1.3632,
+      "step": 6389
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010781485236847063,
+      "loss": 1.2164,
+      "step": 6390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010778724532783413,
+      "loss": 0.9787,
+      "step": 6391
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010775963769005977,
+      "loss": 1.8627,
+      "step": 6392
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010773202945726457,
+      "loss": 1.6602,
+      "step": 6393
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010770442063156556,
+      "loss": 1.6487,
+      "step": 6394
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010767681121507981,
+      "loss": 1.1175,
+      "step": 6395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010764920120992448,
+      "loss": 1.3983,
+      "step": 6396
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010762159061821673,
+      "loss": 1.2949,
+      "step": 6397
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010759397944207379,
+      "loss": 2.6638,
+      "step": 6398
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010756636768361295,
+      "loss": 1.4823,
+      "step": 6399
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010753875534495146,
+      "loss": 1.6104,
+      "step": 6400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010751114242820673,
+      "loss": 0.6485,
+      "step": 6401
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010748352893549611,
+      "loss": 1.4947,
+      "step": 6402
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010745591486893711,
+      "loss": 1.3139,
+      "step": 6403
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001074283002306472,
+      "loss": 1.4512,
+      "step": 6404
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010740068502274387,
+      "loss": 1.7889,
+      "step": 6405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010737306924734473,
+      "loss": 1.4019,
+      "step": 6406
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010734545290656743,
+      "loss": 1.3115,
+      "step": 6407
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010731783600252956,
+      "loss": 1.4821,
+      "step": 6408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010729021853734886,
+      "loss": 1.3491,
+      "step": 6409
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001072626005131431,
+      "loss": 1.415,
+      "step": 6410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010723498193203005,
+      "loss": 1.5771,
+      "step": 6411
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010720736279612758,
+      "loss": 1.0807,
+      "step": 6412
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001071797431075535,
+      "loss": 1.3538,
+      "step": 6413
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010715212286842578,
+      "loss": 1.0615,
+      "step": 6414
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010712450208086237,
+      "loss": 1.5713,
+      "step": 6415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010709688074698128,
+      "loss": 1.6868,
+      "step": 6416
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010706925886890053,
+      "loss": 1.2265,
+      "step": 6417
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010704163644873826,
+      "loss": 1.4595,
+      "step": 6418
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010701401348861255,
+      "loss": 1.6499,
+      "step": 6419
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010698638999064161,
+      "loss": 1.4723,
+      "step": 6420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010695876595694358,
+      "loss": 1.5342,
+      "step": 6421
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001069311413896368,
+      "loss": 1.1054,
+      "step": 6422
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010690351629083952,
+      "loss": 1.5846,
+      "step": 6423
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010687589066267009,
+      "loss": 1.7317,
+      "step": 6424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010684826450724688,
+      "loss": 1.1208,
+      "step": 6425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001068206378266883,
+      "loss": 0.8107,
+      "step": 6426
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001067930106231128,
+      "loss": 1.7144,
+      "step": 6427
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010676538289863886,
+      "loss": 1.1365,
+      "step": 6428
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010673775465538508,
+      "loss": 1.0677,
+      "step": 6429
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010671012589546998,
+      "loss": 1.1857,
+      "step": 6430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010668249662101222,
+      "loss": 1.1965,
+      "step": 6431
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001066548668341304,
+      "loss": 1.6839,
+      "step": 6432
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010662723653694324,
+      "loss": 0.9154,
+      "step": 6433
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010659960573156946,
+      "loss": 0.9777,
+      "step": 6434
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010657197442012787,
+      "loss": 1.2684,
+      "step": 6435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010654434260473727,
+      "loss": 1.3432,
+      "step": 6436
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010651671028751649,
+      "loss": 1.7699,
+      "step": 6437
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010648907747058441,
+      "loss": 1.4221,
+      "step": 6438
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010646144415606,
+      "loss": 2.1799,
+      "step": 6439
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010643381034606217,
+      "loss": 1.7864,
+      "step": 6440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010640617604270997,
+      "loss": 1.4722,
+      "step": 6441
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010637854124812242,
+      "loss": 1.746,
+      "step": 6442
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010635090596441857,
+      "loss": 1.2154,
+      "step": 6443
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001063232701937176,
+      "loss": 1.5771,
+      "step": 6444
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010629563393813862,
+      "loss": 1.1796,
+      "step": 6445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001062679971998008,
+      "loss": 0.9602,
+      "step": 6446
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010624035998082343,
+      "loss": 1.3943,
+      "step": 6447
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010621272228332571,
+      "loss": 1.402,
+      "step": 6448
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000106185084109427,
+      "loss": 1.0872,
+      "step": 6449
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010615744546124658,
+      "loss": 1.78,
+      "step": 6450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010612980634090386,
+      "loss": 1.2281,
+      "step": 6451
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010610216675051823,
+      "loss": 1.2093,
+      "step": 6452
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010607452669220916,
+      "loss": 1.2385,
+      "step": 6453
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010604688616809613,
+      "loss": 1.4832,
+      "step": 6454
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010601924518029865,
+      "loss": 1.1021,
+      "step": 6455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010599160373093624,
+      "loss": 0.7318,
+      "step": 6456
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010596396182212855,
+      "loss": 1.6607,
+      "step": 6457
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010593631945599516,
+      "loss": 1.1827,
+      "step": 6458
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010590867663465568,
+      "loss": 1.4045,
+      "step": 6459
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010588103336022994,
+      "loss": 1.2617,
+      "step": 6460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010585338963483757,
+      "loss": 0.9837,
+      "step": 6461
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010582574546059836,
+      "loss": 1.6968,
+      "step": 6462
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010579810083963211,
+      "loss": 1.257,
+      "step": 6463
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010577045577405864,
+      "loss": 1.5659,
+      "step": 6464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010574281026599779,
+      "loss": 1.3414,
+      "step": 6465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010571516431756952,
+      "loss": 1.9352,
+      "step": 6466
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010568751793089374,
+      "loss": 1.2346,
+      "step": 6467
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010565987110809042,
+      "loss": 1.4268,
+      "step": 6468
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010563222385127951,
+      "loss": 1.5735,
+      "step": 6469
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001056045761625811,
+      "loss": 1.1906,
+      "step": 6470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010557692804411524,
+      "loss": 1.0948,
+      "step": 6471
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010554927949800204,
+      "loss": 1.3198,
+      "step": 6472
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010552163052636164,
+      "loss": 1.3341,
+      "step": 6473
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010549398113131417,
+      "loss": 1.3848,
+      "step": 6474
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010546633131497986,
+      "loss": 0.861,
+      "step": 6475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010543868107947892,
+      "loss": 1.5125,
+      "step": 6476
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010541103042693159,
+      "loss": 0.8249,
+      "step": 6477
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010538337935945824,
+      "loss": 1.7065,
+      "step": 6478
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010535572787917913,
+      "loss": 1.6393,
+      "step": 6479
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010532807598821464,
+      "loss": 1.0609,
+      "step": 6480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010530042368868515,
+      "loss": 1.2086,
+      "step": 6481
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010527277098271111,
+      "loss": 1.3081,
+      "step": 6482
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010524511787241293,
+      "loss": 1.639,
+      "step": 6483
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010521746435991111,
+      "loss": 1.4143,
+      "step": 6484
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010518981044732616,
+      "loss": 1.1444,
+      "step": 6485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010516215613677867,
+      "loss": 1.3502,
+      "step": 6486
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010513450143038914,
+      "loss": 1.7109,
+      "step": 6487
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001051068463302782,
+      "loss": 0.8556,
+      "step": 6488
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010507919083856652,
+      "loss": 1.6292,
+      "step": 6489
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010505153495737472,
+      "loss": 1.8789,
+      "step": 6490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010502387868882353,
+      "loss": 0.896,
+      "step": 6491
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010499622203503368,
+      "loss": 1.5217,
+      "step": 6492
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010496856499812586,
+      "loss": 1.3624,
+      "step": 6493
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010494090758022093,
+      "loss": 1.9848,
+      "step": 6494
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010491324978343966,
+      "loss": 1.4947,
+      "step": 6495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010488559160990288,
+      "loss": 1.3809,
+      "step": 6496
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010485793306173153,
+      "loss": 1.9313,
+      "step": 6497
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010483027414104642,
+      "loss": 1.9285,
+      "step": 6498
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010480261484996854,
+      "loss": 1.5259,
+      "step": 6499
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010477495519061884,
+      "loss": 1.3133,
+      "step": 6500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001047472951651183,
+      "loss": 1.5473,
+      "step": 6501
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001047196347755879,
+      "loss": 0.77,
+      "step": 6502
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001046919740241487,
+      "loss": 2.014,
+      "step": 6503
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001046643129129218,
+      "loss": 1.5108,
+      "step": 6504
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010463665144402827,
+      "loss": 1.7573,
+      "step": 6505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010460898961958922,
+      "loss": 1.2672,
+      "step": 6506
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010458132744172584,
+      "loss": 1.3024,
+      "step": 6507
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010455366491255927,
+      "loss": 1.3044,
+      "step": 6508
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010452600203421073,
+      "loss": 1.0173,
+      "step": 6509
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010449833880880146,
+      "loss": 1.2155,
+      "step": 6510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010447067523845271,
+      "loss": 1.3504,
+      "step": 6511
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010444301132528574,
+      "loss": 1.3534,
+      "step": 6512
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010441534707142193,
+      "loss": 1.2835,
+      "step": 6513
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010438768247898254,
+      "loss": 1.1732,
+      "step": 6514
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010436001755008893,
+      "loss": 1.206,
+      "step": 6515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010433235228686258,
+      "loss": 1.1682,
+      "step": 6516
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010430468669142483,
+      "loss": 1.5482,
+      "step": 6517
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010427702076589716,
+      "loss": 2.1264,
+      "step": 6518
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010424935451240101,
+      "loss": 1.1281,
+      "step": 6519
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010422168793305784,
+      "loss": 1.2656,
+      "step": 6520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010419402102998923,
+      "loss": 1.088,
+      "step": 6521
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010416635380531667,
+      "loss": 2.0133,
+      "step": 6522
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010413868626116179,
+      "loss": 1.3094,
+      "step": 6523
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001041110183996461,
+      "loss": 1.5187,
+      "step": 6524
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010408335022289124,
+      "loss": 1.6641,
+      "step": 6525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010405568173301889,
+      "loss": 0.8506,
+      "step": 6526
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010402801293215063,
+      "loss": 1.3169,
+      "step": 6527
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010400034382240821,
+      "loss": 1.5645,
+      "step": 6528
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010397267440591338,
+      "loss": 0.9034,
+      "step": 6529
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010394500468478776,
+      "loss": 1.0443,
+      "step": 6530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010391733466115322,
+      "loss": 1.0626,
+      "step": 6531
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010388966433713143,
+      "loss": 1.9257,
+      "step": 6532
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010386199371484426,
+      "loss": 1.3157,
+      "step": 6533
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010383432279641352,
+      "loss": 1.0708,
+      "step": 6534
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010380665158396105,
+      "loss": 1.3446,
+      "step": 6535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010377898007960873,
+      "loss": 1.6318,
+      "step": 6536
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001037513082854785,
+      "loss": 1.2668,
+      "step": 6537
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010372363620369218,
+      "loss": 0.849,
+      "step": 6538
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010369596383637179,
+      "loss": 0.7731,
+      "step": 6539
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010366829118563924,
+      "loss": 1.5191,
+      "step": 6540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001036406182536165,
+      "loss": 1.2402,
+      "step": 6541
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010361294504242565,
+      "loss": 1.2071,
+      "step": 6542
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010358527155418865,
+      "loss": 1.2521,
+      "step": 6543
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010355759779102753,
+      "loss": 1.7725,
+      "step": 6544
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001035299237550644,
+      "loss": 1.4874,
+      "step": 6545
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010350224944842135,
+      "loss": 1.0811,
+      "step": 6546
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010347457487322047,
+      "loss": 1.6723,
+      "step": 6547
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010344690003158389,
+      "loss": 1.5819,
+      "step": 6548
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010341922492563374,
+      "loss": 1.1369,
+      "step": 6549
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010339154955749222,
+      "loss": 1.6425,
+      "step": 6550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001033638739292815,
+      "loss": 1.3257,
+      "step": 6551
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010333619804312377,
+      "loss": 1.3539,
+      "step": 6552
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010330852190114133,
+      "loss": 1.383,
+      "step": 6553
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010328084550545638,
+      "loss": 1.3638,
+      "step": 6554
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010325316885819118,
+      "loss": 1.432,
+      "step": 6555
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010322549196146805,
+      "loss": 0.8845,
+      "step": 6556
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010319781481740925,
+      "loss": 1.2217,
+      "step": 6557
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010317013742813715,
+      "loss": 1.5559,
+      "step": 6558
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010314245979577406,
+      "loss": 1.6272,
+      "step": 6559
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010311478192244235,
+      "loss": 1.3478,
+      "step": 6560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010308710381026445,
+      "loss": 1.3178,
+      "step": 6561
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010305942546136272,
+      "loss": 1.5321,
+      "step": 6562
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010303174687785956,
+      "loss": 1.1058,
+      "step": 6563
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010300406806187741,
+      "loss": 1.3017,
+      "step": 6564
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010297638901553875,
+      "loss": 1.5741,
+      "step": 6565
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010294870974096606,
+      "loss": 1.4714,
+      "step": 6566
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010292103024028181,
+      "loss": 1.2548,
+      "step": 6567
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010289335051560851,
+      "loss": 1.2124,
+      "step": 6568
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010286567056906867,
+      "loss": 1.2931,
+      "step": 6569
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010283799040278481,
+      "loss": 1.8483,
+      "step": 6570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010281031001887955,
+      "loss": 1.5508,
+      "step": 6571
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010278262941947541,
+      "loss": 2.4843,
+      "step": 6572
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010275494860669501,
+      "loss": 1.5284,
+      "step": 6573
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010272726758266098,
+      "loss": 1.3336,
+      "step": 6574
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010269958634949588,
+      "loss": 2.0465,
+      "step": 6575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001026719049093224,
+      "loss": 1.3644,
+      "step": 6576
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010264422326426314,
+      "loss": 0.9852,
+      "step": 6577
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010261654141644084,
+      "loss": 1.5877,
+      "step": 6578
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010258885936797815,
+      "loss": 1.4519,
+      "step": 6579
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010256117712099777,
+      "loss": 1.4023,
+      "step": 6580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010253349467762241,
+      "loss": 1.6115,
+      "step": 6581
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010250581203997485,
+      "loss": 1.6928,
+      "step": 6582
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010247812921017776,
+      "loss": 0.8592,
+      "step": 6583
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010245044619035398,
+      "loss": 2.1354,
+      "step": 6584
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001024227629826262,
+      "loss": 1.4879,
+      "step": 6585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010239507958911729,
+      "loss": 1.1202,
+      "step": 6586
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010236739601195004,
+      "loss": 1.7622,
+      "step": 6587
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010233971225324725,
+      "loss": 1.6204,
+      "step": 6588
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010231202831513174,
+      "loss": 1.0554,
+      "step": 6589
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010228434419972636,
+      "loss": 1.5625,
+      "step": 6590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000102256659909154,
+      "loss": 0.963,
+      "step": 6591
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010222897544553753,
+      "loss": 1.1962,
+      "step": 6592
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001022012908109998,
+      "loss": 0.5671,
+      "step": 6593
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010217360600766374,
+      "loss": 1.749,
+      "step": 6594
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010214592103765225,
+      "loss": 1.5147,
+      "step": 6595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010211823590308827,
+      "loss": 1.4467,
+      "step": 6596
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010209055060609471,
+      "loss": 1.35,
+      "step": 6597
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010206286514879459,
+      "loss": 1.1164,
+      "step": 6598
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010203517953331079,
+      "loss": 1.8627,
+      "step": 6599
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010200749376176635,
+      "loss": 0.8367,
+      "step": 6600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010197980783628418,
+      "loss": 1.2843,
+      "step": 6601
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010195212175898734,
+      "loss": 1.866,
+      "step": 6602
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010192443553199883,
+      "loss": 1.0785,
+      "step": 6603
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010189674915744168,
+      "loss": 1.4728,
+      "step": 6604
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010186906263743889,
+      "loss": 1.0861,
+      "step": 6605
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010184137597411357,
+      "loss": 0.9151,
+      "step": 6606
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010181368916958869,
+      "loss": 1.7906,
+      "step": 6607
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010178600222598736,
+      "loss": 1.5821,
+      "step": 6608
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010175831514543266,
+      "loss": 1.8703,
+      "step": 6609
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010173062793004768,
+      "loss": 1.4861,
+      "step": 6610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010170294058195553,
+      "loss": 1.03,
+      "step": 6611
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010167525310327928,
+      "loss": 0.7709,
+      "step": 6612
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010164756549614204,
+      "loss": 1.6259,
+      "step": 6613
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000101619877762667,
+      "loss": 0.956,
+      "step": 6614
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010159218990497723,
+      "loss": 1.1592,
+      "step": 6615
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010156450192519595,
+      "loss": 1.1214,
+      "step": 6616
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010153681382544627,
+      "loss": 1.1397,
+      "step": 6617
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010150912560785133,
+      "loss": 1.4398,
+      "step": 6618
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010148143727453438,
+      "loss": 1.2134,
+      "step": 6619
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010145374882761852,
+      "loss": 1.2163,
+      "step": 6620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000101426060269227,
+      "loss": 1.281,
+      "step": 6621
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010139837160148302,
+      "loss": 1.1603,
+      "step": 6622
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010137068282650974,
+      "loss": 1.3957,
+      "step": 6623
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010134299394643046,
+      "loss": 1.1131,
+      "step": 6624
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001013153049633683,
+      "loss": 1.2828,
+      "step": 6625
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010128761587944657,
+      "loss": 1.2791,
+      "step": 6626
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010125992669678849,
+      "loss": 1.3672,
+      "step": 6627
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010123223741751731,
+      "loss": 1.606,
+      "step": 6628
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001012045480437563,
+      "loss": 1.1393,
+      "step": 6629
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010117685857762869,
+      "loss": 0.9717,
+      "step": 6630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010114916902125777,
+      "loss": 1.626,
+      "step": 6631
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010112147937676683,
+      "loss": 1.8726,
+      "step": 6632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010109378964627914,
+      "loss": 1.4906,
+      "step": 6633
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010106609983191797,
+      "loss": 1.4484,
+      "step": 6634
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010103840993580667,
+      "loss": 1.3448,
+      "step": 6635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010101071996006847,
+      "loss": 1.8619,
+      "step": 6636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010098302990682678,
+      "loss": 1.4967,
+      "step": 6637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010095533977820481,
+      "loss": 1.0941,
+      "step": 6638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010092764957632595,
+      "loss": 1.4406,
+      "step": 6639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010089995930331351,
+      "loss": 0.9177,
+      "step": 6640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001008722689612908,
+      "loss": 0.7696,
+      "step": 6641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010084457855238119,
+      "loss": 0.6855,
+      "step": 6642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010081688807870802,
+      "loss": 1.0674,
+      "step": 6643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010078919754239459,
+      "loss": 1.9077,
+      "step": 6644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001007615069455643,
+      "loss": 1.3606,
+      "step": 6645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010073381629034051,
+      "loss": 1.3614,
+      "step": 6646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010070612557884655,
+      "loss": 1.7308,
+      "step": 6647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010067843481320583,
+      "loss": 0.8307,
+      "step": 6648
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010065074399554168,
+      "loss": 0.962,
+      "step": 6649
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010062305312797748,
+      "loss": 1.1111,
+      "step": 6650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010059536221263662,
+      "loss": 1.0442,
+      "step": 6651
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010056767125164249,
+      "loss": 0.9897,
+      "step": 6652
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010053998024711846,
+      "loss": 0.812,
+      "step": 6653
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010051228920118789,
+      "loss": 1.2419,
+      "step": 6654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010048459811597424,
+      "loss": 1.1912,
+      "step": 6655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010045690699360086,
+      "loss": 1.3957,
+      "step": 6656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010042921583619116,
+      "loss": 1.2985,
+      "step": 6657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010040152464586851,
+      "loss": 0.9116,
+      "step": 6658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010037383342475634,
+      "loss": 0.7417,
+      "step": 6659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010034614217497807,
+      "loss": 1.2337,
+      "step": 6660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001003184508986571,
+      "loss": 1.5412,
+      "step": 6661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010029075959791681,
+      "loss": 1.2681,
+      "step": 6662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010026306827488063,
+      "loss": 1.3566,
+      "step": 6663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010023537693167197,
+      "loss": 1.195,
+      "step": 6664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010020768557041424,
+      "loss": 1.3611,
+      "step": 6665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010017999419323092,
+      "loss": 1.4549,
+      "step": 6666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010015230280224533,
+      "loss": 1.4939,
+      "step": 6667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010012461139958091,
+      "loss": 0.9928,
+      "step": 6668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010009691998736113,
+      "loss": 1.5681,
+      "step": 6669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010006922856770935,
+      "loss": 1.1108,
+      "step": 6670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010004153714274902,
+      "loss": 1.0545,
+      "step": 6671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010001384571460359,
+      "loss": 1.088,
+      "step": 6672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.998615428539644e-05,
+      "loss": 1.3788,
+      "step": 6673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.9958462857251e-05,
+      "loss": 1.8374,
+      "step": 6674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.993077143229067e-05,
+      "loss": 1.2467,
+      "step": 6675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.990308001263891e-05,
+      "loss": 0.7399,
+      "step": 6676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.987538860041908e-05,
+      "loss": 1.403,
+      "step": 6677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.984769719775474e-05,
+      "loss": 1.7268,
+      "step": 6678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.982000580676913e-05,
+      "loss": 0.5528,
+      "step": 6679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.979231442958577e-05,
+      "loss": 1.577,
+      "step": 6680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.976462306832804e-05,
+      "loss": 1.0817,
+      "step": 6681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.973693172511939e-05,
+      "loss": 1.7055,
+      "step": 6682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.970924040208321e-05,
+      "loss": 1.2695,
+      "step": 6683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.968154910134294e-05,
+      "loss": 0.9219,
+      "step": 6684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.965385782502196e-05,
+      "loss": 1.782,
+      "step": 6685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.962616657524367e-05,
+      "loss": 1.372,
+      "step": 6686
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.959847535413151e-05,
+      "loss": 1.104,
+      "step": 6687
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.957078416380888e-05,
+      "loss": 1.9019,
+      "step": 6688
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.954309300639918e-05,
+      "loss": 1.4064,
+      "step": 6689
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.951540188402579e-05,
+      "loss": 1.1933,
+      "step": 6690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.948771079881212e-05,
+      "loss": 1.2252,
+      "step": 6691
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.946001975288157e-05,
+      "loss": 1.2166,
+      "step": 6692
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.943232874835752e-05,
+      "loss": 1.419,
+      "step": 6693
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.940463778736341e-05,
+      "loss": 1.4498,
+      "step": 6694
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.937694687202255e-05,
+      "loss": 1.8179,
+      "step": 6695
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.934925600445834e-05,
+      "loss": 1.6442,
+      "step": 6696
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.932156518679418e-05,
+      "loss": 1.297,
+      "step": 6697
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.929387442115346e-05,
+      "loss": 1.1059,
+      "step": 6698
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.926618370965954e-05,
+      "loss": 1.4856,
+      "step": 6699
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.923849305443573e-05,
+      "loss": 1.0299,
+      "step": 6700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.921080245760543e-05,
+      "loss": 1.2023,
+      "step": 6701
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.918311192129202e-05,
+      "loss": 1.0996,
+      "step": 6702
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.915542144761882e-05,
+      "loss": 1.0387,
+      "step": 6703
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.912773103870921e-05,
+      "loss": 1.6752,
+      "step": 6704
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.910004069668655e-05,
+      "loss": 0.7947,
+      "step": 6705
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.907235042367409e-05,
+      "loss": 1.4139,
+      "step": 6706
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.90446602217952e-05,
+      "loss": 1.4693,
+      "step": 6707
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.901697009317323e-05,
+      "loss": 1.5818,
+      "step": 6708
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.898928003993151e-05,
+      "loss": 1.7689,
+      "step": 6709
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.896159006419338e-05,
+      "loss": 1.0487,
+      "step": 6710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.893390016808207e-05,
+      "loss": 1.0219,
+      "step": 6711
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.89062103537209e-05,
+      "loss": 1.252,
+      "step": 6712
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.887852062323319e-05,
+      "loss": 1.6245,
+      "step": 6713
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.885083097874223e-05,
+      "loss": 1.4475,
+      "step": 6714
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.882314142237134e-05,
+      "loss": 1.5428,
+      "step": 6715
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.879545195624373e-05,
+      "loss": 1.3043,
+      "step": 6716
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.876776258248271e-05,
+      "loss": 1.5055,
+      "step": 6717
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.874007330321153e-05,
+      "loss": 1.2718,
+      "step": 6718
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.871238412055345e-05,
+      "loss": 1.3974,
+      "step": 6719
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.868469503663172e-05,
+      "loss": 1.7375,
+      "step": 6720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.86570060535696e-05,
+      "loss": 1.9173,
+      "step": 6721
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.862931717349028e-05,
+      "loss": 1.1254,
+      "step": 6722
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.8601628398517e-05,
+      "loss": 1.6578,
+      "step": 6723
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.857393973077301e-05,
+      "loss": 1.5126,
+      "step": 6724
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.854625117238149e-05,
+      "loss": 0.8612,
+      "step": 6725
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.851856272546565e-05,
+      "loss": 0.9374,
+      "step": 6726
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.849087439214868e-05,
+      "loss": 1.7099,
+      "step": 6727
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.846318617455376e-05,
+      "loss": 0.8668,
+      "step": 6728
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.843549807480407e-05,
+      "loss": 1.9604,
+      "step": 6729
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.840781009502276e-05,
+      "loss": 1.0608,
+      "step": 6730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.838012223733302e-05,
+      "loss": 1.721,
+      "step": 6731
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.835243450385797e-05,
+      "loss": 1.3018,
+      "step": 6732
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.832474689672076e-05,
+      "loss": 1.4462,
+      "step": 6733
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.82970594180445e-05,
+      "loss": 0.3068,
+      "step": 6734
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.826937206995233e-05,
+      "loss": 1.359,
+      "step": 6735
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.824168485456734e-05,
+      "loss": 1.5617,
+      "step": 6736
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.821399777401265e-05,
+      "loss": 1.5006,
+      "step": 6737
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.818631083041133e-05,
+      "loss": 1.0772,
+      "step": 6738
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.815862402588646e-05,
+      "loss": 0.9777,
+      "step": 6739
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.81309373625611e-05,
+      "loss": 1.7133,
+      "step": 6740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.810325084255832e-05,
+      "loss": 0.8112,
+      "step": 6741
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.80755644680012e-05,
+      "loss": 0.7898,
+      "step": 6742
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.804787824101269e-05,
+      "loss": 1.2814,
+      "step": 6743
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.802019216371583e-05,
+      "loss": 1.0182,
+      "step": 6744
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.799250623823369e-05,
+      "loss": 1.4123,
+      "step": 6745
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.796482046668922e-05,
+      "loss": 1.3725,
+      "step": 6746
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.793713485120545e-05,
+      "loss": 1.415,
+      "step": 6747
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.79094493939053e-05,
+      "loss": 1.5645,
+      "step": 6748
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.788176409691175e-05,
+      "loss": 1.994,
+      "step": 6749
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.785407896234777e-05,
+      "loss": 1.748,
+      "step": 6750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.782639399233628e-05,
+      "loss": 1.9332,
+      "step": 6751
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.779870918900024e-05,
+      "loss": 1.43,
+      "step": 6752
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.77710245544625e-05,
+      "loss": 1.1711,
+      "step": 6753
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.774334009084603e-05,
+      "loss": 1.2682,
+      "step": 6754
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.771565580027366e-05,
+      "loss": 1.6696,
+      "step": 6755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.768797168486828e-05,
+      "loss": 0.9566,
+      "step": 6756
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.766028774675277e-05,
+      "loss": 1.1381,
+      "step": 6757
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.763260398804998e-05,
+      "loss": 1.6278,
+      "step": 6758
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.760492041088272e-05,
+      "loss": 0.9002,
+      "step": 6759
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.75772370173738e-05,
+      "loss": 1.3933,
+      "step": 6760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.754955380964605e-05,
+      "loss": 0.9884,
+      "step": 6761
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.752187078982225e-05,
+      "loss": 1.2065,
+      "step": 6762
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.74941879600252e-05,
+      "loss": 1.2117,
+      "step": 6763
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.74665053223776e-05,
+      "loss": 0.9354,
+      "step": 6764
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.743882287900225e-05,
+      "loss": 1.4025,
+      "step": 6765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.741114063202186e-05,
+      "loss": 0.8795,
+      "step": 6766
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.738345858355917e-05,
+      "loss": 0.46,
+      "step": 6767
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.735577673573688e-05,
+      "loss": 1.1656,
+      "step": 6768
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.732809509067764e-05,
+      "loss": 1.6897,
+      "step": 6769
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.730041365050414e-05,
+      "loss": 0.9426,
+      "step": 6770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.727273241733906e-05,
+      "loss": 1.072,
+      "step": 6771
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.724505139330498e-05,
+      "loss": 1.3637,
+      "step": 6772
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.721737058052459e-05,
+      "loss": 1.0322,
+      "step": 6773
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.71896899811205e-05,
+      "loss": 1.6462,
+      "step": 6774
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.716200959721521e-05,
+      "loss": 1.4976,
+      "step": 6775
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.713432943093137e-05,
+      "loss": 1.7452,
+      "step": 6776
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.710664948439152e-05,
+      "loss": 1.3108,
+      "step": 6777
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.70789697597182e-05,
+      "loss": 1.1947,
+      "step": 6778
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.705129025903397e-05,
+      "loss": 0.8199,
+      "step": 6779
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.702361098446126e-05,
+      "loss": 1.9293,
+      "step": 6780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.699593193812261e-05,
+      "loss": 1.8351,
+      "step": 6781
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.696825312214045e-05,
+      "loss": 1.6968,
+      "step": 6782
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.694057453863729e-05,
+      "loss": 0.7731,
+      "step": 6783
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.691289618973558e-05,
+      "loss": 1.1725,
+      "step": 6784
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.688521807755766e-05,
+      "loss": 1.7756,
+      "step": 6785
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.685754020422597e-05,
+      "loss": 1.5885,
+      "step": 6786
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.682986257186289e-05,
+      "loss": 1.3543,
+      "step": 6787
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.680218518259078e-05,
+      "loss": 0.7539,
+      "step": 6788
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.6774508038532e-05,
+      "loss": 1.6618,
+      "step": 6789
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.674683114180884e-05,
+      "loss": 1.2997,
+      "step": 6790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.671915449454366e-05,
+      "loss": 1.4037,
+      "step": 6791
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.669147809885868e-05,
+      "loss": 0.9578,
+      "step": 6792
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.666380195687622e-05,
+      "loss": 1.2295,
+      "step": 6793
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.663612607071851e-05,
+      "loss": 1.0528,
+      "step": 6794
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.660845044250782e-05,
+      "loss": 1.5735,
+      "step": 6795
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.658077507436629e-05,
+      "loss": 1.1723,
+      "step": 6796
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.655309996841614e-05,
+      "loss": 1.4204,
+      "step": 6797
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.652542512677954e-05,
+      "loss": 1.5886,
+      "step": 6798
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.649775055157866e-05,
+      "loss": 1.5343,
+      "step": 6799
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.647007624493562e-05,
+      "loss": 1.1315,
+      "step": 6800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.644240220897248e-05,
+      "loss": 1.2402,
+      "step": 6801
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.641472844581138e-05,
+      "loss": 1.2708,
+      "step": 6802
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.638705495757436e-05,
+      "loss": 1.0843,
+      "step": 6803
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.635938174638348e-05,
+      "loss": 2.2153,
+      "step": 6804
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.633170881436079e-05,
+      "loss": 1.0189,
+      "step": 6805
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.630403616362824e-05,
+      "loss": 1.0663,
+      "step": 6806
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.627636379630783e-05,
+      "loss": 1.6828,
+      "step": 6807
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.624869171452152e-05,
+      "loss": 1.2699,
+      "step": 6808
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.622101992039125e-05,
+      "loss": 1.444,
+      "step": 6809
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.619334841603895e-05,
+      "loss": 1.4805,
+      "step": 6810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.616567720358654e-05,
+      "loss": 1.1904,
+      "step": 6811
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.613800628515577e-05,
+      "loss": 1.1934,
+      "step": 6812
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.61103356628686e-05,
+      "loss": 1.1358,
+      "step": 6813
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.608266533884681e-05,
+      "loss": 1.1361,
+      "step": 6814
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.605499531521223e-05,
+      "loss": 1.5613,
+      "step": 6815
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.602732559408667e-05,
+      "loss": 1.3557,
+      "step": 6816
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.59996561775918e-05,
+      "loss": 1.1493,
+      "step": 6817
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.597198706784939e-05,
+      "loss": 1.3117,
+      "step": 6818
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.594431826698113e-05,
+      "loss": 1.5177,
+      "step": 6819
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.591664977710875e-05,
+      "loss": 1.1731,
+      "step": 6820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.588898160035395e-05,
+      "loss": 1.273,
+      "step": 6821
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.586131373883826e-05,
+      "loss": 1.0898,
+      "step": 6822
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.583364619468335e-05,
+      "loss": 0.92,
+      "step": 6823
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.58059789700108e-05,
+      "loss": 1.7759,
+      "step": 6824
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.577831206694218e-05,
+      "loss": 1.4466,
+      "step": 6825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.575064548759902e-05,
+      "loss": 0.6971,
+      "step": 6826
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.572297923410288e-05,
+      "loss": 2.5867,
+      "step": 6827
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.569531330857518e-05,
+      "loss": 2.0818,
+      "step": 6828
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.566764771313743e-05,
+      "loss": 1.5081,
+      "step": 6829
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.563998244991105e-05,
+      "loss": 1.0272,
+      "step": 6830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.561231752101749e-05,
+      "loss": 0.9457,
+      "step": 6831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.558465292857812e-05,
+      "loss": 1.1961,
+      "step": 6832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.555698867471427e-05,
+      "loss": 1.2257,
+      "step": 6833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.552932476154733e-05,
+      "loss": 1.3924,
+      "step": 6834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.550166119119856e-05,
+      "loss": 1.1865,
+      "step": 6835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.547399796578927e-05,
+      "loss": 1.493,
+      "step": 6836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.544633508744075e-05,
+      "loss": 1.6179,
+      "step": 6837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.541867255827418e-05,
+      "loss": 1.7273,
+      "step": 6838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.539101038041079e-05,
+      "loss": 1.2445,
+      "step": 6839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.536334855597174e-05,
+      "loss": 1.1808,
+      "step": 6840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.533568708707821e-05,
+      "loss": 1.4084,
+      "step": 6841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.530802597585133e-05,
+      "loss": 1.4434,
+      "step": 6842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.528036522441212e-05,
+      "loss": 1.6074,
+      "step": 6843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.525270483488173e-05,
+      "loss": 1.4379,
+      "step": 6844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.522504480938116e-05,
+      "loss": 1.6136,
+      "step": 6845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.519738515003146e-05,
+      "loss": 1.3739,
+      "step": 6846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.516972585895357e-05,
+      "loss": 1.3369,
+      "step": 6847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.514206693826853e-05,
+      "loss": 1.3542,
+      "step": 6848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.511440839009714e-05,
+      "loss": 1.0191,
+      "step": 6849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.508675021656036e-05,
+      "loss": 1.1434,
+      "step": 6850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.505909241977908e-05,
+      "loss": 1.5128,
+      "step": 6851
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.503143500187414e-05,
+      "loss": 1.5688,
+      "step": 6852
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.500377796496639e-05,
+      "loss": 1.4942,
+      "step": 6853
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.49761213111765e-05,
+      "loss": 1.4596,
+      "step": 6854
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.49484650426253e-05,
+      "loss": 0.9551,
+      "step": 6855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.492080916143351e-05,
+      "loss": 1.4784,
+      "step": 6856
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.489315366972179e-05,
+      "loss": 1.7425,
+      "step": 6857
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.48654985696109e-05,
+      "loss": 1.3879,
+      "step": 6858
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.483784386322138e-05,
+      "loss": 1.0883,
+      "step": 6859
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.481018955267386e-05,
+      "loss": 1.5077,
+      "step": 6860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.478253564008891e-05,
+      "loss": 1.1685,
+      "step": 6861
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.47548821275871e-05,
+      "loss": 1.2797,
+      "step": 6862
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.472722901728891e-05,
+      "loss": 1.3874,
+      "step": 6863
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.469957631131487e-05,
+      "loss": 1.7349,
+      "step": 6864
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.46719240117854e-05,
+      "loss": 1.0769,
+      "step": 6865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.46442721208209e-05,
+      "loss": 0.7686,
+      "step": 6866
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.461662064054178e-05,
+      "loss": 1.6469,
+      "step": 6867
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.458896957306841e-05,
+      "loss": 1.4095,
+      "step": 6868
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.456131892052113e-05,
+      "loss": 1.663,
+      "step": 6869
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.453366868502018e-05,
+      "loss": 1.1781,
+      "step": 6870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.450601886868586e-05,
+      "loss": 1.4225,
+      "step": 6871
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.447836947363838e-05,
+      "loss": 1.1111,
+      "step": 6872
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.445072050199797e-05,
+      "loss": 1.332,
+      "step": 6873
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.442307195588477e-05,
+      "loss": 1.0074,
+      "step": 6874
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.439542383741893e-05,
+      "loss": 1.2646,
+      "step": 6875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.436777614872052e-05,
+      "loss": 1.4466,
+      "step": 6876
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.434012889190962e-05,
+      "loss": 1.1678,
+      "step": 6877
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.431248206910627e-05,
+      "loss": 0.9293,
+      "step": 6878
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.428483568243052e-05,
+      "loss": 1.0697,
+      "step": 6879
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.425718973400222e-05,
+      "loss": 0.3716,
+      "step": 6880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.42295442259414e-05,
+      "loss": 1.1165,
+      "step": 6881
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.42018991603679e-05,
+      "loss": 1.3319,
+      "step": 6882
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.417425453940164e-05,
+      "loss": 1.4949,
+      "step": 6883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.414661036516243e-05,
+      "loss": 1.1085,
+      "step": 6884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.41189666397701e-05,
+      "loss": 1.36,
+      "step": 6885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.409132336534433e-05,
+      "loss": 1.4154,
+      "step": 6886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.406368054400487e-05,
+      "loss": 1.3911,
+      "step": 6887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.403603817787147e-05,
+      "loss": 0.6948,
+      "step": 6888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.400839626906376e-05,
+      "loss": 1.6392,
+      "step": 6889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.39807548197014e-05,
+      "loss": 1.6222,
+      "step": 6890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.39531138319039e-05,
+      "loss": 1.3443,
+      "step": 6891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.392547330779086e-05,
+      "loss": 1.6672,
+      "step": 6892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.389783324948178e-05,
+      "loss": 0.64,
+      "step": 6893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.387019365909615e-05,
+      "loss": 1.4963,
+      "step": 6894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.384255453875346e-05,
+      "loss": 1.5019,
+      "step": 6895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.381491589057305e-05,
+      "loss": 1.4415,
+      "step": 6896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.378727771667431e-05,
+      "loss": 1.4715,
+      "step": 6897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.375964001917661e-05,
+      "loss": 1.1469,
+      "step": 6898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.373200280019922e-05,
+      "loss": 1.2814,
+      "step": 6899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.370436606186142e-05,
+      "loss": 1.5006,
+      "step": 6900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.367672980628245e-05,
+      "loss": 1.2511,
+      "step": 6901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.364909403558146e-05,
+      "loss": 1.3641,
+      "step": 6902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.362145875187762e-05,
+      "loss": 1.3545,
+      "step": 6903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.359382395729006e-05,
+      "loss": 1.3119,
+      "step": 6904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.356618965393784e-05,
+      "loss": 1.1144,
+      "step": 6905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.353855584394005e-05,
+      "loss": 0.9856,
+      "step": 6906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.351092252941562e-05,
+      "loss": 1.254,
+      "step": 6907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.348328971248354e-05,
+      "loss": 1.4117,
+      "step": 6908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.345565739526274e-05,
+      "loss": 1.1678,
+      "step": 6909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.342802557987212e-05,
+      "loss": 0.9948,
+      "step": 6910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.340039426843055e-05,
+      "loss": 1.5826,
+      "step": 6911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.337276346305679e-05,
+      "loss": 1.09,
+      "step": 6912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.334513316586963e-05,
+      "loss": 1.5592,
+      "step": 6913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.331750337898781e-05,
+      "loss": 1.7106,
+      "step": 6914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.328987410453002e-05,
+      "loss": 1.0415,
+      "step": 6915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.326224534461492e-05,
+      "loss": 1.6758,
+      "step": 6916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.323461710136116e-05,
+      "loss": 0.8975,
+      "step": 6917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.320698937688724e-05,
+      "loss": 0.8782,
+      "step": 6918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.317936217331172e-05,
+      "loss": 1.3375,
+      "step": 6919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.315173549275313e-05,
+      "loss": 1.5962,
+      "step": 6920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.312410933732992e-05,
+      "loss": 1.4333,
+      "step": 6921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.309648370916051e-05,
+      "loss": 1.2796,
+      "step": 6922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.306885861036323e-05,
+      "loss": 1.5634,
+      "step": 6923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.304123404305644e-05,
+      "loss": 1.1877,
+      "step": 6924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.301361000935841e-05,
+      "loss": 1.298,
+      "step": 6925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.298598651138745e-05,
+      "loss": 1.7436,
+      "step": 6926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.295836355126178e-05,
+      "loss": 1.4919,
+      "step": 6927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.293074113109949e-05,
+      "loss": 1.2366,
+      "step": 6928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.290311925301876e-05,
+      "loss": 1.3785,
+      "step": 6929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.287549791913765e-05,
+      "loss": 1.0465,
+      "step": 6930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.284787713157424e-05,
+      "loss": 1.2412,
+      "step": 6931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.282025689244655e-05,
+      "loss": 1.9013,
+      "step": 6932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.279263720387247e-05,
+      "loss": 1.7155,
+      "step": 6933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.276501806796997e-05,
+      "loss": 1.3477,
+      "step": 6934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.273739948685691e-05,
+      "loss": 1.2796,
+      "step": 6935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.270978146265115e-05,
+      "loss": 1.4793,
+      "step": 6936
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.268216399747046e-05,
+      "loss": 1.7424,
+      "step": 6937
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.265454709343262e-05,
+      "loss": 1.1559,
+      "step": 6938
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.262693075265528e-05,
+      "loss": 0.891,
+      "step": 6939
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.259931497725614e-05,
+      "loss": 1.1919,
+      "step": 6940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.257169976935282e-05,
+      "loss": 1.2905,
+      "step": 6941
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.254408513106288e-05,
+      "loss": 1.6073,
+      "step": 6942
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.25164710645039e-05,
+      "loss": 1.2514,
+      "step": 6943
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.24888575717933e-05,
+      "loss": 1.2783,
+      "step": 6944
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.246124465504856e-05,
+      "loss": 1.1262,
+      "step": 6945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.243363231638708e-05,
+      "loss": 1.2692,
+      "step": 6946
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.240602055792621e-05,
+      "loss": 1.5948,
+      "step": 6947
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.237840938178328e-05,
+      "loss": 1.5609,
+      "step": 6948
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.235079879007553e-05,
+      "loss": 1.4776,
+      "step": 6949
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.23231887849202e-05,
+      "loss": 1.0954,
+      "step": 6950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.229557936843446e-05,
+      "loss": 0.845,
+      "step": 6951
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.226797054273544e-05,
+      "loss": 1.1762,
+      "step": 6952
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.224036230994023e-05,
+      "loss": 1.3331,
+      "step": 6953
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.221275467216592e-05,
+      "loss": 1.5076,
+      "step": 6954
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.218514763152938e-05,
+      "loss": 1.013,
+      "step": 6955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.215754119014767e-05,
+      "loss": 1.126,
+      "step": 6956
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.212993535013767e-05,
+      "loss": 0.6744,
+      "step": 6957
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.210233011361623e-05,
+      "loss": 1.1184,
+      "step": 6958
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.207472548270021e-05,
+      "loss": 1.2003,
+      "step": 6959
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.204712145950627e-05,
+      "loss": 1.4181,
+      "step": 6960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.201951804615118e-05,
+      "loss": 1.8957,
+      "step": 6961
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.199191524475162e-05,
+      "loss": 1.0721,
+      "step": 6962
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.196431305742421e-05,
+      "loss": 1.4833,
+      "step": 6963
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.193671148628559e-05,
+      "loss": 1.6913,
+      "step": 6964
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.190911053345219e-05,
+      "loss": 1.6066,
+      "step": 6965
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.188151020104052e-05,
+      "loss": 1.2249,
+      "step": 6966
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.185391049116703e-05,
+      "loss": 2.0457,
+      "step": 6967
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.182631140594812e-05,
+      "loss": 1.2356,
+      "step": 6968
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.179871294750013e-05,
+      "loss": 1.2901,
+      "step": 6969
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.177111511793932e-05,
+      "loss": 1.5774,
+      "step": 6970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.174351791938195e-05,
+      "loss": 1.4614,
+      "step": 6971
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.171592135394423e-05,
+      "loss": 1.1709,
+      "step": 6972
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.168832542374228e-05,
+      "loss": 1.2954,
+      "step": 6973
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.166073013089223e-05,
+      "loss": 1.896,
+      "step": 6974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.163313547751012e-05,
+      "loss": 1.3348,
+      "step": 6975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.160554146571192e-05,
+      "loss": 1.9286,
+      "step": 6976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.157794809761361e-05,
+      "loss": 1.7474,
+      "step": 6977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.155035537533108e-05,
+      "loss": 1.3557,
+      "step": 6978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.152276330098018e-05,
+      "loss": 1.9755,
+      "step": 6979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.149517187667675e-05,
+      "loss": 1.0785,
+      "step": 6980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.14675811045365e-05,
+      "loss": 1.0545,
+      "step": 6981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.143999098667514e-05,
+      "loss": 1.3156,
+      "step": 6982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.141240152520832e-05,
+      "loss": 1.2471,
+      "step": 6983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.138481272225164e-05,
+      "loss": 2.0952,
+      "step": 6984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.135722457992071e-05,
+      "loss": 1.1902,
+      "step": 6985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.132963710033093e-05,
+      "loss": 1.7855,
+      "step": 6986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.13020502855978e-05,
+      "loss": 1.4912,
+      "step": 6987
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.127446413783673e-05,
+      "loss": 1.2049,
+      "step": 6988
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.124687865916306e-05,
+      "loss": 2.0344,
+      "step": 6989
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.121929385169209e-05,
+      "loss": 1.3141,
+      "step": 6990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.11917097175391e-05,
+      "loss": 1.3694,
+      "step": 6991
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.116412625881918e-05,
+      "loss": 1.4063,
+      "step": 6992
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.113654347764753e-05,
+      "loss": 0.8143,
+      "step": 6993
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.110896137613927e-05,
+      "loss": 0.9124,
+      "step": 6994
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.108137995640941e-05,
+      "loss": 1.1992,
+      "step": 6995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.105379922057297e-05,
+      "loss": 1.7806,
+      "step": 6996
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.10262191707448e-05,
+      "loss": 1.1041,
+      "step": 6997
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.099863980903985e-05,
+      "loss": 1.2312,
+      "step": 6998
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.097106113757292e-05,
+      "loss": 1.3888,
+      "step": 6999
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.094348315845877e-05,
+      "loss": 1.8409,
+      "step": 7000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.091590587381222e-05,
+      "loss": 1.0753,
+      "step": 7001
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.088832928574781e-05,
+      "loss": 1.5631,
+      "step": 7002
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.08607533963802e-05,
+      "loss": 1.5405,
+      "step": 7003
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.083317820782397e-05,
+      "loss": 1.3686,
+      "step": 7004
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.080560372219361e-05,
+      "loss": 1.1613,
+      "step": 7005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.077802994160358e-05,
+      "loss": 1.7624,
+      "step": 7006
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.07504568681683e-05,
+      "loss": 1.0701,
+      "step": 7007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.072288450400208e-05,
+      "loss": 1.1001,
+      "step": 7008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.069531285121921e-05,
+      "loss": 1.3957,
+      "step": 7009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.066774191193395e-05,
+      "loss": 1.4595,
+      "step": 7010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.064017168826048e-05,
+      "loss": 1.9368,
+      "step": 7011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.061260218231293e-05,
+      "loss": 1.1474,
+      "step": 7012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.058503339620536e-05,
+      "loss": 1.8141,
+      "step": 7013
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.055746533205176e-05,
+      "loss": 2.1307,
+      "step": 7014
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.052989799196613e-05,
+      "loss": 1.5919,
+      "step": 7015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.050233137806238e-05,
+      "loss": 1.5315,
+      "step": 7016
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.047476549245437e-05,
+      "loss": 1.4011,
+      "step": 7017
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.044720033725583e-05,
+      "loss": 1.5069,
+      "step": 7018
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.041963591458053e-05,
+      "loss": 1.0344,
+      "step": 7019
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.039207222654218e-05,
+      "loss": 1.5968,
+      "step": 7020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.036450927525438e-05,
+      "loss": 1.4675,
+      "step": 7021
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.033694706283073e-05,
+      "loss": 1.4602,
+      "step": 7022
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.030938559138466e-05,
+      "loss": 1.4957,
+      "step": 7023
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.028182486302971e-05,
+      "loss": 1.7672,
+      "step": 7024
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.025426487987927e-05,
+      "loss": 1.4901,
+      "step": 7025
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.022670564404664e-05,
+      "loss": 0.9056,
+      "step": 7026
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.019914715764513e-05,
+      "loss": 1.3047,
+      "step": 7027
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.0171589422788e-05,
+      "loss": 1.8232,
+      "step": 7028
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.014403244158835e-05,
+      "loss": 1.2188,
+      "step": 7029
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.011647621615931e-05,
+      "loss": 1.2327,
+      "step": 7030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.008892074861396e-05,
+      "loss": 1.9181,
+      "step": 7031
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.006136604106527e-05,
+      "loss": 1.3912,
+      "step": 7032
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.003381209562626e-05,
+      "loss": 0.7079,
+      "step": 7033
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.000625891440968e-05,
+      "loss": 1.4299,
+      "step": 7034
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.99787064995284e-05,
+      "loss": 0.7968,
+      "step": 7035
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.99511548530952e-05,
+      "loss": 1.0701,
+      "step": 7036
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.992360397722275e-05,
+      "loss": 1.5962,
+      "step": 7037
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.989605387402378e-05,
+      "loss": 0.9464,
+      "step": 7038
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.986850454561076e-05,
+      "loss": 1.2116,
+      "step": 7039
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.984095599409627e-05,
+      "loss": 1.2783,
+      "step": 7040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.981340822159278e-05,
+      "loss": 1.8781,
+      "step": 7041
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.978586123021267e-05,
+      "loss": 0.6485,
+      "step": 7042
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.975831502206829e-05,
+      "loss": 1.5378,
+      "step": 7043
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.973076959927197e-05,
+      "loss": 1.2917,
+      "step": 7044
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.970322496393587e-05,
+      "loss": 0.8436,
+      "step": 7045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.96756811181722e-05,
+      "loss": 1.5696,
+      "step": 7046
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.964813806409303e-05,
+      "loss": 1.4281,
+      "step": 7047
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.962059580381042e-05,
+      "loss": 1.4454,
+      "step": 7048
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.959305433943639e-05,
+      "loss": 1.2723,
+      "step": 7049
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.956551367308279e-05,
+      "loss": 1.0578,
+      "step": 7050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.953797380686153e-05,
+      "loss": 0.7629,
+      "step": 7051
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.951043474288438e-05,
+      "loss": 1.4093,
+      "step": 7052
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.948289648326311e-05,
+      "loss": 1.9133,
+      "step": 7053
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.945535903010937e-05,
+      "loss": 1.2059,
+      "step": 7054
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.942782238553479e-05,
+      "loss": 0.8122,
+      "step": 7055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.94002865516509e-05,
+      "loss": 1.5265,
+      "step": 7056
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.93727515305692e-05,
+      "loss": 1.5958,
+      "step": 7057
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.934521732440113e-05,
+      "loss": 1.2818,
+      "step": 7058
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.931768393525809e-05,
+      "loss": 0.8493,
+      "step": 7059
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.929015136525129e-05,
+      "loss": 0.9614,
+      "step": 7060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.9262619616492e-05,
+      "loss": 1.281,
+      "step": 7061
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.923508869109143e-05,
+      "loss": 0.9741,
+      "step": 7062
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.920755859116067e-05,
+      "loss": 1.4711,
+      "step": 7063
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.918002931881079e-05,
+      "loss": 2.2732,
+      "step": 7064
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.91525008761528e-05,
+      "loss": 1.4735,
+      "step": 7065
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.912497326529755e-05,
+      "loss": 1.0211,
+      "step": 7066
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.909744648835594e-05,
+      "loss": 1.4934,
+      "step": 7067
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.906992054743873e-05,
+      "loss": 1.3011,
+      "step": 7068
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.90423954446567e-05,
+      "loss": 0.6849,
+      "step": 7069
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.901487118212056e-05,
+      "loss": 1.209,
+      "step": 7070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.898734776194079e-05,
+      "loss": 1.64,
+      "step": 7071
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.8959825186228e-05,
+      "loss": 1.475,
+      "step": 7072
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.893230345709265e-05,
+      "loss": 1.1112,
+      "step": 7073
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.890478257664516e-05,
+      "loss": 1.1409,
+      "step": 7074
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.887726254699586e-05,
+      "loss": 1.9903,
+      "step": 7075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.884974337025503e-05,
+      "loss": 1.293,
+      "step": 7076
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.882222504853288e-05,
+      "loss": 1.4423,
+      "step": 7077
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.879470758393957e-05,
+      "loss": 1.1598,
+      "step": 7078
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.876719097858516e-05,
+      "loss": 1.8913,
+      "step": 7079
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.873967523457968e-05,
+      "loss": 1.0662,
+      "step": 7080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.87121603540331e-05,
+      "loss": 0.933,
+      "step": 7081
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.868464633905527e-05,
+      "loss": 1.2429,
+      "step": 7082
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.865713319175599e-05,
+      "loss": 1.046,
+      "step": 7083
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.862962091424506e-05,
+      "loss": 1.4847,
+      "step": 7084
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.860210950863212e-05,
+      "loss": 1.5397,
+      "step": 7085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.857459897702684e-05,
+      "loss": 1.8543,
+      "step": 7086
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.854708932153872e-05,
+      "loss": 1.5219,
+      "step": 7087
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.851958054427725e-05,
+      "loss": 1.5436,
+      "step": 7088
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.849207264735186e-05,
+      "loss": 1.5215,
+      "step": 7089
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.846456563287189e-05,
+      "loss": 1.35,
+      "step": 7090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.843705950294663e-05,
+      "loss": 1.0164,
+      "step": 7091
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.840955425968527e-05,
+      "loss": 0.9936,
+      "step": 7092
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.838204990519697e-05,
+      "loss": 1.5713,
+      "step": 7093
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.835454644159079e-05,
+      "loss": 1.8255,
+      "step": 7094
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.832704387097576e-05,
+      "loss": 1.0211,
+      "step": 7095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.82995421954608e-05,
+      "loss": 1.2643,
+      "step": 7096
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.827204141715485e-05,
+      "loss": 1.013,
+      "step": 7097
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.824454153816657e-05,
+      "loss": 1.1755,
+      "step": 7098
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.82170425606048e-05,
+      "loss": 1.2536,
+      "step": 7099
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.818954448657819e-05,
+      "loss": 1.8722,
+      "step": 7100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.81620473181953e-05,
+      "loss": 1.3107,
+      "step": 7101
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.813455105756473e-05,
+      "loss": 1.2399,
+      "step": 7102
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.810705570679484e-05,
+      "loss": 1.8788,
+      "step": 7103
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.807956126799406e-05,
+      "loss": 0.8191,
+      "step": 7104
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.805206774327067e-05,
+      "loss": 1.3979,
+      "step": 7105
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.802457513473298e-05,
+      "loss": 1.0072,
+      "step": 7106
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.799708344448917e-05,
+      "loss": 1.3512,
+      "step": 7107
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.796959267464726e-05,
+      "loss": 1.3945,
+      "step": 7108
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.794210282731534e-05,
+      "loss": 1.346,
+      "step": 7109
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.791461390460136e-05,
+      "loss": 1.2412,
+      "step": 7110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.788712590861323e-05,
+      "loss": 0.8345,
+      "step": 7111
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.785963884145876e-05,
+      "loss": 1.2533,
+      "step": 7112
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.78321527052457e-05,
+      "loss": 1.0768,
+      "step": 7113
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.780466750208171e-05,
+      "loss": 0.9903,
+      "step": 7114
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.777718323407444e-05,
+      "loss": 1.3126,
+      "step": 7115
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.774969990333137e-05,
+      "loss": 1.4117,
+      "step": 7116
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.772221751196002e-05,
+      "loss": 1.7205,
+      "step": 7117
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.769473606206776e-05,
+      "loss": 0.9792,
+      "step": 7118
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.76672555557619e-05,
+      "loss": 1.377,
+      "step": 7119
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.763977599514968e-05,
+      "loss": 1.4423,
+      "step": 7120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.76122973823383e-05,
+      "loss": 1.0924,
+      "step": 7121
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.758481971943483e-05,
+      "loss": 1.5485,
+      "step": 7122
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.755734300854637e-05,
+      "loss": 1.0454,
+      "step": 7123
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.752986725177978e-05,
+      "loss": 1.4816,
+      "step": 7124
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.750239245124201e-05,
+      "loss": 1.3496,
+      "step": 7125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.747491860903982e-05,
+      "loss": 0.8716,
+      "step": 7126
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.744744572728e-05,
+      "loss": 1.2872,
+      "step": 7127
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.741997380806919e-05,
+      "loss": 1.438,
+      "step": 7128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.739250285351396e-05,
+      "loss": 1.0751,
+      "step": 7129
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.736503286572085e-05,
+      "loss": 1.308,
+      "step": 7130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.733756384679626e-05,
+      "loss": 1.217,
+      "step": 7131
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.731009579884661e-05,
+      "loss": 1.2701,
+      "step": 7132
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.728262872397816e-05,
+      "loss": 1.3156,
+      "step": 7133
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.725516262429718e-05,
+      "loss": 1.1941,
+      "step": 7134
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.722769750190973e-05,
+      "loss": 1.5085,
+      "step": 7135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.720023335892189e-05,
+      "loss": 0.6296,
+      "step": 7136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.71727701974397e-05,
+      "loss": 0.8903,
+      "step": 7137
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.714530801956906e-05,
+      "loss": 1.4233,
+      "step": 7138
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.711784682741583e-05,
+      "loss": 1.504,
+      "step": 7139
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.709038662308573e-05,
+      "loss": 1.4378,
+      "step": 7140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.706292740868446e-05,
+      "loss": 1.1257,
+      "step": 7141
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.703546918631767e-05,
+      "loss": 1.2237,
+      "step": 7142
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.700801195809083e-05,
+      "loss": 2.1454,
+      "step": 7143
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.698055572610952e-05,
+      "loss": 1.303,
+      "step": 7144
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.695310049247901e-05,
+      "loss": 1.3653,
+      "step": 7145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.692564625930466e-05,
+      "loss": 0.9084,
+      "step": 7146
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.689819302869167e-05,
+      "loss": 1.031,
+      "step": 7147
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.687074080274524e-05,
+      "loss": 1.344,
+      "step": 7148
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.684328958357047e-05,
+      "loss": 1.069,
+      "step": 7149
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.681583937327226e-05,
+      "loss": 1.212,
+      "step": 7150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.678839017395563e-05,
+      "loss": 1.8046,
+      "step": 7151
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.676094198772536e-05,
+      "loss": 1.8702,
+      "step": 7152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.673349481668627e-05,
+      "loss": 1.2209,
+      "step": 7153
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.670604866294303e-05,
+      "loss": 1.0595,
+      "step": 7154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.667860352860026e-05,
+      "loss": 1.0777,
+      "step": 7155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.665115941576249e-05,
+      "loss": 1.3306,
+      "step": 7156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.662371632653417e-05,
+      "loss": 1.2014,
+      "step": 7157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.659627426301967e-05,
+      "loss": 0.9702,
+      "step": 7158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.656883322732331e-05,
+      "loss": 1.3278,
+      "step": 7159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.654139322154931e-05,
+      "loss": 1.1876,
+      "step": 7160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.651395424780179e-05,
+      "loss": 1.8261,
+      "step": 7161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.648651630818486e-05,
+      "loss": 1.6899,
+      "step": 7162
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.645907940480243e-05,
+      "loss": 1.4421,
+      "step": 7163
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.643164353975846e-05,
+      "loss": 1.3797,
+      "step": 7164
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.640420871515679e-05,
+      "loss": 0.8783,
+      "step": 7165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.637677493310106e-05,
+      "loss": 0.7982,
+      "step": 7166
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.634934219569505e-05,
+      "loss": 1.1466,
+      "step": 7167
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.63219105050423e-05,
+      "loss": 1.4377,
+      "step": 7168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.629447986324629e-05,
+      "loss": 1.2007,
+      "step": 7169
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.626705027241048e-05,
+      "loss": 1.0432,
+      "step": 7170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.623962173463825e-05,
+      "loss": 1.1204,
+      "step": 7171
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.621219425203275e-05,
+      "loss": 1.2707,
+      "step": 7172
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.618476782669722e-05,
+      "loss": 1.0023,
+      "step": 7173
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.615734246073477e-05,
+      "loss": 1.8102,
+      "step": 7174
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.612991815624842e-05,
+      "loss": 1.504,
+      "step": 7175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.610249491534114e-05,
+      "loss": 0.7923,
+      "step": 7176
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.607507274011572e-05,
+      "loss": 0.6562,
+      "step": 7177
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.604765163267494e-05,
+      "loss": 0.8651,
+      "step": 7178
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.602023159512153e-05,
+      "loss": 1.9217,
+      "step": 7179
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.599281262955804e-05,
+      "loss": 1.4707,
+      "step": 7180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.596539473808711e-05,
+      "loss": 1.3579,
+      "step": 7181
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.593797792281109e-05,
+      "loss": 1.3485,
+      "step": 7182
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.591056218583236e-05,
+      "loss": 0.8439,
+      "step": 7183
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.588314752925321e-05,
+      "loss": 1.3661,
+      "step": 7184
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.585573395517583e-05,
+      "loss": 1.3672,
+      "step": 7185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.582832146570234e-05,
+      "loss": 0.9228,
+      "step": 7186
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.58009100629348e-05,
+      "loss": 1.676,
+      "step": 7187
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.577349974897513e-05,
+      "loss": 1.1566,
+      "step": 7188
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.574609052592518e-05,
+      "loss": 0.9001,
+      "step": 7189
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.571868239588674e-05,
+      "loss": 1.4641,
+      "step": 7190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.569127536096153e-05,
+      "loss": 1.833,
+      "step": 7191
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.566386942325116e-05,
+      "loss": 1.4234,
+      "step": 7192
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.563646458485713e-05,
+      "loss": 1.3286,
+      "step": 7193
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.56090608478809e-05,
+      "loss": 0.9735,
+      "step": 7194
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.558165821442385e-05,
+      "loss": 1.4985,
+      "step": 7195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.55542566865872e-05,
+      "loss": 1.8737,
+      "step": 7196
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.552685626647223e-05,
+      "loss": 1.2671,
+      "step": 7197
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.549945695617998e-05,
+      "loss": 1.5408,
+      "step": 7198
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.547205875781147e-05,
+      "loss": 1.2673,
+      "step": 7199
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.544466167346765e-05,
+      "loss": 1.154,
+      "step": 7200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.541726570524939e-05,
+      "loss": 1.4643,
+      "step": 7201
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.538987085525746e-05,
+      "loss": 1.2802,
+      "step": 7202
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.536247712559245e-05,
+      "loss": 0.7158,
+      "step": 7203
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.533508451835505e-05,
+      "loss": 1.1292,
+      "step": 7204
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.530769303564575e-05,
+      "loss": 1.3107,
+      "step": 7205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.528030267956492e-05,
+      "loss": 1.8413,
+      "step": 7206
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.525291345221296e-05,
+      "loss": 0.949,
+      "step": 7207
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.522552535569011e-05,
+      "loss": 1.7601,
+      "step": 7208
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.519813839209648e-05,
+      "loss": 1.1522,
+      "step": 7209
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.517075256353217e-05,
+      "loss": 1.142,
+      "step": 7210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.514336787209716e-05,
+      "loss": 1.0867,
+      "step": 7211
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.511598431989136e-05,
+      "loss": 0.8967,
+      "step": 7212
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.508860190901464e-05,
+      "loss": 1.5621,
+      "step": 7213
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.506122064156663e-05,
+      "loss": 1.1288,
+      "step": 7214
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.503384051964699e-05,
+      "loss": 0.8511,
+      "step": 7215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.500646154535529e-05,
+      "loss": 1.0462,
+      "step": 7216
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.497908372079099e-05,
+      "loss": 1.2885,
+      "step": 7217
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.495170704805348e-05,
+      "loss": 1.5488,
+      "step": 7218
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.4924331529242e-05,
+      "loss": 1.3662,
+      "step": 7219
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.489695716645578e-05,
+      "loss": 1.5195,
+      "step": 7220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.48695839617939e-05,
+      "loss": 1.5311,
+      "step": 7221
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.484221191735541e-05,
+      "loss": 1.1327,
+      "step": 7222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.481484103523922e-05,
+      "loss": 1.5695,
+      "step": 7223
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.478747131754421e-05,
+      "loss": 1.2771,
+      "step": 7224
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.476010276636909e-05,
+      "loss": 0.8234,
+      "step": 7225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.473273538381251e-05,
+      "loss": 1.4587,
+      "step": 7226
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.470536917197308e-05,
+      "loss": 1.3233,
+      "step": 7227
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.467800413294926e-05,
+      "loss": 1.0534,
+      "step": 7228
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.465064026883949e-05,
+      "loss": 1.4079,
+      "step": 7229
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.4623277581742e-05,
+      "loss": 1.1846,
+      "step": 7230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.459591607375503e-05,
+      "loss": 1.1507,
+      "step": 7231
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.456855574697673e-05,
+      "loss": 1.3032,
+      "step": 7232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.454119660350508e-05,
+      "loss": 1.3041,
+      "step": 7233
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.45138386454381e-05,
+      "loss": 1.3161,
+      "step": 7234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.448648187487356e-05,
+      "loss": 1.1643,
+      "step": 7235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.445912629390924e-05,
+      "loss": 1.7319,
+      "step": 7236
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.443177190464283e-05,
+      "loss": 1.673,
+      "step": 7237
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.44044187091719e-05,
+      "loss": 1.042,
+      "step": 7238
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.437706670959395e-05,
+      "loss": 1.7246,
+      "step": 7239
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.434971590800633e-05,
+      "loss": 1.6259,
+      "step": 7240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.432236630650634e-05,
+      "loss": 2.2057,
+      "step": 7241
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.429501790719124e-05,
+      "loss": 1.8531,
+      "step": 7242
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.426767071215809e-05,
+      "loss": 1.6747,
+      "step": 7243
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.424032472350397e-05,
+      "loss": 1.2101,
+      "step": 7244
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.421297994332583e-05,
+      "loss": 0.9575,
+      "step": 7245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.418563637372041e-05,
+      "loss": 1.7718,
+      "step": 7246
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.415829401678451e-05,
+      "loss": 1.367,
+      "step": 7247
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.413095287461477e-05,
+      "loss": 0.9402,
+      "step": 7248
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.41036129493078e-05,
+      "loss": 1.452,
+      "step": 7249
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.407627424296005e-05,
+      "loss": 1.5498,
+      "step": 7250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.404893675766785e-05,
+      "loss": 1.2828,
+      "step": 7251
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.40216004955275e-05,
+      "loss": 0.962,
+      "step": 7252
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.399426545863521e-05,
+      "loss": 1.1677,
+      "step": 7253
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.396693164908705e-05,
+      "loss": 1.2939,
+      "step": 7254
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.393959906897903e-05,
+      "loss": 1.1472,
+      "step": 7255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.391226772040705e-05,
+      "loss": 0.9854,
+      "step": 7256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.388493760546691e-05,
+      "loss": 1.1428,
+      "step": 7257
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.385760872625432e-05,
+      "loss": 0.9529,
+      "step": 7258
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.383028108486493e-05,
+      "loss": 1.1068,
+      "step": 7259
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.380295468339424e-05,
+      "loss": 1.3513,
+      "step": 7260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.377562952393769e-05,
+      "loss": 1.8,
+      "step": 7261
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.374830560859061e-05,
+      "loss": 1.1109,
+      "step": 7262
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.372098293944825e-05,
+      "loss": 1.327,
+      "step": 7263
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.369366151860574e-05,
+      "loss": 1.0495,
+      "step": 7264
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.366634134815812e-05,
+      "loss": 1.5066,
+      "step": 7265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.36390224302004e-05,
+      "loss": 0.947,
+      "step": 7266
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.361170476682736e-05,
+      "loss": 1.4082,
+      "step": 7267
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.35843883601338e-05,
+      "loss": 1.3188,
+      "step": 7268
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.355707321221437e-05,
+      "loss": 1.7531,
+      "step": 7269
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.352975932516364e-05,
+      "loss": 1.2191,
+      "step": 7270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.350244670107611e-05,
+      "loss": 1.2655,
+      "step": 7271
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.34751353420461e-05,
+      "loss": 1.5245,
+      "step": 7272
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.344782525016793e-05,
+      "loss": 1.5102,
+      "step": 7273
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.342051642753576e-05,
+      "loss": 1.4241,
+      "step": 7274
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.339320887624367e-05,
+      "loss": 1.1411,
+      "step": 7275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.336590259838565e-05,
+      "loss": 1.0241,
+      "step": 7276
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.333859759605564e-05,
+      "loss": 1.3654,
+      "step": 7277
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.331129387134732e-05,
+      "loss": 1.265,
+      "step": 7278
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.328399142635445e-05,
+      "loss": 1.3815,
+      "step": 7279
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.32566902631706e-05,
+      "loss": 1.2328,
+      "step": 7280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.322939038388929e-05,
+      "loss": 1.0472,
+      "step": 7281
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.320209179060395e-05,
+      "loss": 2.0541,
+      "step": 7282
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.317479448540779e-05,
+      "loss": 1.1642,
+      "step": 7283
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.314749847039404e-05,
+      "loss": 1.459,
+      "step": 7284
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.312020374765581e-05,
+      "loss": 1.1932,
+      "step": 7285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.309291031928608e-05,
+      "loss": 1.7075,
+      "step": 7286
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.306561818737784e-05,
+      "loss": 1.1281,
+      "step": 7287
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.303832735402378e-05,
+      "loss": 0.9819,
+      "step": 7288
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.301103782131664e-05,
+      "loss": 1.2121,
+      "step": 7289
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.298374959134905e-05,
+      "loss": 1.4842,
+      "step": 7290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.295646266621348e-05,
+      "loss": 1.1479,
+      "step": 7291
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.292917704800238e-05,
+      "loss": 1.2052,
+      "step": 7292
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.290189273880798e-05,
+      "loss": 1.0288,
+      "step": 7293
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.287460974072255e-05,
+      "loss": 1.4269,
+      "step": 7294
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.284732805583812e-05,
+      "loss": 0.9839,
+      "step": 7295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.282004768624677e-05,
+      "loss": 1.4995,
+      "step": 7296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.279276863404036e-05,
+      "loss": 1.0616,
+      "step": 7297
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.276549090131072e-05,
+      "loss": 1.7125,
+      "step": 7298
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.273821449014948e-05,
+      "loss": 1.2326,
+      "step": 7299
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.27109394026483e-05,
+      "loss": 1.231,
+      "step": 7300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.268366564089865e-05,
+      "loss": 1.4416,
+      "step": 7301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.265639320699193e-05,
+      "loss": 0.831,
+      "step": 7302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.262912210301944e-05,
+      "loss": 1.821,
+      "step": 7303
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.260185233107234e-05,
+      "loss": 2.0553,
+      "step": 7304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.257458389324175e-05,
+      "loss": 1.6652,
+      "step": 7305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.254731679161865e-05,
+      "loss": 1.0162,
+      "step": 7306
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.25200510282939e-05,
+      "loss": 1.1678,
+      "step": 7307
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.249278660535835e-05,
+      "loss": 0.9259,
+      "step": 7308
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.246552352490255e-05,
+      "loss": 1.2373,
+      "step": 7309
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.243826178901718e-05,
+      "loss": 1.3431,
+      "step": 7310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.241100139979267e-05,
+      "loss": 1.3843,
+      "step": 7311
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.238374235931942e-05,
+      "loss": 1.4622,
+      "step": 7312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.235648466968765e-05,
+      "loss": 0.8813,
+      "step": 7313
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.23292283329876e-05,
+      "loss": 1.2835,
+      "step": 7314
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.230197335130923e-05,
+      "loss": 1.1905,
+      "step": 7315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.22747197267425e-05,
+      "loss": 0.695,
+      "step": 7316
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.224746746137732e-05,
+      "loss": 1.696,
+      "step": 7317
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.222021655730343e-05,
+      "loss": 1.5257,
+      "step": 7318
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.219296701661048e-05,
+      "loss": 1.2154,
+      "step": 7319
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.216571884138793e-05,
+      "loss": 1.0196,
+      "step": 7320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.213847203372526e-05,
+      "loss": 1.4333,
+      "step": 7321
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.211122659571181e-05,
+      "loss": 1.3077,
+      "step": 7322
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.208398252943675e-05,
+      "loss": 1.6256,
+      "step": 7323
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.205673983698932e-05,
+      "loss": 1.2771,
+      "step": 7324
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.202949852045838e-05,
+      "loss": 1.1463,
+      "step": 7325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.200225858193292e-05,
+      "loss": 1.4389,
+      "step": 7326
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.19750200235017e-05,
+      "loss": 2.4536,
+      "step": 7327
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.194778284725345e-05,
+      "loss": 1.3061,
+      "step": 7328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.192054705527676e-05,
+      "loss": 1.9716,
+      "step": 7329
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.189331264966006e-05,
+      "loss": 1.7513,
+      "step": 7330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.186607963249178e-05,
+      "loss": 1.4979,
+      "step": 7331
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.183884800586016e-05,
+      "loss": 0.7426,
+      "step": 7332
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.18116177718534e-05,
+      "loss": 0.9124,
+      "step": 7333
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.178438893255951e-05,
+      "loss": 1.5103,
+      "step": 7334
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.175716149006649e-05,
+      "loss": 1.0858,
+      "step": 7335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.17299354464621e-05,
+      "loss": 1.8232,
+      "step": 7336
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.170271080383415e-05,
+      "loss": 0.9377,
+      "step": 7337
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.167548756427024e-05,
+      "loss": 0.7868,
+      "step": 7338
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.164826572985789e-05,
+      "loss": 1.354,
+      "step": 7339
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.162104530268453e-05,
+      "loss": 1.2021,
+      "step": 7340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.159382628483743e-05,
+      "loss": 1.1355,
+      "step": 7341
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.156660867840379e-05,
+      "loss": 1.9807,
+      "step": 7342
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.153939248547073e-05,
+      "loss": 1.7658,
+      "step": 7343
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.15121777081252e-05,
+      "loss": 0.7535,
+      "step": 7344
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.148496434845411e-05,
+      "loss": 1.6798,
+      "step": 7345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.145775240854414e-05,
+      "loss": 1.2861,
+      "step": 7346
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.143054189048202e-05,
+      "loss": 1.4939,
+      "step": 7347
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.140333279635426e-05,
+      "loss": 1.2207,
+      "step": 7348
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.137612512824731e-05,
+      "loss": 1.1452,
+      "step": 7349
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.13489188882475e-05,
+      "loss": 1.3499,
+      "step": 7350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.132171407844106e-05,
+      "loss": 1.1133,
+      "step": 7351
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.129451070091402e-05,
+      "loss": 1.3218,
+      "step": 7352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.126730875775243e-05,
+      "loss": 1.0764,
+      "step": 7353
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.124010825104219e-05,
+      "loss": 1.3395,
+      "step": 7354
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.121290918286907e-05,
+      "loss": 1.7175,
+      "step": 7355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.118571155531875e-05,
+      "loss": 1.2474,
+      "step": 7356
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.115851537047673e-05,
+      "loss": 0.9791,
+      "step": 7357
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.11313206304285e-05,
+      "loss": 1.353,
+      "step": 7358
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.110412733725936e-05,
+      "loss": 0.4958,
+      "step": 7359
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.107693549305456e-05,
+      "loss": 1.2749,
+      "step": 7360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.104974509989924e-05,
+      "loss": 1.0299,
+      "step": 7361
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.102255615987833e-05,
+      "loss": 1.1375,
+      "step": 7362
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.099536867507676e-05,
+      "loss": 1.5856,
+      "step": 7363
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.096818264757933e-05,
+      "loss": 1.5265,
+      "step": 7364
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.094099807947065e-05,
+      "loss": 1.3699,
+      "step": 7365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.091381497283533e-05,
+      "loss": 1.1706,
+      "step": 7366
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.088663332975779e-05,
+      "loss": 1.1999,
+      "step": 7367
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.085945315232233e-05,
+      "loss": 0.9395,
+      "step": 7368
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.083227444261321e-05,
+      "loss": 0.9333,
+      "step": 7369
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.080509720271452e-05,
+      "loss": 1.0247,
+      "step": 7370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.077792143471025e-05,
+      "loss": 1.0562,
+      "step": 7371
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.075074714068429e-05,
+      "loss": 1.775,
+      "step": 7372
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.072357432272038e-05,
+      "loss": 1.3957,
+      "step": 7373
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.06964029829022e-05,
+      "loss": 1.6383,
+      "step": 7374
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.066923312331328e-05,
+      "loss": 1.096,
+      "step": 7375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.064206474603704e-05,
+      "loss": 0.9191,
+      "step": 7376
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.061489785315682e-05,
+      "loss": 0.9429,
+      "step": 7377
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.058773244675579e-05,
+      "loss": 1.4503,
+      "step": 7378
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.056056852891702e-05,
+      "loss": 1.329,
+      "step": 7379
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.053340610172352e-05,
+      "loss": 1.9074,
+      "step": 7380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.050624516725812e-05,
+      "loss": 1.1552,
+      "step": 7381
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.047908572760362e-05,
+      "loss": 1.5725,
+      "step": 7382
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.045192778484256e-05,
+      "loss": 1.1484,
+      "step": 7383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.042477134105746e-05,
+      "loss": 0.9472,
+      "step": 7384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.039761639833078e-05,
+      "loss": 1.1662,
+      "step": 7385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.037046295874476e-05,
+      "loss": 0.962,
+      "step": 7386
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.03433110243816e-05,
+      "loss": 1.3048,
+      "step": 7387
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.031616059732334e-05,
+      "loss": 1.4814,
+      "step": 7388
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.028901167965189e-05,
+      "loss": 1.3604,
+      "step": 7389
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.026186427344908e-05,
+      "loss": 1.8402,
+      "step": 7390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.023471838079659e-05,
+      "loss": 1.6192,
+      "step": 7391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.020757400377607e-05,
+      "loss": 1.2482,
+      "step": 7392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.0180431144469e-05,
+      "loss": 0.9075,
+      "step": 7393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.015328980495666e-05,
+      "loss": 1.3164,
+      "step": 7394
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.012614998732033e-05,
+      "loss": 1.1879,
+      "step": 7395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.009901169364113e-05,
+      "loss": 1.4031,
+      "step": 7396
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.007187492600006e-05,
+      "loss": 1.3324,
+      "step": 7397
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.004473968647804e-05,
+      "loss": 1.4121,
+      "step": 7398
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.001760597715579e-05,
+      "loss": 1.2287,
+      "step": 7399
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.999047380011398e-05,
+      "loss": 1.4966,
+      "step": 7400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.996334315743317e-05,
+      "loss": 0.7848,
+      "step": 7401
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.993621405119376e-05,
+      "loss": 1.012,
+      "step": 7402
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.990908648347606e-05,
+      "loss": 1.4209,
+      "step": 7403
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.988196045636027e-05,
+      "loss": 1.4753,
+      "step": 7404
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.985483597192642e-05,
+      "loss": 1.1374,
+      "step": 7405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.982771303225448e-05,
+      "loss": 1.493,
+      "step": 7406
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.980059163942424e-05,
+      "loss": 1.3233,
+      "step": 7407
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.977347179551547e-05,
+      "loss": 1.231,
+      "step": 7408
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.974635350260773e-05,
+      "loss": 1.7847,
+      "step": 7409
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.971923676278048e-05,
+      "loss": 1.6689,
+      "step": 7410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.969212157811309e-05,
+      "loss": 1.4643,
+      "step": 7411
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.966500795068479e-05,
+      "loss": 0.7515,
+      "step": 7412
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.963789588257469e-05,
+      "loss": 1.5583,
+      "step": 7413
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.961078537586182e-05,
+      "loss": 0.8131,
+      "step": 7414
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.958367643262499e-05,
+      "loss": 0.9322,
+      "step": 7415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.9556569054943e-05,
+      "loss": 1.1754,
+      "step": 7416
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.952946324489446e-05,
+      "loss": 1.0997,
+      "step": 7417
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.950235900455792e-05,
+      "loss": 1.7913,
+      "step": 7418
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.947525633601178e-05,
+      "loss": 1.1166,
+      "step": 7419
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.944815524133425e-05,
+      "loss": 1.7128,
+      "step": 7420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.94210557226035e-05,
+      "loss": 1.3504,
+      "step": 7421
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.939395778189759e-05,
+      "loss": 1.4422,
+      "step": 7422
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.936686142129446e-05,
+      "loss": 1.1243,
+      "step": 7423
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.933976664287183e-05,
+      "loss": 1.2493,
+      "step": 7424
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.931267344870746e-05,
+      "loss": 1.3742,
+      "step": 7425
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.92855818408788e-05,
+      "loss": 1.5229,
+      "step": 7426
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.925849182146332e-05,
+      "loss": 1.0277,
+      "step": 7427
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.92314033925383e-05,
+      "loss": 1.3048,
+      "step": 7428
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.920431655618096e-05,
+      "loss": 0.9544,
+      "step": 7429
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.917723131446838e-05,
+      "loss": 2.1888,
+      "step": 7430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.915014766947742e-05,
+      "loss": 1.2986,
+      "step": 7431
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.912306562328496e-05,
+      "loss": 1.0298,
+      "step": 7432
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.909598517796763e-05,
+      "loss": 1.3564,
+      "step": 7433
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.906890633560206e-05,
+      "loss": 1.2996,
+      "step": 7434
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.90418290982647e-05,
+      "loss": 1.5569,
+      "step": 7435
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.901475346803182e-05,
+      "loss": 0.8875,
+      "step": 7436
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.898767944697964e-05,
+      "loss": 1.1763,
+      "step": 7437
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.896060703718426e-05,
+      "loss": 1.772,
+      "step": 7438
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.893353624072161e-05,
+      "loss": 0.761,
+      "step": 7439
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.890646705966756e-05,
+      "loss": 1.1437,
+      "step": 7440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.887939949609778e-05,
+      "loss": 1.3842,
+      "step": 7441
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.885233355208784e-05,
+      "loss": 1.9056,
+      "step": 7442
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.882526922971322e-05,
+      "loss": 1.3893,
+      "step": 7443
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.879820653104926e-05,
+      "loss": 1.2713,
+      "step": 7444
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.877114545817115e-05,
+      "loss": 1.3001,
+      "step": 7445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.8744086013154e-05,
+      "loss": 1.204,
+      "step": 7446
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.871702819807274e-05,
+      "loss": 1.1784,
+      "step": 7447
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.868997201500221e-05,
+      "loss": 1.5514,
+      "step": 7448
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.866291746601714e-05,
+      "loss": 1.4486,
+      "step": 7449
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.86358645531921e-05,
+      "loss": 1.4146,
+      "step": 7450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.860881327860159e-05,
+      "loss": 1.1736,
+      "step": 7451
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.858176364431982e-05,
+      "loss": 1.5802,
+      "step": 7452
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.855471565242112e-05,
+      "loss": 0.9905,
+      "step": 7453
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.852766930497952e-05,
+      "loss": 1.6548,
+      "step": 7454
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.8500624604069e-05,
+      "loss": 1.4752,
+      "step": 7455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.847358155176336e-05,
+      "loss": 1.2966,
+      "step": 7456
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.844654015013637e-05,
+      "loss": 1.6993,
+      "step": 7457
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.841950040126152e-05,
+      "loss": 1.3435,
+      "step": 7458
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.839246230721224e-05,
+      "loss": 1.4118,
+      "step": 7459
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.836542587006197e-05,
+      "loss": 1.1022,
+      "step": 7460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.833839109188381e-05,
+      "loss": 1.1532,
+      "step": 7461
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.83113579747509e-05,
+      "loss": 1.3879,
+      "step": 7462
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.82843265207361e-05,
+      "loss": 1.006,
+      "step": 7463
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.825729673191228e-05,
+      "loss": 1.0869,
+      "step": 7464
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.823026861035209e-05,
+      "loss": 1.7218,
+      "step": 7465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.820324215812809e-05,
+      "loss": 1.493,
+      "step": 7466
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.817621737731278e-05,
+      "loss": 1.0267,
+      "step": 7467
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.814919426997837e-05,
+      "loss": 1.1763,
+      "step": 7468
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.812217283819708e-05,
+      "loss": 1.2331,
+      "step": 7469
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.809515308404093e-05,
+      "loss": 1.2103,
+      "step": 7470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.806813500958185e-05,
+      "loss": 1.4477,
+      "step": 7471
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.804111861689165e-05,
+      "loss": 1.4578,
+      "step": 7472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.801410390804195e-05,
+      "loss": 1.0511,
+      "step": 7473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.798709088510429e-05,
+      "loss": 1.2124,
+      "step": 7474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.796007955015007e-05,
+      "loss": 0.8403,
+      "step": 7475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.793306990525056e-05,
+      "loss": 1.1814,
+      "step": 7476
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.790606195247692e-05,
+      "loss": 1.847,
+      "step": 7477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.787905569390017e-05,
+      "loss": 1.0836,
+      "step": 7478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.785205113159113e-05,
+      "loss": 1.3419,
+      "step": 7479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.78250482676206e-05,
+      "loss": 0.9974,
+      "step": 7480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.779804710405918e-05,
+      "loss": 0.8133,
+      "step": 7481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.777104764297737e-05,
+      "loss": 1.3275,
+      "step": 7482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.774404988644555e-05,
+      "loss": 1.234,
+      "step": 7483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.771705383653392e-05,
+      "loss": 1.0204,
+      "step": 7484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.769005949531258e-05,
+      "loss": 1.3811,
+      "step": 7485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.76630668648515e-05,
+      "loss": 1.3368,
+      "step": 7486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.76360759472205e-05,
+      "loss": 1.4724,
+      "step": 7487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.760908674448938e-05,
+      "loss": 1.2193,
+      "step": 7488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.758209925872757e-05,
+      "loss": 0.8724,
+      "step": 7489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.75551134920046e-05,
+      "loss": 0.998,
+      "step": 7490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.752812944638974e-05,
+      "loss": 1.2743,
+      "step": 7491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.75011471239522e-05,
+      "loss": 0.8245,
+      "step": 7492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.747416652676102e-05,
+      "loss": 1.0876,
+      "step": 7493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.744718765688514e-05,
+      "loss": 1.1941,
+      "step": 7494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.742021051639327e-05,
+      "loss": 1.353,
+      "step": 7495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.739323510735407e-05,
+      "loss": 1.6186,
+      "step": 7496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.736626143183609e-05,
+      "loss": 1.0043,
+      "step": 7497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.733928949190771e-05,
+      "loss": 1.3082,
+      "step": 7498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.731231928963723e-05,
+      "loss": 1.3518,
+      "step": 7499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.728535082709267e-05,
+      "loss": 1.5703,
+      "step": 7500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.725838410634203e-05,
+      "loss": 1.9999,
+      "step": 7501
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.723141912945318e-05,
+      "loss": 2.4097,
+      "step": 7502
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.720445589849383e-05,
+      "loss": 0.9,
+      "step": 7503
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.717749441553164e-05,
+      "loss": 1.0374,
+      "step": 7504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.715053468263392e-05,
+      "loss": 1.1899,
+      "step": 7505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.712357670186808e-05,
+      "loss": 1.8335,
+      "step": 7506
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.709662047530123e-05,
+      "loss": 0.8463,
+      "step": 7507
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.706966600500047e-05,
+      "loss": 1.4988,
+      "step": 7508
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.704271329303272e-05,
+      "loss": 1.2669,
+      "step": 7509
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.70157623414647e-05,
+      "loss": 1.2171,
+      "step": 7510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.698881315236309e-05,
+      "loss": 1.3336,
+      "step": 7511
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.696186572779438e-05,
+      "loss": 1.1452,
+      "step": 7512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.693492006982494e-05,
+      "loss": 1.4186,
+      "step": 7513
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.690797618052101e-05,
+      "loss": 0.7446,
+      "step": 7514
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.688103406194868e-05,
+      "loss": 1.2869,
+      "step": 7515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.685409371617392e-05,
+      "loss": 1.4237,
+      "step": 7516
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.682715514526256e-05,
+      "loss": 1.1445,
+      "step": 7517
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.680021835128029e-05,
+      "loss": 1.1499,
+      "step": 7518
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.677328333629264e-05,
+      "loss": 1.38,
+      "step": 7519
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.674635010236507e-05,
+      "loss": 1.054,
+      "step": 7520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.671941865156283e-05,
+      "loss": 0.9927,
+      "step": 7521
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.669248898595107e-05,
+      "loss": 1.2602,
+      "step": 7522
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.666556110759481e-05,
+      "loss": 1.1062,
+      "step": 7523
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.66386350185589e-05,
+      "loss": 0.9561,
+      "step": 7524
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.661171072090814e-05,
+      "loss": 1.1768,
+      "step": 7525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.658478821670703e-05,
+      "loss": 0.917,
+      "step": 7526
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.655786750802003e-05,
+      "loss": 1.1352,
+      "step": 7527
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.653094859691154e-05,
+      "loss": 1.0001,
+      "step": 7528
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.65040314854457e-05,
+      "loss": 0.723,
+      "step": 7529
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.647711617568656e-05,
+      "loss": 1.3068,
+      "step": 7530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.64502026696981e-05,
+      "loss": 1.2705,
+      "step": 7531
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.642329096954393e-05,
+      "loss": 1.2992,
+      "step": 7532
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.63963810772878e-05,
+      "loss": 1.4155,
+      "step": 7533
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.636947299499312e-05,
+      "loss": 1.3585,
+      "step": 7534
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.634256672472332e-05,
+      "loss": 0.9056,
+      "step": 7535
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.631566226854163e-05,
+      "loss": 1.2709,
+      "step": 7536
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.628875962851104e-05,
+      "loss": 1.0636,
+      "step": 7537
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.626185880669454e-05,
+      "loss": 1.0991,
+      "step": 7538
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.623495980515489e-05,
+      "loss": 1.4443,
+      "step": 7539
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.620806262595477e-05,
+      "loss": 1.4412,
+      "step": 7540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.61811672711567e-05,
+      "loss": 1.5813,
+      "step": 7541
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.615427374282306e-05,
+      "loss": 2.3064,
+      "step": 7542
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.612738204301606e-05,
+      "loss": 1.48,
+      "step": 7543
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.610049217379782e-05,
+      "loss": 1.0781,
+      "step": 7544
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.607360413723029e-05,
+      "loss": 1.0369,
+      "step": 7545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.604671793537529e-05,
+      "loss": 0.9923,
+      "step": 7546
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.601983357029451e-05,
+      "loss": 1.2627,
+      "step": 7547
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.599295104404944e-05,
+      "loss": 1.4931,
+      "step": 7548
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.596607035870149e-05,
+      "loss": 1.2364,
+      "step": 7549
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.593919151631193e-05,
+      "loss": 1.4418,
+      "step": 7550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.591231451894187e-05,
+      "loss": 1.5957,
+      "step": 7551
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.588543936865228e-05,
+      "loss": 1.1806,
+      "step": 7552
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.585856606750396e-05,
+      "loss": 1.6773,
+      "step": 7553
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.583169461755762e-05,
+      "loss": 0.6374,
+      "step": 7554
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.580482502087378e-05,
+      "loss": 1.3958,
+      "step": 7555
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.577795727951286e-05,
+      "loss": 1.1384,
+      "step": 7556
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.575109139553516e-05,
+      "loss": 1.3759,
+      "step": 7557
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.572422737100073e-05,
+      "loss": 1.2234,
+      "step": 7558
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.569736520796956e-05,
+      "loss": 1.0423,
+      "step": 7559
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.56705049085015e-05,
+      "loss": 1.8068,
+      "step": 7560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.56436464746562e-05,
+      "loss": 1.6309,
+      "step": 7561
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.561678990849331e-05,
+      "loss": 1.5268,
+      "step": 7562
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.55899352120721e-05,
+      "loss": 0.9828,
+      "step": 7563
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.556308238745188e-05,
+      "loss": 1.5035,
+      "step": 7564
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.553623143669178e-05,
+      "loss": 1.7136,
+      "step": 7565
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.550938236185077e-05,
+      "loss": 1.3812,
+      "step": 7566
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.548253516498768e-05,
+      "loss": 1.5238,
+      "step": 7567
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.545568984816121e-05,
+      "loss": 1.5002,
+      "step": 7568
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.542884641342986e-05,
+      "loss": 1.0614,
+      "step": 7569
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.540200486285202e-05,
+      "loss": 1.2262,
+      "step": 7570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.537516519848596e-05,
+      "loss": 1.4129,
+      "step": 7571
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.534832742238979e-05,
+      "loss": 1.4855,
+      "step": 7572
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.532149153662151e-05,
+      "loss": 1.7742,
+      "step": 7573
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.529465754323888e-05,
+      "loss": 0.5077,
+      "step": 7574
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.526782544429958e-05,
+      "loss": 1.5385,
+      "step": 7575
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.524099524186114e-05,
+      "loss": 1.5893,
+      "step": 7576
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.521416693798095e-05,
+      "loss": 1.4555,
+      "step": 7577
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.518734053471625e-05,
+      "loss": 1.5268,
+      "step": 7578
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.51605160341241e-05,
+      "loss": 1.2291,
+      "step": 7579
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.513369343826148e-05,
+      "loss": 1.3927,
+      "step": 7580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.510687274918517e-05,
+      "loss": 1.0947,
+      "step": 7581
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.508005396895182e-05,
+      "loss": 2.5439,
+      "step": 7582
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.505323709961795e-05,
+      "loss": 2.153,
+      "step": 7583
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.50264221432399e-05,
+      "loss": 1.1315,
+      "step": 7584
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.49996091018739e-05,
+      "loss": 1.5242,
+      "step": 7585
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.497279797757598e-05,
+      "loss": 1.0183,
+      "step": 7586
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.494598877240208e-05,
+      "loss": 1.7033,
+      "step": 7587
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.4919181488408e-05,
+      "loss": 1.0922,
+      "step": 7588
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.489237612764932e-05,
+      "loss": 1.5624,
+      "step": 7589
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.486557269218153e-05,
+      "loss": 1.534,
+      "step": 7590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.483877118405996e-05,
+      "loss": 1.1985,
+      "step": 7591
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.481197160533979e-05,
+      "loss": 1.9125,
+      "step": 7592
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.478517395807605e-05,
+      "loss": 1.1947,
+      "step": 7593
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.475837824432366e-05,
+      "loss": 1.1877,
+      "step": 7594
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.473158446613729e-05,
+      "loss": 0.9704,
+      "step": 7595
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.470479262557155e-05,
+      "loss": 1.2275,
+      "step": 7596
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.46780027246809e-05,
+      "loss": 1.385,
+      "step": 7597
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.465121476551965e-05,
+      "loss": 1.2104,
+      "step": 7598
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.462442875014191e-05,
+      "loss": 1.5814,
+      "step": 7599
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.459764468060166e-05,
+      "loss": 1.4707,
+      "step": 7600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.457086255895276e-05,
+      "loss": 1.2279,
+      "step": 7601
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.454408238724886e-05,
+      "loss": 1.2677,
+      "step": 7602
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.451730416754359e-05,
+      "loss": 0.9944,
+      "step": 7603
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.449052790189028e-05,
+      "loss": 0.7765,
+      "step": 7604
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.446375359234225e-05,
+      "loss": 0.5703,
+      "step": 7605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.443698124095251e-05,
+      "loss": 1.3289,
+      "step": 7606
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.441021084977402e-05,
+      "loss": 1.7183,
+      "step": 7607
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.43834424208596e-05,
+      "loss": 0.9393,
+      "step": 7608
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.435667595626186e-05,
+      "loss": 0.8718,
+      "step": 7609
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.432991145803337e-05,
+      "loss": 1.4887,
+      "step": 7610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.430314892822638e-05,
+      "loss": 1.897,
+      "step": 7611
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.427638836889315e-05,
+      "loss": 1.3276,
+      "step": 7612
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.424962978208567e-05,
+      "loss": 1.1486,
+      "step": 7613
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.422287316985585e-05,
+      "loss": 1.3254,
+      "step": 7614
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.419611853425546e-05,
+      "loss": 1.2813,
+      "step": 7615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.416936587733602e-05,
+      "loss": 1.285,
+      "step": 7616
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.4142615201149e-05,
+      "loss": 1.3625,
+      "step": 7617
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.41158665077457e-05,
+      "loss": 1.7632,
+      "step": 7618
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.408911979917723e-05,
+      "loss": 1.2829,
+      "step": 7619
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.406237507749456e-05,
+      "loss": 1.5116,
+      "step": 7620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.403563234474855e-05,
+      "loss": 1.3929,
+      "step": 7621
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.400889160298984e-05,
+      "loss": 1.4985,
+      "step": 7622
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.398215285426897e-05,
+      "loss": 1.6854,
+      "step": 7623
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.395541610063628e-05,
+      "loss": 1.5374,
+      "step": 7624
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.392868134414201e-05,
+      "loss": 1.2962,
+      "step": 7625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.390194858683626e-05,
+      "loss": 1.1018,
+      "step": 7626
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.387521783076885e-05,
+      "loss": 1.2591,
+      "step": 7627
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.384848907798958e-05,
+      "loss": 0.8898,
+      "step": 7628
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.382176233054808e-05,
+      "loss": 0.9036,
+      "step": 7629
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.379503759049373e-05,
+      "loss": 0.9036,
+      "step": 7630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.376831485987594e-05,
+      "loss": 1.0706,
+      "step": 7631
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.37415941407437e-05,
+      "loss": 0.95,
+      "step": 7632
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.371487543514608e-05,
+      "loss": 1.1987,
+      "step": 7633
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.368815874513191e-05,
+      "loss": 1.2896,
+      "step": 7634
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.366144407274986e-05,
+      "loss": 1.8479,
+      "step": 7635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.363473142004844e-05,
+      "loss": 1.5535,
+      "step": 7636
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.360802078907607e-05,
+      "loss": 1.1386,
+      "step": 7637
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.358131218188089e-05,
+      "loss": 1.3936,
+      "step": 7638
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.355460560051095e-05,
+      "loss": 1.2989,
+      "step": 7639
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.352790104701421e-05,
+      "loss": 1.1679,
+      "step": 7640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.350119852343838e-05,
+      "loss": 0.8041,
+      "step": 7641
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.347449803183113e-05,
+      "loss": 1.4258,
+      "step": 7642
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.344779957423976e-05,
+      "loss": 1.5559,
+      "step": 7643
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.342110315271162e-05,
+      "loss": 1.5597,
+      "step": 7644
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.339440876929384e-05,
+      "loss": 1.4306,
+      "step": 7645
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.336771642603334e-05,
+      "loss": 1.276,
+      "step": 7646
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.334102612497703e-05,
+      "loss": 1.1379,
+      "step": 7647
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.331433786817145e-05,
+      "loss": 1.0764,
+      "step": 7648
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.328765165766315e-05,
+      "loss": 1.2298,
+      "step": 7649
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.326096749549846e-05,
+      "loss": 1.7183,
+      "step": 7650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.323428538372355e-05,
+      "loss": 2.1736,
+      "step": 7651
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.320760532438448e-05,
+      "loss": 1.3845,
+      "step": 7652
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.31809273195271e-05,
+      "loss": 1.4454,
+      "step": 7653
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.31542513711971e-05,
+      "loss": 1.6815,
+      "step": 7654
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.312757748144006e-05,
+      "loss": 1.6979,
+      "step": 7655
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.310090565230134e-05,
+      "loss": 1.0285,
+      "step": 7656
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.307423588582621e-05,
+      "loss": 1.4498,
+      "step": 7657
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.304756818405977e-05,
+      "loss": 1.3313,
+      "step": 7658
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.302090254904688e-05,
+      "loss": 1.4035,
+      "step": 7659
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.299423898283233e-05,
+      "loss": 1.0427,
+      "step": 7660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.296757748746073e-05,
+      "loss": 1.1078,
+      "step": 7661
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.294091806497649e-05,
+      "loss": 0.6854,
+      "step": 7662
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.291426071742397e-05,
+      "loss": 1.2026,
+      "step": 7663
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.288760544684722e-05,
+      "loss": 2.0298,
+      "step": 7664
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.286095225529024e-05,
+      "loss": 1.7268,
+      "step": 7665
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.283430114479682e-05,
+      "loss": 1.2632,
+      "step": 7666
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.280765211741063e-05,
+      "loss": 1.5066,
+      "step": 7667
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.278100517517518e-05,
+      "loss": 1.6889,
+      "step": 7668
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.27543603201337e-05,
+      "loss": 1.5042,
+      "step": 7669
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.272771755432946e-05,
+      "loss": 1.1693,
+      "step": 7670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.270107687980543e-05,
+      "loss": 1.448,
+      "step": 7671
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.267443829860446e-05,
+      "loss": 1.1818,
+      "step": 7672
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.264780181276924e-05,
+      "loss": 1.1048,
+      "step": 7673
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.262116742434233e-05,
+      "loss": 1.3932,
+      "step": 7674
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.259453513536601e-05,
+      "loss": 1.0342,
+      "step": 7675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.256790494788254e-05,
+      "loss": 1.6003,
+      "step": 7676
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.254127686393393e-05,
+      "loss": 1.9987,
+      "step": 7677
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.251465088556211e-05,
+      "loss": 1.3652,
+      "step": 7678
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.248802701480883e-05,
+      "loss": 1.4163,
+      "step": 7679
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.246140525371556e-05,
+      "loss": 1.496,
+      "step": 7680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.243478560432373e-05,
+      "loss": 1.6736,
+      "step": 7681
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.240816806867458e-05,
+      "loss": 1.1747,
+      "step": 7682
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.238155264880919e-05,
+      "loss": 1.3383,
+      "step": 7683
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.235493934676847e-05,
+      "loss": 1.0213,
+      "step": 7684
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.232832816459316e-05,
+      "loss": 1.1201,
+      "step": 7685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.230171910432384e-05,
+      "loss": 1.1978,
+      "step": 7686
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.227511216800093e-05,
+      "loss": 1.1285,
+      "step": 7687
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.224850735766473e-05,
+      "loss": 1.5692,
+      "step": 7688
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.222190467535533e-05,
+      "loss": 0.6735,
+      "step": 7689
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.219530412311262e-05,
+      "loss": 1.2441,
+      "step": 7690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.216870570297638e-05,
+      "loss": 1.4782,
+      "step": 7691
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.214210941698624e-05,
+      "loss": 1.8536,
+      "step": 7692
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.211551526718166e-05,
+      "loss": 1.2767,
+      "step": 7693
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.208892325560189e-05,
+      "loss": 1.2371,
+      "step": 7694
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.206233338428605e-05,
+      "loss": 1.5091,
+      "step": 7695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.203574565527309e-05,
+      "loss": 0.7642,
+      "step": 7696
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.200916007060181e-05,
+      "loss": 1.2563,
+      "step": 7697
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.198257663231082e-05,
+      "loss": 1.3196,
+      "step": 7698
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.195599534243858e-05,
+      "loss": 1.5728,
+      "step": 7699
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.192941620302342e-05,
+      "loss": 0.8367,
+      "step": 7700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.19028392161034e-05,
+      "loss": 0.9677,
+      "step": 7701
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.187626438371653e-05,
+      "loss": 1.4566,
+      "step": 7702
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.184969170790062e-05,
+      "loss": 1.2426,
+      "step": 7703
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.182312119069326e-05,
+      "loss": 1.4394,
+      "step": 7704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.179655283413198e-05,
+      "loss": 1.5352,
+      "step": 7705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.1769986640254e-05,
+      "loss": 1.0321,
+      "step": 7706
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.17434226110965e-05,
+      "loss": 1.2818,
+      "step": 7707
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.171686074869646e-05,
+      "loss": 1.1119,
+      "step": 7708
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.169030105509067e-05,
+      "loss": 1.604,
+      "step": 7709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.166374353231578e-05,
+      "loss": 1.5424,
+      "step": 7710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.163718818240827e-05,
+      "loss": 1.176,
+      "step": 7711
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.161063500740442e-05,
+      "loss": 0.7278,
+      "step": 7712
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.158408400934036e-05,
+      "loss": 1.6363,
+      "step": 7713
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.155753519025205e-05,
+      "loss": 0.8667,
+      "step": 7714
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.153098855217535e-05,
+      "loss": 1.0794,
+      "step": 7715
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.150444409714592e-05,
+      "loss": 1.5676,
+      "step": 7716
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.14779018271991e-05,
+      "loss": 1.0723,
+      "step": 7717
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.14513617443703e-05,
+      "loss": 1.1429,
+      "step": 7718
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.142482385069463e-05,
+      "loss": 1.4739,
+      "step": 7719
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.139828814820702e-05,
+      "loss": 1.4793,
+      "step": 7720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.137175463894232e-05,
+      "loss": 1.3241,
+      "step": 7721
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.134522332493514e-05,
+      "loss": 1.2375,
+      "step": 7722
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.131869420821993e-05,
+      "loss": 1.3721,
+      "step": 7723
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.129216729083096e-05,
+      "loss": 1.2902,
+      "step": 7724
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.126564257480243e-05,
+      "loss": 1.4635,
+      "step": 7725
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.123912006216823e-05,
+      "loss": 1.3379,
+      "step": 7726
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.12125997549622e-05,
+      "loss": 1.5852,
+      "step": 7727
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.11860816552179e-05,
+      "loss": 1.203,
+      "step": 7728
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.11595657649688e-05,
+      "loss": 1.44,
+      "step": 7729
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.113305208624818e-05,
+      "loss": 1.4266,
+      "step": 7730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.110654062108916e-05,
+      "loss": 1.3528,
+      "step": 7731
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.108003137152469e-05,
+      "loss": 1.3742,
+      "step": 7732
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.10535243395875e-05,
+      "loss": 0.9673,
+      "step": 7733
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.102701952731021e-05,
+      "loss": 1.7623,
+      "step": 7734
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.100051693672524e-05,
+      "loss": 1.7172,
+      "step": 7735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.09740165698649e-05,
+      "loss": 1.2106,
+      "step": 7736
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.094751842876122e-05,
+      "loss": 1.4544,
+      "step": 7737
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.092102251544614e-05,
+      "loss": 1.0687,
+      "step": 7738
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.089452883195138e-05,
+      "loss": 1.7096,
+      "step": 7739
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.086803738030857e-05,
+      "loss": 1.651,
+      "step": 7740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.084154816254905e-05,
+      "loss": 1.5683,
+      "step": 7741
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.081506118070416e-05,
+      "loss": 1.5664,
+      "step": 7742
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.078857643680482e-05,
+      "loss": 1.3143,
+      "step": 7743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.076209393288198e-05,
+      "loss": 1.4684,
+      "step": 7744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.07356136709664e-05,
+      "loss": 1.0999,
+      "step": 7745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.070913565308859e-05,
+      "loss": 0.7152,
+      "step": 7746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.068265988127893e-05,
+      "loss": 0.8851,
+      "step": 7747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.065618635756768e-05,
+      "loss": 1.4161,
+      "step": 7748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.062971508398475e-05,
+      "loss": 1.7867,
+      "step": 7749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.060324606256009e-05,
+      "loss": 1.2339,
+      "step": 7750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.057677929532331e-05,
+      "loss": 1.1793,
+      "step": 7751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.055031478430398e-05,
+      "loss": 1.3189,
+      "step": 7752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.052385253153148e-05,
+      "loss": 1.152,
+      "step": 7753
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.04973925390349e-05,
+      "loss": 1.1353,
+      "step": 7754
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.047093480884324e-05,
+      "loss": 2.7116,
+      "step": 7755
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.044447934298535e-05,
+      "loss": 0.8145,
+      "step": 7756
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.041802614348984e-05,
+      "loss": 0.9439,
+      "step": 7757
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.039157521238523e-05,
+      "loss": 1.074,
+      "step": 7758
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.036512655169978e-05,
+      "loss": 1.2992,
+      "step": 7759
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.033868016346161e-05,
+      "loss": 1.2568,
+      "step": 7760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.03122360496987e-05,
+      "loss": 0.9322,
+      "step": 7761
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.02857942124388e-05,
+      "loss": 0.4216,
+      "step": 7762
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.025935465370951e-05,
+      "loss": 1.0009,
+      "step": 7763
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.023291737553829e-05,
+      "loss": 0.9577,
+      "step": 7764
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.020648237995235e-05,
+      "loss": 0.9872,
+      "step": 7765
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.018004966897878e-05,
+      "loss": 1.6133,
+      "step": 7766
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.015361924464446e-05,
+      "loss": 1.1962,
+      "step": 7767
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.012719110897617e-05,
+      "loss": 1.1123,
+      "step": 7768
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.010076526400043e-05,
+      "loss": 1.1305,
+      "step": 7769
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.007434171174358e-05,
+      "loss": 0.8419,
+      "step": 7770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.004792045423187e-05,
+      "loss": 1.1767,
+      "step": 7771
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.002150149349128e-05,
+      "loss": 1.6164,
+      "step": 7772
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.999508483154768e-05,
+      "loss": 1.5904,
+      "step": 7773
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.99686704704268e-05,
+      "loss": 1.664,
+      "step": 7774
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.994225841215398e-05,
+      "loss": 1.2201,
+      "step": 7775
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.991584865875466e-05,
+      "loss": 0.9249,
+      "step": 7776
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.988944121225396e-05,
+      "loss": 1.0106,
+      "step": 7777
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.986303607467683e-05,
+      "loss": 1.4107,
+      "step": 7778
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.983663324804809e-05,
+      "loss": 1.3058,
+      "step": 7779
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.981023273439226e-05,
+      "loss": 1.0515,
+      "step": 7780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.978383453573386e-05,
+      "loss": 1.1645,
+      "step": 7781
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.975743865409707e-05,
+      "loss": 0.9888,
+      "step": 7782
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.973104509150604e-05,
+      "loss": 1.0326,
+      "step": 7783
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.970465384998464e-05,
+      "loss": 1.1882,
+      "step": 7784
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.96782649315566e-05,
+      "loss": 1.0874,
+      "step": 7785
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.965187833824545e-05,
+      "loss": 1.6053,
+      "step": 7786
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.962549407207452e-05,
+      "loss": 1.5387,
+      "step": 7787
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.959911213506704e-05,
+      "loss": 1.6276,
+      "step": 7788
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.957273252924599e-05,
+      "loss": 0.8599,
+      "step": 7789
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.954635525663429e-05,
+      "loss": 1.1805,
+      "step": 7790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.951998031925447e-05,
+      "loss": 1.3589,
+      "step": 7791
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.949360771912903e-05,
+      "loss": 0.8414,
+      "step": 7792
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.946723745828028e-05,
+      "loss": 2.5076,
+      "step": 7793
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.944086953873035e-05,
+      "loss": 1.2844,
+      "step": 7794
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.941450396250115e-05,
+      "loss": 1.6843,
+      "step": 7795
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.938814073161441e-05,
+      "loss": 1.1383,
+      "step": 7796
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.936177984809174e-05,
+      "loss": 1.1474,
+      "step": 7797
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.933542131395452e-05,
+      "loss": 1.7898,
+      "step": 7798
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.930906513122396e-05,
+      "loss": 1.4491,
+      "step": 7799
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.92827113019211e-05,
+      "loss": 1.1801,
+      "step": 7800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.92563598280668e-05,
+      "loss": 1.0223,
+      "step": 7801
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.923001071168169e-05,
+      "loss": 1.4026,
+      "step": 7802
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.920366395478629e-05,
+      "loss": 0.8972,
+      "step": 7803
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.91773195594009e-05,
+      "loss": 1.4527,
+      "step": 7804
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.915097752754567e-05,
+      "loss": 1.6587,
+      "step": 7805
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.912463786124055e-05,
+      "loss": 1.2503,
+      "step": 7806
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.909830056250527e-05,
+      "loss": 1.1637,
+      "step": 7807
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.907196563335942e-05,
+      "loss": 1.1584,
+      "step": 7808
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.904563307582244e-05,
+      "loss": 1.1806,
+      "step": 7809
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.90193028919135e-05,
+      "loss": 1.0635,
+      "step": 7810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.899297508365173e-05,
+      "loss": 1.2006,
+      "step": 7811
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.896664965305584e-05,
+      "loss": 1.5357,
+      "step": 7812
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.89403266021446e-05,
+      "loss": 1.5771,
+      "step": 7813
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.89140059329365e-05,
+      "loss": 1.9559,
+      "step": 7814
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.888768764744984e-05,
+      "loss": 1.5574,
+      "step": 7815
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.886137174770277e-05,
+      "loss": 1.5026,
+      "step": 7816
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.883505823571317e-05,
+      "loss": 0.7018,
+      "step": 7817
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.880874711349884e-05,
+      "loss": 1.0619,
+      "step": 7818
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.878243838307734e-05,
+      "loss": 1.6517,
+      "step": 7819
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.875613204646608e-05,
+      "loss": 1.6105,
+      "step": 7820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.872982810568226e-05,
+      "loss": 1.0678,
+      "step": 7821
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.870352656274298e-05,
+      "loss": 1.0521,
+      "step": 7822
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.867722741966496e-05,
+      "loss": 1.7351,
+      "step": 7823
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.86509306784649e-05,
+      "loss": 1.1888,
+      "step": 7824
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.862463634115932e-05,
+      "loss": 1.1214,
+      "step": 7825
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.859834440976444e-05,
+      "loss": 1.1437,
+      "step": 7826
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.857205488629646e-05,
+      "loss": 1.3144,
+      "step": 7827
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.85457677727712e-05,
+      "loss": 1.3059,
+      "step": 7828
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.851948307120444e-05,
+      "loss": 1.2972,
+      "step": 7829
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.849320078361174e-05,
+      "loss": 1.5655,
+      "step": 7830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.846692091200845e-05,
+      "loss": 1.3127,
+      "step": 7831
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.84406434584098e-05,
+      "loss": 1.1975,
+      "step": 7832
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.841436842483068e-05,
+      "loss": 1.2397,
+      "step": 7833
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.8388095813286e-05,
+      "loss": 1.3376,
+      "step": 7834
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.836182562579031e-05,
+      "loss": 0.8948,
+      "step": 7835
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.833555786435812e-05,
+      "loss": 1.6421,
+      "step": 7836
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.830929253100364e-05,
+      "loss": 0.7945,
+      "step": 7837
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.828302962774096e-05,
+      "loss": 1.1034,
+      "step": 7838
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.82567691565839e-05,
+      "loss": 1.521,
+      "step": 7839
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.823051111954621e-05,
+      "loss": 1.2062,
+      "step": 7840
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.820425551864137e-05,
+      "loss": 1.4629,
+      "step": 7841
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.817800235588272e-05,
+      "loss": 1.4041,
+      "step": 7842
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.815175163328342e-05,
+      "loss": 1.2248,
+      "step": 7843
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.812550335285635e-05,
+      "loss": 0.9951,
+      "step": 7844
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.809925751661427e-05,
+      "loss": 2.091,
+      "step": 7845
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.80730141265698e-05,
+      "loss": 0.7089,
+      "step": 7846
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.804677318473529e-05,
+      "loss": 1.4045,
+      "step": 7847
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.802053469312299e-05,
+      "loss": 0.7953,
+      "step": 7848
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.799429865374483e-05,
+      "loss": 1.3417,
+      "step": 7849
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.796806506861265e-05,
+      "loss": 1.211,
+      "step": 7850
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.794183393973809e-05,
+      "loss": 1.1881,
+      "step": 7851
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.791560526913261e-05,
+      "loss": 1.0018,
+      "step": 7852
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.788937905880744e-05,
+      "loss": 1.3879,
+      "step": 7853
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.786315531077372e-05,
+      "loss": 1.743,
+      "step": 7854
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.783693402704221e-05,
+      "loss": 1.2436,
+      "step": 7855
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.781071520962365e-05,
+      "loss": 1.3134,
+      "step": 7856
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.778449886052853e-05,
+      "loss": 1.4819,
+      "step": 7857
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.775828498176717e-05,
+      "loss": 1.4066,
+      "step": 7858
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.773207357534976e-05,
+      "loss": 1.0427,
+      "step": 7859
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.77058646432861e-05,
+      "loss": 1.6081,
+      "step": 7860
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.767965818758602e-05,
+      "loss": 1.3804,
+      "step": 7861
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.765345421025902e-05,
+      "loss": 0.7984,
+      "step": 7862
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.762725271331449e-05,
+      "loss": 1.1694,
+      "step": 7863
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.760105369876162e-05,
+      "loss": 1.0074,
+      "step": 7864
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.757485716860934e-05,
+      "loss": 1.721,
+      "step": 7865
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.754866312486646e-05,
+      "loss": 1.5351,
+      "step": 7866
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.752247156954159e-05,
+      "loss": 1.561,
+      "step": 7867
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.749628250464314e-05,
+      "loss": 1.2823,
+      "step": 7868
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.747009593217933e-05,
+      "loss": 1.3991,
+      "step": 7869
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.744391185415816e-05,
+      "loss": 1.6885,
+      "step": 7870
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.741773027258749e-05,
+      "loss": 1.4019,
+      "step": 7871
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.739155118947495e-05,
+      "loss": 0.9713,
+      "step": 7872
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.7365374606828e-05,
+      "loss": 1.2767,
+      "step": 7873
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.73392005266539e-05,
+      "loss": 1.4432,
+      "step": 7874
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.731302895095972e-05,
+      "loss": 1.1266,
+      "step": 7875
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.728685988175234e-05,
+      "loss": 1.0573,
+      "step": 7876
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.726069332103841e-05,
+      "loss": 1.319,
+      "step": 7877
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.723452927082446e-05,
+      "loss": 1.5321,
+      "step": 7878
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.720836773311678e-05,
+      "loss": 1.2698,
+      "step": 7879
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.718220870992149e-05,
+      "loss": 1.5595,
+      "step": 7880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.715605220324449e-05,
+      "loss": 1.0761,
+      "step": 7881
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.712989821509147e-05,
+      "loss": 1.2067,
+      "step": 7882
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.7103746747468e-05,
+      "loss": 0.9774,
+      "step": 7883
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.707759780237938e-05,
+      "loss": 1.4066,
+      "step": 7884
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.705145138183084e-05,
+      "loss": 1.4027,
+      "step": 7885
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.702530748782722e-05,
+      "loss": 0.8817,
+      "step": 7886
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.699916612237328e-05,
+      "loss": 0.7987,
+      "step": 7887
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.697302728747363e-05,
+      "loss": 1.133,
+      "step": 7888
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.694689098513263e-05,
+      "loss": 1.4577,
+      "step": 7889
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.692075721735444e-05,
+      "loss": 0.9046,
+      "step": 7890
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.689462598614306e-05,
+      "loss": 0.7415,
+      "step": 7891
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.686849729350225e-05,
+      "loss": 1.2689,
+      "step": 7892
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.684237114143556e-05,
+      "loss": 1.7053,
+      "step": 7893
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.68162475319464e-05,
+      "loss": 1.2114,
+      "step": 7894
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.679012646703804e-05,
+      "loss": 1.2276,
+      "step": 7895
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.676400794871345e-05,
+      "loss": 0.6564,
+      "step": 7896
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.673789197897538e-05,
+      "loss": 1.4687,
+      "step": 7897
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.671177855982649e-05,
+      "loss": 1.0654,
+      "step": 7898
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.668566769326918e-05,
+      "loss": 1.5341,
+      "step": 7899
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.665955938130571e-05,
+      "loss": 1.6893,
+      "step": 7900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.663345362593807e-05,
+      "loss": 2.0401,
+      "step": 7901
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.660735042916809e-05,
+      "loss": 0.9347,
+      "step": 7902
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.658124979299738e-05,
+      "loss": 1.5639,
+      "step": 7903
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.655515171942742e-05,
+      "loss": 1.4121,
+      "step": 7904
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.652905621045944e-05,
+      "loss": 0.9772,
+      "step": 7905
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.65029632680945e-05,
+      "loss": 2.2232,
+      "step": 7906
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.64768728943334e-05,
+      "loss": 1.3829,
+      "step": 7907
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.645078509117682e-05,
+      "loss": 1.4046,
+      "step": 7908
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.64246998606252e-05,
+      "loss": 1.1997,
+      "step": 7909
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.639861720467879e-05,
+      "loss": 1.2542,
+      "step": 7910
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.637253712533769e-05,
+      "loss": 0.9743,
+      "step": 7911
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.634645962460173e-05,
+      "loss": 0.7843,
+      "step": 7912
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.632038470447055e-05,
+      "loss": 1.6055,
+      "step": 7913
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.629431236694364e-05,
+      "loss": 1.311,
+      "step": 7914
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.626824261402027e-05,
+      "loss": 1.4053,
+      "step": 7915
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.624217544769951e-05,
+      "loss": 1.1185,
+      "step": 7916
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.621611086998025e-05,
+      "loss": 1.2896,
+      "step": 7917
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.619004888286107e-05,
+      "loss": 1.1669,
+      "step": 7918
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.616398948834054e-05,
+      "loss": 1.13,
+      "step": 7919
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.613793268841689e-05,
+      "loss": 2.6098,
+      "step": 7920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.611187848508821e-05,
+      "loss": 1.0577,
+      "step": 7921
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.608582688035243e-05,
+      "loss": 1.4093,
+      "step": 7922
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.60597778762071e-05,
+      "loss": 1.9025,
+      "step": 7923
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.603373147464978e-05,
+      "loss": 1.0462,
+      "step": 7924
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.600768767767771e-05,
+      "loss": 1.102,
+      "step": 7925
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.598164648728802e-05,
+      "loss": 1.4359,
+      "step": 7926
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.595560790547757e-05,
+      "loss": 1.1429,
+      "step": 7927
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.592957193424305e-05,
+      "loss": 0.854,
+      "step": 7928
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.590353857558088e-05,
+      "loss": 0.6851,
+      "step": 7929
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.587750783148739e-05,
+      "loss": 1.3998,
+      "step": 7930
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.585147970395862e-05,
+      "loss": 1.0792,
+      "step": 7931
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.582545419499044e-05,
+      "loss": 1.7438,
+      "step": 7932
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.579943130657864e-05,
+      "loss": 1.6589,
+      "step": 7933
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.577341104071856e-05,
+      "loss": 1.1621,
+      "step": 7934
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.574739339940552e-05,
+      "loss": 0.9735,
+      "step": 7935
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.57213783846346e-05,
+      "loss": 1.1501,
+      "step": 7936
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.569536599840065e-05,
+      "loss": 1.1441,
+      "step": 7937
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.566935624269838e-05,
+      "loss": 0.9919,
+      "step": 7938
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.564334911952224e-05,
+      "loss": 1.0964,
+      "step": 7939
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.561734463086647e-05,
+      "loss": 1.6253,
+      "step": 7940
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.559134277872514e-05,
+      "loss": 1.9717,
+      "step": 7941
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.556534356509215e-05,
+      "loss": 1.3327,
+      "step": 7942
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.553934699196111e-05,
+      "loss": 1.2665,
+      "step": 7943
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.551335306132552e-05,
+      "loss": 1.3915,
+      "step": 7944
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.54873617751786e-05,
+      "loss": 1.4151,
+      "step": 7945
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.546137313551343e-05,
+      "loss": 1.0661,
+      "step": 7946
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.543538714432282e-05,
+      "loss": 1.1031,
+      "step": 7947
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.540940380359945e-05,
+      "loss": 1.5733,
+      "step": 7948
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.538342311533578e-05,
+      "loss": 1.348,
+      "step": 7949
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.535744508152401e-05,
+      "loss": 1.6202,
+      "step": 7950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.533146970415617e-05,
+      "loss": 1.2164,
+      "step": 7951
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.530549698522412e-05,
+      "loss": 1.3329,
+      "step": 7952
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.527952692671946e-05,
+      "loss": 1.642,
+      "step": 7953
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.525355953063368e-05,
+      "loss": 1.2696,
+      "step": 7954
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.522759479895788e-05,
+      "loss": 1.6378,
+      "step": 7955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.520163273368317e-05,
+      "loss": 1.4094,
+      "step": 7956
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.517567333680034e-05,
+      "loss": 0.9996,
+      "step": 7957
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.514971661029999e-05,
+      "loss": 1.3554,
+      "step": 7958
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.512376255617255e-05,
+      "loss": 1.3378,
+      "step": 7959
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.509781117640817e-05,
+      "loss": 1.0378,
+      "step": 7960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.507186247299683e-05,
+      "loss": 1.2918,
+      "step": 7961
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.504591644792834e-05,
+      "loss": 1.3519,
+      "step": 7962
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.501997310319233e-05,
+      "loss": 1.8963,
+      "step": 7963
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.49940324407781e-05,
+      "loss": 1.206,
+      "step": 7964
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.49680944626749e-05,
+      "loss": 0.7855,
+      "step": 7965
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.494215917087161e-05,
+      "loss": 1.1212,
+      "step": 7966
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.491622656735703e-05,
+      "loss": 1.3659,
+      "step": 7967
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.48902966541197e-05,
+      "loss": 1.4015,
+      "step": 7968
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.486436943314794e-05,
+      "loss": 0.997,
+      "step": 7969
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.483844490642998e-05,
+      "loss": 1.7342,
+      "step": 7970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.481252307595367e-05,
+      "loss": 1.003,
+      "step": 7971
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.478660394370674e-05,
+      "loss": 1.4103,
+      "step": 7972
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.476068751167675e-05,
+      "loss": 1.0328,
+      "step": 7973
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.473477378185095e-05,
+      "loss": 0.8564,
+      "step": 7974
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.470886275621653e-05,
+      "loss": 1.6638,
+      "step": 7975
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.46829544367603e-05,
+      "loss": 1.6269,
+      "step": 7976
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.465704882546899e-05,
+      "loss": 1.6567,
+      "step": 7977
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.46311459243291e-05,
+      "loss": 1.3987,
+      "step": 7978
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.460524573532687e-05,
+      "loss": 0.9605,
+      "step": 7979
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.457934826044837e-05,
+      "loss": 1.3451,
+      "step": 7980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.455345350167949e-05,
+      "loss": 1.7836,
+      "step": 7981
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.452756146100585e-05,
+      "loss": 0.9084,
+      "step": 7982
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.450167214041289e-05,
+      "loss": 0.8438,
+      "step": 7983
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.447578554188585e-05,
+      "loss": 0.793,
+      "step": 7984
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.444990166740977e-05,
+      "loss": 0.8297,
+      "step": 7985
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.442402051896944e-05,
+      "loss": 1.0835,
+      "step": 7986
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.439814209854947e-05,
+      "loss": 1.0974,
+      "step": 7987
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.437226640813427e-05,
+      "loss": 1.3756,
+      "step": 7988
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.4346393449708e-05,
+      "loss": 1.6253,
+      "step": 7989
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.432052322525467e-05,
+      "loss": 1.5666,
+      "step": 7990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.429465573675808e-05,
+      "loss": 1.341,
+      "step": 7991
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.42687909862017e-05,
+      "loss": 1.1348,
+      "step": 7992
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.42429289755689e-05,
+      "loss": 1.1275,
+      "step": 7993
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.421706970684288e-05,
+      "loss": 1.5629,
+      "step": 7994
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.41912131820065e-05,
+      "loss": 0.9708,
+      "step": 7995
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.416535940304256e-05,
+      "loss": 1.1303,
+      "step": 7996
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.413950837193347e-05,
+      "loss": 0.9632,
+      "step": 7997
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.411366009066157e-05,
+      "loss": 1.48,
+      "step": 7998
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.408781456120896e-05,
+      "loss": 1.3034,
+      "step": 7999
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.406197178555747e-05,
+      "loss": 1.1706,
+      "step": 8000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.403613176568882e-05,
+      "loss": 1.4259,
+      "step": 8001
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.401029450358446e-05,
+      "loss": 1.322,
+      "step": 8002
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.39844600012256e-05,
+      "loss": 1.3293,
+      "step": 8003
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.395862826059326e-05,
+      "loss": 0.8769,
+      "step": 8004
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.393279928366826e-05,
+      "loss": 0.8025,
+      "step": 8005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.390697307243123e-05,
+      "loss": 1.0985,
+      "step": 8006
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.388114962886258e-05,
+      "loss": 2.3979,
+      "step": 8007
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.385532895494245e-05,
+      "loss": 1.556,
+      "step": 8008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.382951105265079e-05,
+      "loss": 0.8312,
+      "step": 8009
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.38036959239674e-05,
+      "loss": 1.5397,
+      "step": 8010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.377788357087182e-05,
+      "loss": 1.2172,
+      "step": 8011
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.37520739953434e-05,
+      "loss": 1.2694,
+      "step": 8012
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.37262671993612e-05,
+      "loss": 1.3573,
+      "step": 8013
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.370046318490414e-05,
+      "loss": 1.3177,
+      "step": 8014
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.367466195395094e-05,
+      "loss": 1.6357,
+      "step": 8015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.364886350848006e-05,
+      "loss": 1.4937,
+      "step": 8016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.362306785046976e-05,
+      "loss": 1.2265,
+      "step": 8017
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.359727498189814e-05,
+      "loss": 1.5443,
+      "step": 8018
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.357148490474296e-05,
+      "loss": 1.9432,
+      "step": 8019
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.354569762098189e-05,
+      "loss": 2.1203,
+      "step": 8020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.351991313259228e-05,
+      "loss": 1.5255,
+      "step": 8021
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.349413144155141e-05,
+      "loss": 1.1358,
+      "step": 8022
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.346835254983623e-05,
+      "loss": 1.789,
+      "step": 8023
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.344257645942347e-05,
+      "loss": 1.1693,
+      "step": 8024
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.341680317228971e-05,
+      "loss": 1.1597,
+      "step": 8025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.339103269041127e-05,
+      "loss": 1.3042,
+      "step": 8026
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.336526501576427e-05,
+      "loss": 1.592,
+      "step": 8027
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.333950015032468e-05,
+      "loss": 1.3327,
+      "step": 8028
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.331373809606809e-05,
+      "loss": 1.0389,
+      "step": 8029
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.328797885496997e-05,
+      "loss": 0.9891,
+      "step": 8030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.326222242900568e-05,
+      "loss": 1.1619,
+      "step": 8031
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.323646882015017e-05,
+      "loss": 1.3805,
+      "step": 8032
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.321071803037832e-05,
+      "loss": 1.3366,
+      "step": 8033
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.318497006166475e-05,
+      "loss": 1.1602,
+      "step": 8034
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.315922491598381e-05,
+      "loss": 1.2298,
+      "step": 8035
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.313348259530967e-05,
+      "loss": 1.4328,
+      "step": 8036
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.31077431016163e-05,
+      "loss": 1.4153,
+      "step": 8037
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.308200643687746e-05,
+      "loss": 2.0405,
+      "step": 8038
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.305627260306672e-05,
+      "loss": 1.428,
+      "step": 8039
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.303054160215731e-05,
+      "loss": 1.2756,
+      "step": 8040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.300481343612234e-05,
+      "loss": 1.2938,
+      "step": 8041
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.297908810693472e-05,
+      "loss": 1.7571,
+      "step": 8042
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.295336561656706e-05,
+      "loss": 1.9515,
+      "step": 8043
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.292764596699187e-05,
+      "loss": 1.2509,
+      "step": 8044
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.290192916018129e-05,
+      "loss": 1.1875,
+      "step": 8045
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.287621519810738e-05,
+      "loss": 0.7568,
+      "step": 8046
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.28505040827419e-05,
+      "loss": 1.5548,
+      "step": 8047
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.282479581605641e-05,
+      "loss": 1.1609,
+      "step": 8048
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.279909040002233e-05,
+      "loss": 1.4797,
+      "step": 8049
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.277338783661068e-05,
+      "loss": 1.527,
+      "step": 8050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.274768812779243e-05,
+      "loss": 1.5381,
+      "step": 8051
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.272199127553828e-05,
+      "loss": 1.3674,
+      "step": 8052
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.26962972818187e-05,
+      "loss": 1.4259,
+      "step": 8053
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.267060614860393e-05,
+      "loss": 1.6187,
+      "step": 8054
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.264491787786404e-05,
+      "loss": 1.3415,
+      "step": 8055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.26192324715688e-05,
+      "loss": 1.1189,
+      "step": 8056
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.259354993168782e-05,
+      "loss": 1.004,
+      "step": 8057
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.256787026019049e-05,
+      "loss": 1.0439,
+      "step": 8058
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.254219345904594e-05,
+      "loss": 1.31,
+      "step": 8059
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.251651953022317e-05,
+      "loss": 1.3729,
+      "step": 8060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.24908484756908e-05,
+      "loss": 1.4727,
+      "step": 8061
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.24651802974174e-05,
+      "loss": 0.8609,
+      "step": 8062
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.243951499737121e-05,
+      "loss": 1.606,
+      "step": 8063
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.241385257752029e-05,
+      "loss": 1.577,
+      "step": 8064
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.238819303983253e-05,
+      "loss": 1.426,
+      "step": 8065
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.236253638627545e-05,
+      "loss": 0.9762,
+      "step": 8066
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.233688261881649e-05,
+      "loss": 1.2353,
+      "step": 8067
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.231123173942278e-05,
+      "loss": 1.1632,
+      "step": 8068
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.228558375006133e-05,
+      "loss": 1.8332,
+      "step": 8069
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.225993865269884e-05,
+      "loss": 1.2913,
+      "step": 8070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.223429644930183e-05,
+      "loss": 1.2924,
+      "step": 8071
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.220865714183655e-05,
+      "loss": 1.0274,
+      "step": 8072
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.218302073226908e-05,
+      "loss": 1.2879,
+      "step": 8073
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.215738722256525e-05,
+      "loss": 0.7913,
+      "step": 8074
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.213175661469066e-05,
+      "loss": 1.0332,
+      "step": 8075
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.210612891061079e-05,
+      "loss": 1.0369,
+      "step": 8076
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.208050411229072e-05,
+      "loss": 1.0673,
+      "step": 8077
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.205488222169541e-05,
+      "loss": 0.8833,
+      "step": 8078
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.202926324078961e-05,
+      "loss": 1.2847,
+      "step": 8079
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.200364717153782e-05,
+      "loss": 1.4027,
+      "step": 8080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.197803401590433e-05,
+      "loss": 1.523,
+      "step": 8081
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.195242377585316e-05,
+      "loss": 1.3613,
+      "step": 8082
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.192681645334817e-05,
+      "loss": 0.9964,
+      "step": 8083
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.190121205035295e-05,
+      "loss": 1.8835,
+      "step": 8084
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.187561056883091e-05,
+      "loss": 1.3264,
+      "step": 8085
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.185001201074521e-05,
+      "loss": 1.1949,
+      "step": 8086
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.182441637805876e-05,
+      "loss": 1.0944,
+      "step": 8087
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.179882367273429e-05,
+      "loss": 1.5283,
+      "step": 8088
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.177323389673426e-05,
+      "loss": 0.922,
+      "step": 8089
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.174764705202098e-05,
+      "loss": 1.4774,
+      "step": 8090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.172206314055645e-05,
+      "loss": 1.7463,
+      "step": 8091
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.169648216430253e-05,
+      "loss": 1.2415,
+      "step": 8092
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.167090412522075e-05,
+      "loss": 1.1377,
+      "step": 8093
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.164532902527249e-05,
+      "loss": 1.1263,
+      "step": 8094
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.161975686641891e-05,
+      "loss": 1.2766,
+      "step": 8095
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.15941876506209e-05,
+      "loss": 1.3039,
+      "step": 8096
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.15686213798392e-05,
+      "loss": 1.1716,
+      "step": 8097
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.154305805603417e-05,
+      "loss": 1.4834,
+      "step": 8098
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.151749768116612e-05,
+      "loss": 1.5921,
+      "step": 8099
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.149194025719505e-05,
+      "loss": 0.9621,
+      "step": 8100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.146638578608072e-05,
+      "loss": 1.0845,
+      "step": 8101
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.144083426978273e-05,
+      "loss": 0.7139,
+      "step": 8102
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.141528571026035e-05,
+      "loss": 0.9881,
+      "step": 8103
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.138974010947271e-05,
+      "loss": 1.29,
+      "step": 8104
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.136419746937865e-05,
+      "loss": 1.4324,
+      "step": 8105
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.133865779193689e-05,
+      "loss": 1.0317,
+      "step": 8106
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.131312107910581e-05,
+      "loss": 1.5673,
+      "step": 8107
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.128758733284363e-05,
+      "loss": 1.026,
+      "step": 8108
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.126205655510828e-05,
+      "loss": 0.8253,
+      "step": 8109
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.12365287478575e-05,
+      "loss": 1.3625,
+      "step": 8110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.121100391304883e-05,
+      "loss": 1.9435,
+      "step": 8111
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.118548205263952e-05,
+      "loss": 1.5796,
+      "step": 8112
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.115996316858669e-05,
+      "loss": 1.4291,
+      "step": 8113
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.11344472628471e-05,
+      "loss": 1.6081,
+      "step": 8114
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.110893433737735e-05,
+      "loss": 1.1014,
+      "step": 8115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.108342439413383e-05,
+      "loss": 1.6194,
+      "step": 8116
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.10579174350727e-05,
+      "loss": 1.4932,
+      "step": 8117
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.1032413462149854e-05,
+      "loss": 0.9569,
+      "step": 8118
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.100691247732097e-05,
+      "loss": 1.1536,
+      "step": 8119
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.09814144825415e-05,
+      "loss": 1.5843,
+      "step": 8120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0955919479766674e-05,
+      "loss": 1.2405,
+      "step": 8121
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.09304274709515e-05,
+      "loss": 1.9337,
+      "step": 8122
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.090493845805072e-05,
+      "loss": 1.295,
+      "step": 8123
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0879452443018913e-05,
+      "loss": 0.7548,
+      "step": 8124
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.085396942781034e-05,
+      "loss": 1.2784,
+      "step": 8125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0828489414379094e-05,
+      "loss": 1.3095,
+      "step": 8126
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.080301240467903e-05,
+      "loss": 1.5933,
+      "step": 8127
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0777538400663744e-05,
+      "loss": 0.9891,
+      "step": 8128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0752067404286664e-05,
+      "loss": 0.9895,
+      "step": 8129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0726599417500886e-05,
+      "loss": 1.7609,
+      "step": 8130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.070113444225937e-05,
+      "loss": 0.6967,
+      "step": 8131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.067567248051479e-05,
+      "loss": 1.5011,
+      "step": 8132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.065021353421963e-05,
+      "loss": 1.5787,
+      "step": 8133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0624757605326145e-05,
+      "loss": 0.8207,
+      "step": 8134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.059930469578626e-05,
+      "loss": 0.8094,
+      "step": 8135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0573854807551766e-05,
+      "loss": 1.0376,
+      "step": 8136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0548407942574236e-05,
+      "loss": 1.3645,
+      "step": 8137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0522964102804956e-05,
+      "loss": 1.1716,
+      "step": 8138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.049752329019504e-05,
+      "loss": 1.4225,
+      "step": 8139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.047208550669523e-05,
+      "loss": 1.1148,
+      "step": 8140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0446650754256204e-05,
+      "loss": 1.6611,
+      "step": 8141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.042121903482832e-05,
+      "loss": 0.9477,
+      "step": 8142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0395790350361716e-05,
+      "loss": 1.3628,
+      "step": 8143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0370364702806305e-05,
+      "loss": 1.356,
+      "step": 8144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.034494209411183e-05,
+      "loss": 1.1037,
+      "step": 8145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.031952252622763e-05,
+      "loss": 0.9146,
+      "step": 8146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.029410600110296e-05,
+      "loss": 1.3171,
+      "step": 8147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.02686925206868e-05,
+      "loss": 0.2195,
+      "step": 8148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0243282086927885e-05,
+      "loss": 1.1835,
+      "step": 8149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0217874701774766e-05,
+      "loss": 1.5013,
+      "step": 8150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0192470367175665e-05,
+      "loss": 1.4859,
+      "step": 8151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0167069085078656e-05,
+      "loss": 0.708,
+      "step": 8152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.014167085743153e-05,
+      "loss": 1.0042,
+      "step": 8153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0116275686181876e-05,
+      "loss": 1.1635,
+      "step": 8154
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.009088357327706e-05,
+      "loss": 1.0783,
+      "step": 8155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0065494520664125e-05,
+      "loss": 1.3435,
+      "step": 8156
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.004010853028999e-05,
+      "loss": 1.1064,
+      "step": 8157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0014725604101264e-05,
+      "loss": 1.0222,
+      "step": 8158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.998934574404437e-05,
+      "loss": 1.515,
+      "step": 8159
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.996396895206546e-05,
+      "loss": 1.6435,
+      "step": 8160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.993859523011051e-05,
+      "loss": 1.3322,
+      "step": 8161
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.991322458012514e-05,
+      "loss": 1.3413,
+      "step": 8162
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.988785700405486e-05,
+      "loss": 1.5429,
+      "step": 8163
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.986249250384486e-05,
+      "loss": 1.1888,
+      "step": 8164
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.983713108144018e-05,
+      "loss": 1.3156,
+      "step": 8165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9811772738785544e-05,
+      "loss": 1.1068,
+      "step": 8166
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.978641747782545e-05,
+      "loss": 1.2698,
+      "step": 8167
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.976106530050421e-05,
+      "loss": 1.3422,
+      "step": 8168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9735716208765834e-05,
+      "loss": 1.1882,
+      "step": 8169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.971037020455414e-05,
+      "loss": 0.824,
+      "step": 8170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.968502728981278e-05,
+      "loss": 1.2427,
+      "step": 8171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.965968746648495e-05,
+      "loss": 1.2859,
+      "step": 8172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.96343507365138e-05,
+      "loss": 1.3635,
+      "step": 8173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.960901710184221e-05,
+      "loss": 0.6334,
+      "step": 8174
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.95836865644128e-05,
+      "loss": 1.3107,
+      "step": 8175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.955835912616799e-05,
+      "loss": 0.8915,
+      "step": 8176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.953303478904982e-05,
+      "loss": 1.3103,
+      "step": 8177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.950771355500029e-05,
+      "loss": 1.1101,
+      "step": 8178
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9482395425961015e-05,
+      "loss": 1.2175,
+      "step": 8179
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.945708040387346e-05,
+      "loss": 1.5324,
+      "step": 8180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.943176849067881e-05,
+      "loss": 1.5034,
+      "step": 8181
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.940645968831807e-05,
+      "loss": 1.2811,
+      "step": 8182
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.938115399873188e-05,
+      "loss": 1.3274,
+      "step": 8183
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.935585142386075e-05,
+      "loss": 0.8764,
+      "step": 8184
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.933055196564492e-05,
+      "loss": 1.0944,
+      "step": 8185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.930525562602439e-05,
+      "loss": 1.2935,
+      "step": 8186
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.927996240693894e-05,
+      "loss": 1.3005,
+      "step": 8187
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.925467231032806e-05,
+      "loss": 0.5087,
+      "step": 8188
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.922938533813106e-05,
+      "loss": 2.0661,
+      "step": 8189
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.920410149228697e-05,
+      "loss": 1.2748,
+      "step": 8190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.91788207747346e-05,
+      "loss": 1.4044,
+      "step": 8191
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.915354318741252e-05,
+      "loss": 1.5522,
+      "step": 8192
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.912826873225903e-05,
+      "loss": 1.2389,
+      "step": 8193
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9102997411212234e-05,
+      "loss": 1.0712,
+      "step": 8194
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9077729226209965e-05,
+      "loss": 1.8131,
+      "step": 8195
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.905246417918984e-05,
+      "loss": 1.887,
+      "step": 8196
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.90272022720892e-05,
+      "loss": 1.2908,
+      "step": 8197
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9001943506845216e-05,
+      "loss": 0.9584,
+      "step": 8198
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.89766878853947e-05,
+      "loss": 1.1225,
+      "step": 8199
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.8951435409674324e-05,
+      "loss": 1.1684,
+      "step": 8200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.8926186081620495e-05,
+      "loss": 1.3743,
+      "step": 8201
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.890093990316936e-05,
+      "loss": 0.7584,
+      "step": 8202
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.887569687625685e-05,
+      "loss": 1.4853,
+      "step": 8203
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.885045700281859e-05,
+      "loss": 1.3803,
+      "step": 8204
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.882522028479007e-05,
+      "loss": 1.849,
+      "step": 8205
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.879998672410645e-05,
+      "loss": 1.0282,
+      "step": 8206
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.8774756322702685e-05,
+      "loss": 0.2922,
+      "step": 8207
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.874952908251352e-05,
+      "loss": 1.3985,
+      "step": 8208
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.872430500547333e-05,
+      "loss": 1.1199,
+      "step": 8209
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8699084093516386e-05,
+      "loss": 1.4705,
+      "step": 8210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.867386634857666e-05,
+      "loss": 1.3359,
+      "step": 8211
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.864865177258789e-05,
+      "loss": 0.8882,
+      "step": 8212
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.862344036748357e-05,
+      "loss": 1.2328,
+      "step": 8213
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.859823213519701e-05,
+      "loss": 1.9007,
+      "step": 8214
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.85730270776611e-05,
+      "loss": 1.2319,
+      "step": 8215
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.854782519680866e-05,
+      "loss": 1.2919,
+      "step": 8216
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8522626494572196e-05,
+      "loss": 0.7092,
+      "step": 8217
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.849743097288398e-05,
+      "loss": 1.0635,
+      "step": 8218
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8472238633676105e-05,
+      "loss": 1.2682,
+      "step": 8219
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.844704947888028e-05,
+      "loss": 1.4769,
+      "step": 8220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.842186351042807e-05,
+      "loss": 0.883,
+      "step": 8221
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8396680730250775e-05,
+      "loss": 1.1857,
+      "step": 8222
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.837150114027946e-05,
+      "loss": 0.8508,
+      "step": 8223
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.834632474244497e-05,
+      "loss": 1.2423,
+      "step": 8224
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.832115153867778e-05,
+      "loss": 0.8708,
+      "step": 8225
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.82959815309082e-05,
+      "loss": 1.7035,
+      "step": 8226
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.827081472106644e-05,
+      "loss": 0.9373,
+      "step": 8227
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.824565111108222e-05,
+      "loss": 1.3187,
+      "step": 8228
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.822049070288521e-05,
+      "loss": 1.0372,
+      "step": 8229
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.819533349840465e-05,
+      "loss": 1.0065,
+      "step": 8230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.817017949956967e-05,
+      "loss": 0.985,
+      "step": 8231
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8145028708309127e-05,
+      "loss": 1.1102,
+      "step": 8232
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.811988112655162e-05,
+      "loss": 1.0378,
+      "step": 8233
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8094736756225496e-05,
+      "loss": 1.4469,
+      "step": 8234
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.806959559925891e-05,
+      "loss": 1.2534,
+      "step": 8235
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.804445765757964e-05,
+      "loss": 0.9178,
+      "step": 8236
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.801932293311536e-05,
+      "loss": 1.038,
+      "step": 8237
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.799419142779342e-05,
+      "loss": 1.5052,
+      "step": 8238
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.796906314354093e-05,
+      "loss": 1.8699,
+      "step": 8239
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.79439380822848e-05,
+      "loss": 0.6262,
+      "step": 8240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.791881624595164e-05,
+      "loss": 0.9663,
+      "step": 8241
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7893697636467824e-05,
+      "loss": 1.0084,
+      "step": 8242
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.78685822557595e-05,
+      "loss": 0.8806,
+      "step": 8243
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.784347010575253e-05,
+      "loss": 1.0086,
+      "step": 8244
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.781836118837263e-05,
+      "loss": 1.4087,
+      "step": 8245
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.779325550554506e-05,
+      "loss": 1.4691,
+      "step": 8246
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.776815305919505e-05,
+      "loss": 1.0722,
+      "step": 8247
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.774305385124746e-05,
+      "loss": 1.3544,
+      "step": 8248
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.771795788362696e-05,
+      "loss": 0.9931,
+      "step": 8249
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.769286515825794e-05,
+      "loss": 1.3235,
+      "step": 8250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7667775677064526e-05,
+      "loss": 1.294,
+      "step": 8251
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7642689441970644e-05,
+      "loss": 1.2599,
+      "step": 8252
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7617606454899944e-05,
+      "loss": 1.0231,
+      "step": 8253
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.759252671777582e-05,
+      "loss": 1.1268,
+      "step": 8254
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.756745023252142e-05,
+      "loss": 1.0866,
+      "step": 8255
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.754237700105969e-05,
+      "loss": 1.0417,
+      "step": 8256
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.751730702531322e-05,
+      "loss": 1.2658,
+      "step": 8257
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.749224030720442e-05,
+      "loss": 1.1429,
+      "step": 8258
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.746717684865548e-05,
+      "loss": 1.0808,
+      "step": 8259
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7442116651588274e-05,
+      "loss": 1.3484,
+      "step": 8260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7417059717924524e-05,
+      "loss": 1.177,
+      "step": 8261
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.739200604958553e-05,
+      "loss": 1.4182,
+      "step": 8262
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.736695564849246e-05,
+      "loss": 1.5809,
+      "step": 8263
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.734190851656629e-05,
+      "loss": 0.8909,
+      "step": 8264
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7316864655727644e-05,
+      "loss": 1.2999,
+      "step": 8265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7291824067896946e-05,
+      "loss": 1.628,
+      "step": 8266
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7266786754994287e-05,
+      "loss": 1.2542,
+      "step": 8267
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.724175271893958e-05,
+      "loss": 1.0193,
+      "step": 8268
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7216721961652485e-05,
+      "loss": 1.3374,
+      "step": 8269
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7191694485052415e-05,
+      "loss": 0.9399,
+      "step": 8270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.71666702910585e-05,
+      "loss": 1.305,
+      "step": 8271
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.714164938158967e-05,
+      "loss": 1.2693,
+      "step": 8272
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.711663175856449e-05,
+      "loss": 1.1464,
+      "step": 8273
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7091617423901386e-05,
+      "loss": 0.8484,
+      "step": 8274
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.70666063795185e-05,
+      "loss": 1.0845,
+      "step": 8275
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7041598627333737e-05,
+      "loss": 1.7081,
+      "step": 8276
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7016594169264683e-05,
+      "loss": 1.5146,
+      "step": 8277
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.699159300722876e-05,
+      "loss": 1.2289,
+      "step": 8278
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.696659514314308e-05,
+      "loss": 1.2485,
+      "step": 8279
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6941600578924506e-05,
+      "loss": 1.3062,
+      "step": 8280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6916609316489674e-05,
+      "loss": 1.1078,
+      "step": 8281
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.689162135775499e-05,
+      "loss": 1.6009,
+      "step": 8282
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.686663670463649e-05,
+      "loss": 1.4238,
+      "step": 8283
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6841655359050075e-05,
+      "loss": 0.7783,
+      "step": 8284
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.681667732291134e-05,
+      "loss": 1.6764,
+      "step": 8285
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6791702598135656e-05,
+      "loss": 1.3203,
+      "step": 8286
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.676673118663811e-05,
+      "loss": 1.3775,
+      "step": 8287
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.674176309033357e-05,
+      "loss": 1.1484,
+      "step": 8288
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.671679831113659e-05,
+      "loss": 1.0477,
+      "step": 8289
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6691836850961544e-05,
+      "loss": 1.7044,
+      "step": 8290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6666878711722494e-05,
+      "loss": 1.4447,
+      "step": 8291
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.664192389533327e-05,
+      "loss": 1.6573,
+      "step": 8292
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6616972403707504e-05,
+      "loss": 1.9267,
+      "step": 8293
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.65920242387584e-05,
+      "loss": 1.2224,
+      "step": 8294
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6567079402399094e-05,
+      "loss": 2.5275,
+      "step": 8295
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.654213789654239e-05,
+      "loss": 1.1905,
+      "step": 8296
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.651719972310081e-05,
+      "loss": 1.2462,
+      "step": 8297
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6492264883986734e-05,
+      "loss": 0.9842,
+      "step": 8298
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6467333381112084e-05,
+      "loss": 0.8176,
+      "step": 8299
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6442405216388716e-05,
+      "loss": 1.28,
+      "step": 8300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6417480391728105e-05,
+      "loss": 1.3211,
+      "step": 8301
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6392558909041605e-05,
+      "loss": 1.4546,
+      "step": 8302
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.63676407702402e-05,
+      "loss": 1.5209,
+      "step": 8303
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.634272597723469e-05,
+      "loss": 1.5971,
+      "step": 8304
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6317814531935495e-05,
+      "loss": 1.1919,
+      "step": 8305
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.629290643625289e-05,
+      "loss": 1.2734,
+      "step": 8306
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.626800169209689e-05,
+      "loss": 1.6408,
+      "step": 8307
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6243100301377227e-05,
+      "loss": 1.6316,
+      "step": 8308
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.621820226600341e-05,
+      "loss": 1.5617,
+      "step": 8309
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.619330758788455e-05,
+      "loss": 1.296,
+      "step": 8310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.61684162689297e-05,
+      "loss": 1.8284,
+      "step": 8311
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.614352831104753e-05,
+      "loss": 1.2283,
+      "step": 8312
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.61186437161465e-05,
+      "loss": 1.4983,
+      "step": 8313
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6093762486134805e-05,
+      "loss": 1.1309,
+      "step": 8314
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6068884622920346e-05,
+      "loss": 0.6832,
+      "step": 8315
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6044010128410826e-05,
+      "loss": 1.5652,
+      "step": 8316
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6019139004513656e-05,
+      "loss": 1.277,
+      "step": 8317
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.599427125313599e-05,
+      "loss": 1.3601,
+      "step": 8318
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.596940687618474e-05,
+      "loss": 0.9829,
+      "step": 8319
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.59445458755665e-05,
+      "loss": 1.5016,
+      "step": 8320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5919688253187655e-05,
+      "loss": 1.1838,
+      "step": 8321
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5894834010954364e-05,
+      "loss": 2.0036,
+      "step": 8322
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.586998315077245e-05,
+      "loss": 1.5258,
+      "step": 8323
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.584513567454754e-05,
+      "loss": 0.6703,
+      "step": 8324
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.582029158418498e-05,
+      "loss": 1.091,
+      "step": 8325
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.579545088158984e-05,
+      "loss": 1.5233,
+      "step": 8326
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.577061356866693e-05,
+      "loss": 1.3203,
+      "step": 8327
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.574577964732084e-05,
+      "loss": 0.8494,
+      "step": 8328
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.572094911945587e-05,
+      "loss": 1.2917,
+      "step": 8329
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5696121986976066e-05,
+      "loss": 0.9087,
+      "step": 8330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.567129825178519e-05,
+      "loss": 1.412,
+      "step": 8331
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.564647791578674e-05,
+      "loss": 1.3673,
+      "step": 8332
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.562166098088404e-05,
+      "loss": 1.27,
+      "step": 8333
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.559684744898004e-05,
+      "loss": 1.0234,
+      "step": 8334
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5572037321977556e-05,
+      "loss": 1.4287,
+      "step": 8335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.554723060177895e-05,
+      "loss": 0.649,
+      "step": 8336
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5522427290286515e-05,
+      "loss": 1.3187,
+      "step": 8337
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.549762738940214e-05,
+      "loss": 1.0579,
+      "step": 8338
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5472830901027615e-05,
+      "loss": 0.9324,
+      "step": 8339
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5448037827064325e-05,
+      "loss": 0.831,
+      "step": 8340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.542324816941349e-05,
+      "loss": 1.2685,
+      "step": 8341
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5398461929975906e-05,
+      "loss": 1.3138,
+      "step": 8342
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.537367911065232e-05,
+      "loss": 1.022,
+      "step": 8343
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.534889971334306e-05,
+      "loss": 0.9733,
+      "step": 8344
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5324123739948275e-05,
+      "loss": 1.4043,
+      "step": 8345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5299351192367864e-05,
+      "loss": 0.9802,
+      "step": 8346
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5274582072501336e-05,
+      "loss": 1.5374,
+      "step": 8347
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5249816382248064e-05,
+      "loss": 1.2941,
+      "step": 8348
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.522505412350712e-05,
+      "loss": 1.1527,
+      "step": 8349
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.520029529817733e-05,
+      "loss": 0.7836,
+      "step": 8350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5175539908157203e-05,
+      "loss": 0.9913,
+      "step": 8351
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5150787955345054e-05,
+      "loss": 1.4817,
+      "step": 8352
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5126039441638886e-05,
+      "loss": 0.4656,
+      "step": 8353
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.510129436893645e-05,
+      "loss": 0.8758,
+      "step": 8354
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.507655273913524e-05,
+      "loss": 1.7606,
+      "step": 8355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.505181455413252e-05,
+      "loss": 1.2595,
+      "step": 8356
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.502707981582518e-05,
+      "loss": 1.0855,
+      "step": 8357
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5002348526109936e-05,
+      "loss": 0.9065,
+      "step": 8358
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.497762068688324e-05,
+      "loss": 1.1531,
+      "step": 8359
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.495289630004126e-05,
+      "loss": 1.3817,
+      "step": 8360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.4928175367479896e-05,
+      "loss": 1.0398,
+      "step": 8361
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.490345789109479e-05,
+      "loss": 1.4862,
+      "step": 8362
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.48787438727813e-05,
+      "loss": 1.1815,
+      "step": 8363
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.485403331443455e-05,
+      "loss": 1.1141,
+      "step": 8364
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.482932621794938e-05,
+      "loss": 1.1304,
+      "step": 8365
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.4804622585220376e-05,
+      "loss": 0.8774,
+      "step": 8366
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.4779922418141874e-05,
+      "loss": 1.1941,
+      "step": 8367
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.475522571860785e-05,
+      "loss": 1.5461,
+      "step": 8368
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.473053248851213e-05,
+      "loss": 1.407,
+      "step": 8369
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.4705842729748214e-05,
+      "loss": 2.1232,
+      "step": 8370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.468115644420937e-05,
+      "loss": 1.2385,
+      "step": 8371
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.465647363378859e-05,
+      "loss": 0.8556,
+      "step": 8372
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.4631794300378535e-05,
+      "loss": 1.0678,
+      "step": 8373
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.4607118445871696e-05,
+      "loss": 0.9088,
+      "step": 8374
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.4582446072160235e-05,
+      "loss": 1.0802,
+      "step": 8375
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.4557777181136036e-05,
+      "loss": 1.4094,
+      "step": 8376
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.4533111774690836e-05,
+      "loss": 1.338,
+      "step": 8377
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.4508449854716006e-05,
+      "loss": 1.3564,
+      "step": 8378
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.448379142310258e-05,
+      "loss": 1.5548,
+      "step": 8379
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.445913648174146e-05,
+      "loss": 1.4269,
+      "step": 8380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.44344850325232e-05,
+      "loss": 1.0112,
+      "step": 8381
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.440983707733811e-05,
+      "loss": 1.4643,
+      "step": 8382
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.438519261807631e-05,
+      "loss": 1.0986,
+      "step": 8383
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.436055165662746e-05,
+      "loss": 1.3022,
+      "step": 8384
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.433591419488111e-05,
+      "loss": 1.4802,
+      "step": 8385
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.4311280234726506e-05,
+      "loss": 1.1274,
+      "step": 8386
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.428664977805262e-05,
+      "loss": 1.2149,
+      "step": 8387
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.426202282674815e-05,
+      "loss": 0.9794,
+      "step": 8388
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.4237399382701514e-05,
+      "loss": 1.2061,
+      "step": 8389
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.42127794478009e-05,
+      "loss": 1.4412,
+      "step": 8390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.418816302393419e-05,
+      "loss": 1.5969,
+      "step": 8391
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.416355011298901e-05,
+      "loss": 1.2423,
+      "step": 8392
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.413894071685272e-05,
+      "loss": 1.4698,
+      "step": 8393
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.4114334837412426e-05,
+      "loss": 1.7973,
+      "step": 8394
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.408973247655489e-05,
+      "loss": 1.7375,
+      "step": 8395
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.4065133636166686e-05,
+      "loss": 1.5072,
+      "step": 8396
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.404053831813409e-05,
+      "loss": 1.0295,
+      "step": 8397
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.401594652434312e-05,
+      "loss": 1.1535,
+      "step": 8398
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.399135825667949e-05,
+      "loss": 1.1547,
+      "step": 8399
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.396677351702869e-05,
+      "loss": 1.4958,
+      "step": 8400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.39421923072759e-05,
+      "loss": 1.5064,
+      "step": 8401
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.391761462930605e-05,
+      "loss": 0.825,
+      "step": 8402
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.3893040485003784e-05,
+      "loss": 1.3078,
+      "step": 8403
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.386846987625355e-05,
+      "loss": 1.2203,
+      "step": 8404
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.384390280493935e-05,
+      "loss": 1.7642,
+      "step": 8405
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.381933927294509e-05,
+      "loss": 0.9695,
+      "step": 8406
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.3794779282154316e-05,
+      "loss": 1.22,
+      "step": 8407
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.377022283445034e-05,
+      "loss": 0.8485,
+      "step": 8408
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.374566993171622e-05,
+      "loss": 1.2982,
+      "step": 8409
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.372112057583465e-05,
+      "loss": 1.0528,
+      "step": 8410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.3696574768688125e-05,
+      "loss": 1.2818,
+      "step": 8411
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.367203251215886e-05,
+      "loss": 1.3219,
+      "step": 8412
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.3647493808128777e-05,
+      "loss": 0.9775,
+      "step": 8413
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.3622958658479595e-05,
+      "loss": 0.919,
+      "step": 8414
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.359842706509271e-05,
+      "loss": 1.5274,
+      "step": 8415
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.357389902984917e-05,
+      "loss": 1.3057,
+      "step": 8416
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.354937455462987e-05,
+      "loss": 1.1263,
+      "step": 8417
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.352485364131537e-05,
+      "loss": 1.0986,
+      "step": 8418
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.350033629178597e-05,
+      "loss": 1.2657,
+      "step": 8419
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.3475822507921735e-05,
+      "loss": 1.4041,
+      "step": 8420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.345131229160235e-05,
+      "loss": 1.1995,
+      "step": 8421
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.342680564470733e-05,
+      "loss": 1.5423,
+      "step": 8422
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.3402302569115895e-05,
+      "loss": 0.7176,
+      "step": 8423
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.3377803066706966e-05,
+      "loss": 2.1336,
+      "step": 8424
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.3353307139359196e-05,
+      "loss": 1.6696,
+      "step": 8425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.332881478895098e-05,
+      "loss": 0.8872,
+      "step": 8426
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.330432601736044e-05,
+      "loss": 0.8014,
+      "step": 8427
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.3279840826465375e-05,
+      "loss": 1.8282,
+      "step": 8428
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.3255359218143394e-05,
+      "loss": 0.8722,
+      "step": 8429
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.323088119427175e-05,
+      "loss": 1.5077,
+      "step": 8430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.32064067567275e-05,
+      "loss": 1.4018,
+      "step": 8431
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.318193590738731e-05,
+      "loss": 0.9388,
+      "step": 8432
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.3157468648127696e-05,
+      "loss": 1.0992,
+      "step": 8433
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.313300498082481e-05,
+      "loss": 1.8684,
+      "step": 8434
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.3108544907354583e-05,
+      "loss": 1.1911,
+      "step": 8435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.308408842959265e-05,
+      "loss": 1.2614,
+      "step": 8436
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.3059635549414374e-05,
+      "loss": 1.1187,
+      "step": 8437
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.3035186268694834e-05,
+      "loss": 1.2748,
+      "step": 8438
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.301074058930885e-05,
+      "loss": 0.7113,
+      "step": 8439
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.2986298513130925e-05,
+      "loss": 1.0989,
+      "step": 8440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.296186004203538e-05,
+      "loss": 1.4685,
+      "step": 8441
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.293742517789612e-05,
+      "loss": 1.486,
+      "step": 8442
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.2912993922586864e-05,
+      "loss": 0.9778,
+      "step": 8443
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.288856627798104e-05,
+      "loss": 1.5615,
+      "step": 8444
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.286414224595182e-05,
+      "loss": 1.2232,
+      "step": 8445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.283972182837209e-05,
+      "loss": 1.308,
+      "step": 8446
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.2815305027114395e-05,
+      "loss": 1.4846,
+      "step": 8447
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.279089184405106e-05,
+      "loss": 1.1096,
+      "step": 8448
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.276648228105414e-05,
+      "loss": 1.0291,
+      "step": 8449
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.274207633999539e-05,
+      "loss": 0.9281,
+      "step": 8450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.271767402274628e-05,
+      "loss": 1.5917,
+      "step": 8451
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.2693275331178116e-05,
+      "loss": 1.171,
+      "step": 8452
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.2668880267161705e-05,
+      "loss": 0.7503,
+      "step": 8453
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.264448883256774e-05,
+      "loss": 1.1067,
+      "step": 8454
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.26201010292666e-05,
+      "loss": 1.027,
+      "step": 8455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.259571685912837e-05,
+      "loss": 1.5605,
+      "step": 8456
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2571336324022914e-05,
+      "loss": 1.5513,
+      "step": 8457
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2546959425819676e-05,
+      "loss": 1.8222,
+      "step": 8458
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2522586166387967e-05,
+      "loss": 1.0165,
+      "step": 8459
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.249821654759677e-05,
+      "loss": 1.0843,
+      "step": 8460
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.247385057131477e-05,
+      "loss": 0.8461,
+      "step": 8461
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.244948823941038e-05,
+      "loss": 1.2613,
+      "step": 8462
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2425129553751774e-05,
+      "loss": 0.8136,
+      "step": 8463
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2400774516206785e-05,
+      "loss": 1.3369,
+      "step": 8464
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.237642312864299e-05,
+      "loss": 1.4892,
+      "step": 8465
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.235207539292772e-05,
+      "loss": 1.0675,
+      "step": 8466
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2327731310927975e-05,
+      "loss": 1.0684,
+      "step": 8467
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.230339088451054e-05,
+      "loss": 1.1616,
+      "step": 8468
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.227905411554179e-05,
+      "loss": 1.0551,
+      "step": 8469
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.225472100588796e-05,
+      "loss": 0.9509,
+      "step": 8470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2230391557414936e-05,
+      "loss": 1.539,
+      "step": 8471
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2206065771988355e-05,
+      "loss": 1.6099,
+      "step": 8472
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.218174365147357e-05,
+      "loss": 1.3932,
+      "step": 8473
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2157425197735543e-05,
+      "loss": 1.5345,
+      "step": 8474
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.213311041263915e-05,
+      "loss": 1.3026,
+      "step": 8475
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.210879929804886e-05,
+      "loss": 1.607,
+      "step": 8476
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2084491855828884e-05,
+      "loss": 1.1626,
+      "step": 8477
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2060188087843185e-05,
+      "loss": 1.2745,
+      "step": 8478
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.203588799595534e-05,
+      "loss": 1.5941,
+      "step": 8479
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.201159158202876e-05,
+      "loss": 1.3346,
+      "step": 8480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1987298847926524e-05,
+      "loss": 1.6334,
+      "step": 8481
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.196300979551144e-05,
+      "loss": 0.2366,
+      "step": 8482
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.193872442664603e-05,
+      "loss": 1.5921,
+      "step": 8483
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.191444274319257e-05,
+      "loss": 1.4177,
+      "step": 8484
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.189016474701294e-05,
+      "loss": 1.3722,
+      "step": 8485
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.186589043996884e-05,
+      "loss": 1.8798,
+      "step": 8486
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.184161982392168e-05,
+      "loss": 1.2094,
+      "step": 8487
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.181735290073253e-05,
+      "loss": 1.3862,
+      "step": 8488
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.179308967226233e-05,
+      "loss": 1.401,
+      "step": 8489
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.176883014037148e-05,
+      "loss": 1.2536,
+      "step": 8490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.17445743069203e-05,
+      "loss": 1.3629,
+      "step": 8491
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.172032217376876e-05,
+      "loss": 1.58,
+      "step": 8492
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1696073742776555e-05,
+      "loss": 1.2834,
+      "step": 8493
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.167182901580312e-05,
+      "loss": 1.2569,
+      "step": 8494
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1647587994707505e-05,
+      "loss": 1.3619,
+      "step": 8495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.162335068134858e-05,
+      "loss": 1.0776,
+      "step": 8496
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1599117077584916e-05,
+      "loss": 1.3942,
+      "step": 8497
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.157488718527478e-05,
+      "loss": 1.4237,
+      "step": 8498
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.155066100627615e-05,
+      "loss": 1.3428,
+      "step": 8499
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.152643854244671e-05,
+      "loss": 1.3475,
+      "step": 8500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1502219795643894e-05,
+      "loss": 1.1464,
+      "step": 8501
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.147800476772484e-05,
+      "loss": 1.186,
+      "step": 8502
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.145379346054638e-05,
+      "loss": 1.4435,
+      "step": 8503
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.142958587596508e-05,
+      "loss": 1.1781,
+      "step": 8504
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1405382015837244e-05,
+      "loss": 1.6865,
+      "step": 8505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1381181882018806e-05,
+      "loss": 1.3998,
+      "step": 8506
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1356985476365484e-05,
+      "loss": 1.1508,
+      "step": 8507
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.133279280073271e-05,
+      "loss": 0.8677,
+      "step": 8508
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.13086038569756e-05,
+      "loss": 1.6184,
+      "step": 8509
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.128441864694904e-05,
+      "loss": 1.2036,
+      "step": 8510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.12602371725075e-05,
+      "loss": 1.314,
+      "step": 8511
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1236059435505334e-05,
+      "loss": 0.9911,
+      "step": 8512
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.121188543779649e-05,
+      "loss": 1.5636,
+      "step": 8513
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1187715181234705e-05,
+      "loss": 1.4114,
+      "step": 8514
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1163548667673386e-05,
+      "loss": 1.1902,
+      "step": 8515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.113938589896562e-05,
+      "loss": 1.6628,
+      "step": 8516
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.111522687696425e-05,
+      "loss": 1.7631,
+      "step": 8517
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.109107160352185e-05,
+      "loss": 1.4179,
+      "step": 8518
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.106692008049065e-05,
+      "loss": 1.1144,
+      "step": 8519
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1042772309722675e-05,
+      "loss": 0.9729,
+      "step": 8520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.10186282930696e-05,
+      "loss": 1.2609,
+      "step": 8521
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.09944880323828e-05,
+      "loss": 1.4369,
+      "step": 8522
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0970351529513386e-05,
+      "loss": 1.1071,
+      "step": 8523
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0946218786312184e-05,
+      "loss": 0.693,
+      "step": 8524
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.092208980462976e-05,
+      "loss": 0.8786,
+      "step": 8525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.089796458631633e-05,
+      "loss": 1.324,
+      "step": 8526
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0873843133221855e-05,
+      "loss": 1.3421,
+      "step": 8527
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.084972544719603e-05,
+      "loss": 1.4569,
+      "step": 8528
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0825611530088214e-05,
+      "loss": 1.9404,
+      "step": 8529
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0801501383747505e-05,
+      "loss": 1.6026,
+      "step": 8530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0777395010022743e-05,
+      "loss": 0.997,
+      "step": 8531
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0753292410762365e-05,
+      "loss": 1.1988,
+      "step": 8532
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.072919358781463e-05,
+      "loss": 0.7941,
+      "step": 8533
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0705098543027474e-05,
+      "loss": 1.5193,
+      "step": 8534
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.068100727824855e-05,
+      "loss": 1.0323,
+      "step": 8535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.065691979532521e-05,
+      "loss": 1.2891,
+      "step": 8536
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.06328360961045e-05,
+      "loss": 1.3151,
+      "step": 8537
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0608756182433226e-05,
+      "loss": 1.4046,
+      "step": 8538
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.058468005615786e-05,
+      "loss": 1.4694,
+      "step": 8539
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0560607719124586e-05,
+      "loss": 0.8301,
+      "step": 8540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.053653917317931e-05,
+      "loss": 0.7701,
+      "step": 8541
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.051247442016769e-05,
+      "loss": 1.048,
+      "step": 8542
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.048841346193498e-05,
+      "loss": 1.3984,
+      "step": 8543
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.046435630032622e-05,
+      "loss": 1.4762,
+      "step": 8544
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.044030293718618e-05,
+      "loss": 1.0242,
+      "step": 8545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.04162533743593e-05,
+      "loss": 0.7935,
+      "step": 8546
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.039220761368977e-05,
+      "loss": 1.2973,
+      "step": 8547
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.036816565702138e-05,
+      "loss": 1.0649,
+      "step": 8548
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.03441275061977e-05,
+      "loss": 1.3621,
+      "step": 8549
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0320093163062096e-05,
+      "loss": 1.4478,
+      "step": 8550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.029606262945753e-05,
+      "loss": 1.7342,
+      "step": 8551
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.027203590722671e-05,
+      "loss": 1.1359,
+      "step": 8552
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0248012998212e-05,
+      "loss": 1.0242,
+      "step": 8553
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.022399390425552e-05,
+      "loss": 1.8661,
+      "step": 8554
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.01999786271991e-05,
+      "loss": 1.4439,
+      "step": 8555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0175967168884273e-05,
+      "loss": 1.65,
+      "step": 8556
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.015195953115228e-05,
+      "loss": 1.4011,
+      "step": 8557
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.012795571584409e-05,
+      "loss": 1.5891,
+      "step": 8558
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0103955724800276e-05,
+      "loss": 1.1254,
+      "step": 8559
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.007995955986124e-05,
+      "loss": 1.34,
+      "step": 8560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.005596722286705e-05,
+      "loss": 0.9938,
+      "step": 8561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.003197871565746e-05,
+      "loss": 1.3544,
+      "step": 8562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.000799404007194e-05,
+      "loss": 1.2675,
+      "step": 8563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9984013197949686e-05,
+      "loss": 0.9698,
+      "step": 8564
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.996003619112959e-05,
+      "loss": 1.3021,
+      "step": 8565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.993606302145023e-05,
+      "loss": 0.8371,
+      "step": 8566
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9912093690749914e-05,
+      "loss": 1.3356,
+      "step": 8567
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.988812820086669e-05,
+      "loss": 1.227,
+      "step": 8568
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.986416655363818e-05,
+      "loss": 1.5941,
+      "step": 8569
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.984020875090184e-05,
+      "loss": 1.3878,
+      "step": 8570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.981625479449481e-05,
+      "loss": 1.735,
+      "step": 8571
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.979230468625389e-05,
+      "loss": 1.5895,
+      "step": 8572
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9768358428015616e-05,
+      "loss": 1.8491,
+      "step": 8573
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9744416021616235e-05,
+      "loss": 0.9871,
+      "step": 8574
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.972047746889168e-05,
+      "loss": 0.8192,
+      "step": 8575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.969654277167759e-05,
+      "loss": 1.7191,
+      "step": 8576
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.967261193180933e-05,
+      "loss": 1.5587,
+      "step": 8577
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.964868495112194e-05,
+      "loss": 1.2334,
+      "step": 8578
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.962476183145022e-05,
+      "loss": 0.6337,
+      "step": 8579
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.960084257462856e-05,
+      "loss": 0.7485,
+      "step": 8580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.957692718249115e-05,
+      "loss": 1.2458,
+      "step": 8581
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.955301565687187e-05,
+      "loss": 1.3499,
+      "step": 8582
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.952910799960429e-05,
+      "loss": 1.2738,
+      "step": 8583
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.950520421252172e-05,
+      "loss": 0.8405,
+      "step": 8584
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.948130429745707e-05,
+      "loss": 1.3165,
+      "step": 8585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.945740825624301e-05,
+      "loss": 1.6101,
+      "step": 8586
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.943351609071201e-05,
+      "loss": 1.4841,
+      "step": 8587
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9409627802696115e-05,
+      "loss": 1.0558,
+      "step": 8588
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.938574339402716e-05,
+      "loss": 1.2236,
+      "step": 8589
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9361862866536554e-05,
+      "loss": 1.4751,
+      "step": 8590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9337986222055544e-05,
+      "loss": 1.5443,
+      "step": 8591
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.931411346241501e-05,
+      "loss": 1.0778,
+      "step": 8592
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9290244589445564e-05,
+      "loss": 1.226,
+      "step": 8593
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.926637960497751e-05,
+      "loss": 1.5435,
+      "step": 8594
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9242518510840865e-05,
+      "loss": 1.2651,
+      "step": 8595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.921866130886529e-05,
+      "loss": 1.3092,
+      "step": 8596
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.919480800088021e-05,
+      "loss": 1.2111,
+      "step": 8597
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.917095858871473e-05,
+      "loss": 1.0986,
+      "step": 8598
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.914711307419768e-05,
+      "loss": 1.2373,
+      "step": 8599
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.912327145915755e-05,
+      "loss": 1.1106,
+      "step": 8600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.909943374542256e-05,
+      "loss": 1.5924,
+      "step": 8601
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9075599934820627e-05,
+      "loss": 1.8007,
+      "step": 8602
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.905177002917936e-05,
+      "loss": 1.7036,
+      "step": 8603
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.902794403032607e-05,
+      "loss": 0.9375,
+      "step": 8604
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.90041219400878e-05,
+      "loss": 1.4865,
+      "step": 8605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.89803037602912e-05,
+      "loss": 1.4719,
+      "step": 8606
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8956489492762726e-05,
+      "loss": 1.1797,
+      "step": 8607
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8932679139328486e-05,
+      "loss": 1.727,
+      "step": 8608
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8908872701814304e-05,
+      "loss": 0.7179,
+      "step": 8609
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8885070182045665e-05,
+      "loss": 1.0138,
+      "step": 8610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.886127158184782e-05,
+      "loss": 1.5046,
+      "step": 8611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.883747690304565e-05,
+      "loss": 1.5563,
+      "step": 8612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.88136861474638e-05,
+      "loss": 1.0149,
+      "step": 8613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.878989931692655e-05,
+      "loss": 1.0134,
+      "step": 8614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.876611641325792e-05,
+      "loss": 1.7409,
+      "step": 8615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.874233743828167e-05,
+      "loss": 1.5106,
+      "step": 8616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.871856239382112e-05,
+      "loss": 1.7631,
+      "step": 8617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.869479128169942e-05,
+      "loss": 1.4794,
+      "step": 8618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.867102410373937e-05,
+      "loss": 1.0159,
+      "step": 8619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8647260861763475e-05,
+      "loss": 0.8973,
+      "step": 8620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8623501557593985e-05,
+      "loss": 1.052,
+      "step": 8621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.859974619305271e-05,
+      "loss": 1.5595,
+      "step": 8622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8575994769961285e-05,
+      "loss": 1.3934,
+      "step": 8623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.855224729014097e-05,
+      "loss": 1.8097,
+      "step": 8624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8528503755412845e-05,
+      "loss": 0.7897,
+      "step": 8625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8504764167597596e-05,
+      "loss": 1.592,
+      "step": 8626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.848102852851553e-05,
+      "loss": 1.3027,
+      "step": 8627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8457296839986755e-05,
+      "loss": 1.1866,
+      "step": 8628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.843356910383108e-05,
+      "loss": 1.5563,
+      "step": 8629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8409845321867975e-05,
+      "loss": 0.9212,
+      "step": 8630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.838612549591661e-05,
+      "loss": 1.2704,
+      "step": 8631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.83624096277959e-05,
+      "loss": 1.1195,
+      "step": 8632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.833869771932433e-05,
+      "loss": 1.1268,
+      "step": 8633
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.831498977232022e-05,
+      "loss": 0.9839,
+      "step": 8634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.829128578860151e-05,
+      "loss": 1.3424,
+      "step": 8635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.826758576998589e-05,
+      "loss": 1.1926,
+      "step": 8636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.824388971829068e-05,
+      "loss": 1.0113,
+      "step": 8637
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.822019763533294e-05,
+      "loss": 0.8815,
+      "step": 8638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.819650952292942e-05,
+      "loss": 1.5042,
+      "step": 8639
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.817282538289656e-05,
+      "loss": 0.8729,
+      "step": 8640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.81491452170505e-05,
+      "loss": 1.2452,
+      "step": 8641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.812546902720709e-05,
+      "loss": 1.8553,
+      "step": 8642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.81017968151818e-05,
+      "loss": 0.1871,
+      "step": 8643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8078128582789886e-05,
+      "loss": 1.3219,
+      "step": 8644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.805446433184625e-05,
+      "loss": 1.0399,
+      "step": 8645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.803080406416552e-05,
+      "loss": 1.2996,
+      "step": 8646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8007147781562e-05,
+      "loss": 0.9451,
+      "step": 8647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.798349548584968e-05,
+      "loss": 1.2785,
+      "step": 8648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.795984717884227e-05,
+      "loss": 1.4304,
+      "step": 8649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.793620286235314e-05,
+      "loss": 1.0893,
+      "step": 8650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.791256253819537e-05,
+      "loss": 1.0897,
+      "step": 8651
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.788892620818176e-05,
+      "loss": 0.9247,
+      "step": 8652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.786529387412479e-05,
+      "loss": 1.1534,
+      "step": 8653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.784166553783658e-05,
+      "loss": 1.0742,
+      "step": 8654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.781804120112899e-05,
+      "loss": 1.5904,
+      "step": 8655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.779442086581359e-05,
+      "loss": 1.0098,
+      "step": 8656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.777080453370163e-05,
+      "loss": 1.3376,
+      "step": 8657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.774719220660406e-05,
+      "loss": 1.5067,
+      "step": 8658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.772358388633146e-05,
+      "loss": 0.8476,
+      "step": 8659
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7699979574694165e-05,
+      "loss": 0.573,
+      "step": 8660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.767637927350217e-05,
+      "loss": 1.01,
+      "step": 8661
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.765278298456525e-05,
+      "loss": 1.0486,
+      "step": 8662
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.762919070969277e-05,
+      "loss": 1.3505,
+      "step": 8663
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7605602450693854e-05,
+      "loss": 0.8948,
+      "step": 8664
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.758201820937722e-05,
+      "loss": 1.4051,
+      "step": 8665
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.755843798755137e-05,
+      "loss": 1.2735,
+      "step": 8666
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7534861787024476e-05,
+      "loss": 1.2661,
+      "step": 8667
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7511289609604405e-05,
+      "loss": 1.4926,
+      "step": 8668
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7487721457098735e-05,
+      "loss": 1.3034,
+      "step": 8669
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.746415733131463e-05,
+      "loss": 1.3656,
+      "step": 8670
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.744059723405906e-05,
+      "loss": 0.5207,
+      "step": 8671
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7417041167138665e-05,
+      "loss": 1.182,
+      "step": 8672
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.739348913235975e-05,
+      "loss": 1.1487,
+      "step": 8673
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.736994113152832e-05,
+      "loss": 1.1847,
+      "step": 8674
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7346397166450075e-05,
+      "loss": 1.7146,
+      "step": 8675
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.732285723893039e-05,
+      "loss": 1.3614,
+      "step": 8676
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7299321350774375e-05,
+      "loss": 0.8124,
+      "step": 8677
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.727578950378676e-05,
+      "loss": 0.9603,
+      "step": 8678
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.725226169977205e-05,
+      "loss": 1.1413,
+      "step": 8679
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.722873794053434e-05,
+      "loss": 1.0226,
+      "step": 8680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.720521822787748e-05,
+      "loss": 1.3596,
+      "step": 8681
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7181702563605004e-05,
+      "loss": 1.2201,
+      "step": 8682
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.715819094952013e-05,
+      "loss": 1.4044,
+      "step": 8683
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.713468338742577e-05,
+      "loss": 1.6595,
+      "step": 8684
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7111179879124525e-05,
+      "loss": 1.1304,
+      "step": 8685
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.708768042641866e-05,
+      "loss": 1.1754,
+      "step": 8686
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.706418503111015e-05,
+      "loss": 2.4063,
+      "step": 8687
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.704069369500068e-05,
+      "loss": 1.1253,
+      "step": 8688
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.701720641989159e-05,
+      "loss": 1.5583,
+      "step": 8689
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.699372320758395e-05,
+      "loss": 1.4505,
+      "step": 8690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.697024405987842e-05,
+      "loss": 0.9988,
+      "step": 8691
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6946768978575465e-05,
+      "loss": 1.0114,
+      "step": 8692
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.692329796547517e-05,
+      "loss": 1.1349,
+      "step": 8693
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.689983102237733e-05,
+      "loss": 1.4077,
+      "step": 8694
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6876368151081475e-05,
+      "loss": 0.5675,
+      "step": 8695
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.68529093533867e-05,
+      "loss": 1.3183,
+      "step": 8696
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.682945463109188e-05,
+      "loss": 1.0462,
+      "step": 8697
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6806003985995594e-05,
+      "loss": 1.269,
+      "step": 8698
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6782557419896e-05,
+      "loss": 1.3595,
+      "step": 8699
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.675911493459111e-05,
+      "loss": 1.2576,
+      "step": 8700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6735676531878525e-05,
+      "loss": 1.1303,
+      "step": 8701
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6712242213555456e-05,
+      "loss": 1.0662,
+      "step": 8702
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.668881198141892e-05,
+      "loss": 1.5036,
+      "step": 8703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.666538583726561e-05,
+      "loss": 1.2826,
+      "step": 8704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.664196378289184e-05,
+      "loss": 1.5546,
+      "step": 8705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.661854582009372e-05,
+      "loss": 1.0893,
+      "step": 8706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.659513195066687e-05,
+      "loss": 1.2299,
+      "step": 8707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6571722176406755e-05,
+      "loss": 0.5587,
+      "step": 8708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6548316499108465e-05,
+      "loss": 0.8077,
+      "step": 8709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6524914920566795e-05,
+      "loss": 1.0144,
+      "step": 8710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6501517442576214e-05,
+      "loss": 0.9129,
+      "step": 8711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.647812406693086e-05,
+      "loss": 1.0685,
+      "step": 8712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.645473479542458e-05,
+      "loss": 1.0854,
+      "step": 8713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6431349629850906e-05,
+      "loss": 1.5269,
+      "step": 8714
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6407968572003036e-05,
+      "loss": 1.3123,
+      "step": 8715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6384591623673915e-05,
+      "loss": 1.2995,
+      "step": 8716
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6361218786656034e-05,
+      "loss": 1.5505,
+      "step": 8717
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6337850062741715e-05,
+      "loss": 1.2021,
+      "step": 8718
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6314485453722886e-05,
+      "loss": 1.2943,
+      "step": 8719
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6291124961391205e-05,
+      "loss": 1.8155,
+      "step": 8720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.626776858753796e-05,
+      "loss": 1.5162,
+      "step": 8721
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.624441633395419e-05,
+      "loss": 1.5339,
+      "step": 8722
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.622106820243054e-05,
+      "loss": 0.9142,
+      "step": 8723
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.619772419475741e-05,
+      "loss": 1.2221,
+      "step": 8724
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.617438431272485e-05,
+      "loss": 1.453,
+      "step": 8725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.615104855812259e-05,
+      "loss": 1.1402,
+      "step": 8726
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6127716932740084e-05,
+      "loss": 1.2855,
+      "step": 8727
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.610438943836638e-05,
+      "loss": 0.8608,
+      "step": 8728
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6081066076790295e-05,
+      "loss": 0.8146,
+      "step": 8729
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.605774684980029e-05,
+      "loss": 1.4667,
+      "step": 8730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.603443175918452e-05,
+      "loss": 1.3618,
+      "step": 8731
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.601112080673088e-05,
+      "loss": 1.3001,
+      "step": 8732
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.598781399422679e-05,
+      "loss": 1.3805,
+      "step": 8733
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5964511323459514e-05,
+      "loss": 0.8493,
+      "step": 8734
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5941212796215916e-05,
+      "loss": 0.7579,
+      "step": 8735
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5917918414282526e-05,
+      "loss": 0.9732,
+      "step": 8736
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5894628179445674e-05,
+      "loss": 1.2807,
+      "step": 8737
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5871342093491286e-05,
+      "loss": 0.8562,
+      "step": 8738
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.584806015820491e-05,
+      "loss": 1.3055,
+      "step": 8739
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.582478237537188e-05,
+      "loss": 0.9814,
+      "step": 8740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5801508746777145e-05,
+      "loss": 0.8018,
+      "step": 8741
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.577823927420539e-05,
+      "loss": 1.1165,
+      "step": 8742
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.575497395944097e-05,
+      "loss": 1.3802,
+      "step": 8743
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5731712804267854e-05,
+      "loss": 0.7634,
+      "step": 8744
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.570845581046976e-05,
+      "loss": 1.0238,
+      "step": 8745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.568520297983009e-05,
+      "loss": 0.7521,
+      "step": 8746
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5661954314131884e-05,
+      "loss": 1.118,
+      "step": 8747
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5638709815157897e-05,
+      "loss": 0.8854,
+      "step": 8748
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5615469484690554e-05,
+      "loss": 0.712,
+      "step": 8749
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5592233324511946e-05,
+      "loss": 1.2808,
+      "step": 8750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.556900133640387e-05,
+      "loss": 1.6932,
+      "step": 8751
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5545773522147795e-05,
+      "loss": 1.2777,
+      "step": 8752
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.552254988352486e-05,
+      "loss": 1.1446,
+      "step": 8753
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.549933042231591e-05,
+      "loss": 1.8015,
+      "step": 8754
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.547611514030139e-05,
+      "loss": 1.4282,
+      "step": 8755
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5452904039261526e-05,
+      "loss": 1.4659,
+      "step": 8756
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.542969712097617e-05,
+      "loss": 1.0808,
+      "step": 8757
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.540649438722486e-05,
+      "loss": 1.0891,
+      "step": 8758
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5383295839786874e-05,
+      "loss": 1.1329,
+      "step": 8759
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.536010148044097e-05,
+      "loss": 0.7041,
+      "step": 8760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.533691131096588e-05,
+      "loss": 0.9673,
+      "step": 8761
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.531372533313978e-05,
+      "loss": 1.5576,
+      "step": 8762
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.529054354874064e-05,
+      "loss": 1.4773,
+      "step": 8763
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.526736595954608e-05,
+      "loss": 1.2694,
+      "step": 8764
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.524419256733335e-05,
+      "loss": 1.4197,
+      "step": 8765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5221023373879444e-05,
+      "loss": 1.5809,
+      "step": 8766
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.519785838096101e-05,
+      "loss": 1.4117,
+      "step": 8767
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.517469759035439e-05,
+      "loss": 1.5565,
+      "step": 8768
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.51515410038356e-05,
+      "loss": 1.2101,
+      "step": 8769
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.512838862318026e-05,
+      "loss": 0.8471,
+      "step": 8770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.510524045016376e-05,
+      "loss": 1.3573,
+      "step": 8771
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5082096486561165e-05,
+      "loss": 1.3656,
+      "step": 8772
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.505895673414716e-05,
+      "loss": 0.3655,
+      "step": 8773
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5035821194696104e-05,
+      "loss": 1.0915,
+      "step": 8774
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5012689869982194e-05,
+      "loss": 1.2003,
+      "step": 8775
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.498956276177904e-05,
+      "loss": 2.0894,
+      "step": 8776
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.496643987186011e-05,
+      "loss": 1.2705,
+      "step": 8777
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4943321201998525e-05,
+      "loss": 1.3657,
+      "step": 8778
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.492020675396702e-05,
+      "loss": 0.9794,
+      "step": 8779
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.489709652953811e-05,
+      "loss": 0.7904,
+      "step": 8780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.487399053048384e-05,
+      "loss": 0.7981,
+      "step": 8781
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.485088875857606e-05,
+      "loss": 1.1464,
+      "step": 8782
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4827791215586236e-05,
+      "loss": 1.9558,
+      "step": 8783
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.480469790328553e-05,
+      "loss": 1.3129,
+      "step": 8784
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.478160882344477e-05,
+      "loss": 1.2096,
+      "step": 8785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4758523977834456e-05,
+      "loss": 0.9549,
+      "step": 8786
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4735443368224775e-05,
+      "loss": 1.1087,
+      "step": 8787
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.471236699638558e-05,
+      "loss": 1.4274,
+      "step": 8788
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.468929486408642e-05,
+      "loss": 1.793,
+      "step": 8789
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.466622697309647e-05,
+      "loss": 1.3864,
+      "step": 8790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4643163325184675e-05,
+      "loss": 1.0673,
+      "step": 8791
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4620103922119516e-05,
+      "loss": 1.0153,
+      "step": 8792
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.459704876566924e-05,
+      "loss": 1.5057,
+      "step": 8793
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.457399785760176e-05,
+      "loss": 0.6596,
+      "step": 8794
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.455095119968466e-05,
+      "loss": 0.938,
+      "step": 8795
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.452790879368523e-05,
+      "loss": 1.3342,
+      "step": 8796
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.450487064137029e-05,
+      "loss": 1.8564,
+      "step": 8797
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.448183674450653e-05,
+      "loss": 0.977,
+      "step": 8798
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4458807104860213e-05,
+      "loss": 0.9959,
+      "step": 8799
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4435781724197266e-05,
+      "loss": 1.0712,
+      "step": 8800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.441276060428337e-05,
+      "loss": 0.8656,
+      "step": 8801
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4389743746883726e-05,
+      "loss": 1.5397,
+      "step": 8802
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.436673115376334e-05,
+      "loss": 1.302,
+      "step": 8803
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4343722826686864e-05,
+      "loss": 1.108,
+      "step": 8804
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.432071876741861e-05,
+      "loss": 1.1358,
+      "step": 8805
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.429771897772259e-05,
+      "loss": 1.2029,
+      "step": 8806
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.42747234593624e-05,
+      "loss": 1.1807,
+      "step": 8807
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4251732214101404e-05,
+      "loss": 0.8173,
+      "step": 8808
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.422874524370261e-05,
+      "loss": 1.098,
+      "step": 8809
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4205762549928686e-05,
+      "loss": 1.3964,
+      "step": 8810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.418278413454196e-05,
+      "loss": 1.5381,
+      "step": 8811
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4159809999304536e-05,
+      "loss": 1.1955,
+      "step": 8812
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.413684014597801e-05,
+      "loss": 1.1102,
+      "step": 8813
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.4113874576323786e-05,
+      "loss": 1.5772,
+      "step": 8814
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.409091329210289e-05,
+      "loss": 1.6679,
+      "step": 8815
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.406795629507604e-05,
+      "loss": 1.2291,
+      "step": 8816
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.404500358700363e-05,
+      "loss": 1.0469,
+      "step": 8817
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.402205516964566e-05,
+      "loss": 1.0771,
+      "step": 8818
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3999111044761877e-05,
+      "loss": 1.3419,
+      "step": 8819
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.397617121411166e-05,
+      "loss": 1.0057,
+      "step": 8820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.395323567945407e-05,
+      "loss": 1.5588,
+      "step": 8821
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.393030444254787e-05,
+      "loss": 1.4986,
+      "step": 8822
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3907377505151425e-05,
+      "loss": 1.1667,
+      "step": 8823
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.388445486902284e-05,
+      "loss": 1.4769,
+      "step": 8824
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3861536535919825e-05,
+      "loss": 1.3906,
+      "step": 8825
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.383862250759981e-05,
+      "loss": 1.1879,
+      "step": 8826
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3815712785819876e-05,
+      "loss": 1.3466,
+      "step": 8827
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3792807372336816e-05,
+      "loss": 1.3627,
+      "step": 8828
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3769906268906966e-05,
+      "loss": 1.0734,
+      "step": 8829
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.374700947728647e-05,
+      "loss": 0.6715,
+      "step": 8830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.372411699923109e-05,
+      "loss": 1.3381,
+      "step": 8831
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3701228836496234e-05,
+      "loss": 1.4426,
+      "step": 8832
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.367834499083706e-05,
+      "loss": 1.0254,
+      "step": 8833
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.365546546400825e-05,
+      "loss": 1.5321,
+      "step": 8834
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.363259025776425e-05,
+      "loss": 1.2948,
+      "step": 8835
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3609719373859224e-05,
+      "loss": 1.3645,
+      "step": 8836
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.358685281404693e-05,
+      "loss": 1.9249,
+      "step": 8837
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3563990580080835e-05,
+      "loss": 0.1303,
+      "step": 8838
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3541132673713984e-05,
+      "loss": 1.6054,
+      "step": 8839
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3518279096699174e-05,
+      "loss": 0.9829,
+      "step": 8840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3495429850788884e-05,
+      "loss": 1.4582,
+      "step": 8841
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3472584937735204e-05,
+      "loss": 1.1531,
+      "step": 8842
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.344974435928992e-05,
+      "loss": 1.0819,
+      "step": 8843
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.342690811720452e-05,
+      "loss": 0.68,
+      "step": 8844
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.340407621323006e-05,
+      "loss": 1.6554,
+      "step": 8845
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.338124864911736e-05,
+      "loss": 0.9943,
+      "step": 8846
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3358425426616855e-05,
+      "loss": 1.8348,
+      "step": 8847
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.333560654747867e-05,
+      "loss": 1.2672,
+      "step": 8848
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3312792013452606e-05,
+      "loss": 1.7305,
+      "step": 8849
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.328998182628811e-05,
+      "loss": 0.82,
+      "step": 8850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3267175987734284e-05,
+      "loss": 1.4021,
+      "step": 8851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.324437449953994e-05,
+      "loss": 1.7604,
+      "step": 8852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.322157736345353e-05,
+      "loss": 1.1283,
+      "step": 8853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.319878458122319e-05,
+      "loss": 1.1906,
+      "step": 8854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.317599615459664e-05,
+      "loss": 1.4011,
+      "step": 8855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.315321208532137e-05,
+      "loss": 1.1739,
+      "step": 8856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.31304323751445e-05,
+      "loss": 1.8611,
+      "step": 8857
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.310765702581281e-05,
+      "loss": 1.0076,
+      "step": 8858
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.308488603907275e-05,
+      "loss": 1.366,
+      "step": 8859
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.306211941667043e-05,
+      "loss": 1.5094,
+      "step": 8860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3039357160351633e-05,
+      "loss": 0.9504,
+      "step": 8861
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.301659927186181e-05,
+      "loss": 1.0766,
+      "step": 8862
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2993845752946064e-05,
+      "loss": 1.1188,
+      "step": 8863
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.297109660534916e-05,
+      "loss": 1.0936,
+      "step": 8864
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.294835183081558e-05,
+      "loss": 1.4256,
+      "step": 8865
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2925611431089366e-05,
+      "loss": 1.8025,
+      "step": 8866
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.29028754079143e-05,
+      "loss": 1.5149,
+      "step": 8867
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.288014376303384e-05,
+      "loss": 1.4618,
+      "step": 8868
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2857416498191086e-05,
+      "loss": 1.285,
+      "step": 8869
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.283469361512881e-05,
+      "loss": 1.1391,
+      "step": 8870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2811975115589376e-05,
+      "loss": 1.6847,
+      "step": 8871
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.278926100131487e-05,
+      "loss": 1.5706,
+      "step": 8872
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2766551274047126e-05,
+      "loss": 1.2178,
+      "step": 8873
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2743845935527526e-05,
+      "loss": 1.7493,
+      "step": 8874
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2721144987497173e-05,
+      "loss": 1.1344,
+      "step": 8875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.269844843169675e-05,
+      "loss": 1.239,
+      "step": 8876
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2675756269866684e-05,
+      "loss": 1.7143,
+      "step": 8877
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2653068503747074e-05,
+      "loss": 1.313,
+      "step": 8878
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.263038513507762e-05,
+      "loss": 2.61,
+      "step": 8879
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.260770616559774e-05,
+      "loss": 0.9759,
+      "step": 8880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.258503159704651e-05,
+      "loss": 1.113,
+      "step": 8881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.256236143116259e-05,
+      "loss": 1.0113,
+      "step": 8882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2539695669684396e-05,
+      "loss": 1.4141,
+      "step": 8883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.251703431434998e-05,
+      "loss": 1.2674,
+      "step": 8884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2494377366897046e-05,
+      "loss": 1.0697,
+      "step": 8885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2471724829062954e-05,
+      "loss": 1.2564,
+      "step": 8886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.244907670258473e-05,
+      "loss": 1.5531,
+      "step": 8887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.242643298919909e-05,
+      "loss": 1.1026,
+      "step": 8888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.240379369064239e-05,
+      "loss": 1.8073,
+      "step": 8889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.238115880865061e-05,
+      "loss": 1.2697,
+      "step": 8890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.23585283449595e-05,
+      "loss": 1.2942,
+      "step": 8891
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.233590230130431e-05,
+      "loss": 0.9561,
+      "step": 8892
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2313280679420074e-05,
+      "loss": 1.6756,
+      "step": 8893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.229066348104147e-05,
+      "loss": 1.0043,
+      "step": 8894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.22680507079028e-05,
+      "loss": 1.1562,
+      "step": 8895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.224544236173804e-05,
+      "loss": 1.5115,
+      "step": 8896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.222283844428085e-05,
+      "loss": 1.4109,
+      "step": 8897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.220023895726453e-05,
+      "loss": 0.9817,
+      "step": 8898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2177643902422036e-05,
+      "loss": 1.0205,
+      "step": 8899
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2155053281486e-05,
+      "loss": 0.7944,
+      "step": 8900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.21324670961887e-05,
+      "loss": 1.0809,
+      "step": 8901
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.210988534826211e-05,
+      "loss": 1.0111,
+      "step": 8902
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.208730803943778e-05,
+      "loss": 1.5526,
+      "step": 8903
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.206473517144699e-05,
+      "loss": 1.2467,
+      "step": 8904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.204216674602066e-05,
+      "loss": 1.1263,
+      "step": 8905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2019602764889385e-05,
+      "loss": 1.0718,
+      "step": 8906
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1997043229783425e-05,
+      "loss": 0.9911,
+      "step": 8907
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.197448814243263e-05,
+      "loss": 1.5316,
+      "step": 8908
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1951937504566585e-05,
+      "loss": 1.5678,
+      "step": 8909
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.192939131791446e-05,
+      "loss": 1.1555,
+      "step": 8910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.190684958420521e-05,
+      "loss": 2.1213,
+      "step": 8911
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.188431230516737e-05,
+      "loss": 1.6641,
+      "step": 8912
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.186177948252907e-05,
+      "loss": 1.2427,
+      "step": 8913
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.183925111801819e-05,
+      "loss": 0.868,
+      "step": 8914
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.181672721336223e-05,
+      "loss": 0.913,
+      "step": 8915
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.179420777028836e-05,
+      "loss": 1.5993,
+      "step": 8916
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.177169279052342e-05,
+      "loss": 1.205,
+      "step": 8917
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.174918227579392e-05,
+      "loss": 0.88,
+      "step": 8918
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.172667622782593e-05,
+      "loss": 1.1217,
+      "step": 8919
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1704174648345284e-05,
+      "loss": 0.9323,
+      "step": 8920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1681677539077424e-05,
+      "loss": 1.0291,
+      "step": 8921
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1659184901747486e-05,
+      "loss": 1.4238,
+      "step": 8922
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.163669673808023e-05,
+      "loss": 1.3242,
+      "step": 8923
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.161421304980008e-05,
+      "loss": 1.2947,
+      "step": 8924
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1591733838631106e-05,
+      "loss": 1.2181,
+      "step": 8925
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1569259106297075e-05,
+      "loss": 1.2433,
+      "step": 8926
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.154678885452137e-05,
+      "loss": 1.4377,
+      "step": 8927
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1524323085027086e-05,
+      "loss": 1.0948,
+      "step": 8928
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.150186179953686e-05,
+      "loss": 1.2757,
+      "step": 8929
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.147940499977308e-05,
+      "loss": 0.8333,
+      "step": 8930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1456952687457776e-05,
+      "loss": 0.6703,
+      "step": 8931
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.143450486431263e-05,
+      "loss": 1.1602,
+      "step": 8932
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.141206153205898e-05,
+      "loss": 1.186,
+      "step": 8933
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.13896226924178e-05,
+      "loss": 1.5271,
+      "step": 8934
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.136718834710973e-05,
+      "loss": 1.1342,
+      "step": 8935
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1344758497855097e-05,
+      "loss": 1.4373,
+      "step": 8936
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.132233314637384e-05,
+      "loss": 1.716,
+      "step": 8937
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.129991229438557e-05,
+      "loss": 1.2083,
+      "step": 8938
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1277495943609576e-05,
+      "loss": 1.3471,
+      "step": 8939
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.125508409576473e-05,
+      "loss": 0.6741,
+      "step": 8940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1232676752569634e-05,
+      "loss": 1.1005,
+      "step": 8941
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1210273915742505e-05,
+      "loss": 1.4154,
+      "step": 8942
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1187875587001236e-05,
+      "loss": 1.0331,
+      "step": 8943
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.116548176806341e-05,
+      "loss": 1.1794,
+      "step": 8944
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.114309246064613e-05,
+      "loss": 0.9727,
+      "step": 8945
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1120707666466285e-05,
+      "loss": 1.5861,
+      "step": 8946
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.109832738724034e-05,
+      "loss": 1.3953,
+      "step": 8947
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.107595162468452e-05,
+      "loss": 1.0716,
+      "step": 8948
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.105358038051463e-05,
+      "loss": 1.8161,
+      "step": 8949
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.103121365644608e-05,
+      "loss": 1.0803,
+      "step": 8950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1008851454193987e-05,
+      "loss": 1.0158,
+      "step": 8951
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.098649377547314e-05,
+      "loss": 1.3567,
+      "step": 8952
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.096414062199796e-05,
+      "loss": 1.3062,
+      "step": 8953
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.09417919954825e-05,
+      "loss": 1.2838,
+      "step": 8954
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.091944789764055e-05,
+      "loss": 1.2704,
+      "step": 8955
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.089710833018541e-05,
+      "loss": 1.1038,
+      "step": 8956
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0874773294830136e-05,
+      "loss": 1.084,
+      "step": 8957
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.085244279328743e-05,
+      "loss": 1.262,
+      "step": 8958
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.083011682726961e-05,
+      "loss": 1.009,
+      "step": 8959
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.080779539848868e-05,
+      "loss": 1.1477,
+      "step": 8960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0785478508656286e-05,
+      "loss": 1.1272,
+      "step": 8961
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.076316615948369e-05,
+      "loss": 1.4613,
+      "step": 8962
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.074085835268188e-05,
+      "loss": 1.6252,
+      "step": 8963
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.071855508996141e-05,
+      "loss": 1.3163,
+      "step": 8964
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.069625637303259e-05,
+      "loss": 0.757,
+      "step": 8965
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.067396220360525e-05,
+      "loss": 1.3673,
+      "step": 8966
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.065167258338897e-05,
+      "loss": 1.5679,
+      "step": 8967
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0629387514092944e-05,
+      "loss": 1.113,
+      "step": 8968
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0607106997426035e-05,
+      "loss": 1.0995,
+      "step": 8969
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0584831035096745e-05,
+      "loss": 1.3092,
+      "step": 8970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0562559628813236e-05,
+      "loss": 1.3016,
+      "step": 8971
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0540292780283295e-05,
+      "loss": 1.1102,
+      "step": 8972
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.05180304912144e-05,
+      "loss": 0.9204,
+      "step": 8973
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0495772763313643e-05,
+      "loss": 1.4077,
+      "step": 8974
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.047351959828778e-05,
+      "loss": 1.0814,
+      "step": 8975
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.045127099784325e-05,
+      "loss": 1.0581,
+      "step": 8976
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0429026963686053e-05,
+      "loss": 1.1042,
+      "step": 8977
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0406787497521924e-05,
+      "loss": 1.648,
+      "step": 8978
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.038455260105623e-05,
+      "loss": 1.1956,
+      "step": 8979
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.036232227599395e-05,
+      "loss": 0.9024,
+      "step": 8980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.034009652403978e-05,
+      "loss": 1.7629,
+      "step": 8981
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.031787534689798e-05,
+      "loss": 1.5931,
+      "step": 8982
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.029565874627252e-05,
+      "loss": 1.5505,
+      "step": 8983
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0273446723867e-05,
+      "loss": 1.1356,
+      "step": 8984
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0251239281384655e-05,
+      "loss": 0.8534,
+      "step": 8985
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.022903642052846e-05,
+      "loss": 1.6752,
+      "step": 8986
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.020683814300089e-05,
+      "loss": 0.8532,
+      "step": 8987
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.018464445050416e-05,
+      "loss": 1.4333,
+      "step": 8988
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.016245534474012e-05,
+      "loss": 1.252,
+      "step": 8989
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0140270827410274e-05,
+      "loss": 1.3637,
+      "step": 8990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.011809090021575e-05,
+      "loss": 1.3236,
+      "step": 8991
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.009591556485739e-05,
+      "loss": 1.4993,
+      "step": 8992
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0073744823035544e-05,
+      "loss": 1.674,
+      "step": 8993
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.005157867645034e-05,
+      "loss": 1.3881,
+      "step": 8994
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.00294171268015e-05,
+      "loss": 1.4986,
+      "step": 8995
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0007260175788444e-05,
+      "loss": 0.9444,
+      "step": 8996
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9985107825110166e-05,
+      "loss": 2.2745,
+      "step": 8997
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.996296007646535e-05,
+      "loss": 1.298,
+      "step": 8998
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.994081693155233e-05,
+      "loss": 1.3492,
+      "step": 8999
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9918678392069055e-05,
+      "loss": 1.1175,
+      "step": 9000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.989654445971316e-05,
+      "loss": 0.7905,
+      "step": 9001
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.987441513618193e-05,
+      "loss": 1.73,
+      "step": 9002
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.985229042317222e-05,
+      "loss": 1.3407,
+      "step": 9003
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.983017032238062e-05,
+      "loss": 1.4311,
+      "step": 9004
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9808054835503314e-05,
+      "loss": 1.1479,
+      "step": 9005
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9785943964236184e-05,
+      "loss": 1.3207,
+      "step": 9006
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.976383771027469e-05,
+      "loss": 1.1837,
+      "step": 9007
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.974173607531399e-05,
+      "loss": 1.3581,
+      "step": 9008
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.971963906104889e-05,
+      "loss": 1.0951,
+      "step": 9009
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.969754666917378e-05,
+      "loss": 1.4221,
+      "step": 9010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.967545890138278e-05,
+      "loss": 1.2262,
+      "step": 9011
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9653375759369585e-05,
+      "loss": 1.1004,
+      "step": 9012
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9631297244827616e-05,
+      "loss": 1.1875,
+      "step": 9013
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.96092233594498e-05,
+      "loss": 1.0047,
+      "step": 9014
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9587154104928856e-05,
+      "loss": 1.1446,
+      "step": 9015
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.956508948295706e-05,
+      "loss": 0.854,
+      "step": 9016
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.954302949522639e-05,
+      "loss": 0.9767,
+      "step": 9017
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.952097414342845e-05,
+      "loss": 1.432,
+      "step": 9018
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.949892342925442e-05,
+      "loss": 1.4134,
+      "step": 9019
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9476877354395216e-05,
+      "loss": 1.1037,
+      "step": 9020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9454835920541356e-05,
+      "loss": 1.3969,
+      "step": 9021
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.943279912938299e-05,
+      "loss": 1.5285,
+      "step": 9022
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9410766982609995e-05,
+      "loss": 0.6331,
+      "step": 9023
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.938873948191183e-05,
+      "loss": 1.5245,
+      "step": 9024
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.936671662897753e-05,
+      "loss": 1.1453,
+      "step": 9025
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.934469842549587e-05,
+      "loss": 1.8863,
+      "step": 9026
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9322684873155245e-05,
+      "loss": 1.2692,
+      "step": 9027
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.93006759736437e-05,
+      "loss": 1.0625,
+      "step": 9028
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9278671728648906e-05,
+      "loss": 0.651,
+      "step": 9029
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.925667213985815e-05,
+      "loss": 0.8925,
+      "step": 9030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.923467720895842e-05,
+      "loss": 1.7597,
+      "step": 9031
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.921268693763632e-05,
+      "loss": 1.0151,
+      "step": 9032
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.919070132757809e-05,
+      "loss": 1.4654,
+      "step": 9033
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9168720380469634e-05,
+      "loss": 1.2421,
+      "step": 9034
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.914674409799647e-05,
+      "loss": 1.1245,
+      "step": 9035
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.912477248184378e-05,
+      "loss": 1.3017,
+      "step": 9036
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.910280553369637e-05,
+      "loss": 1.0718,
+      "step": 9037
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.908084325523873e-05,
+      "loss": 0.955,
+      "step": 9038
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.905888564815496e-05,
+      "loss": 1.5239,
+      "step": 9039
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.903693271412875e-05,
+      "loss": 1.2118,
+      "step": 9040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.901498445484353e-05,
+      "loss": 0.8814,
+      "step": 9041
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.89930408719823e-05,
+      "loss": 1.2974,
+      "step": 9042
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.897110196722774e-05,
+      "loss": 1.5175,
+      "step": 9043
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.894916774226216e-05,
+      "loss": 1.5553,
+      "step": 9044
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8927238198767516e-05,
+      "loss": 1.2996,
+      "step": 9045
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8905313338425384e-05,
+      "loss": 0.9369,
+      "step": 9046
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.888339316291701e-05,
+      "loss": 1.0669,
+      "step": 9047
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.886147767392326e-05,
+      "loss": 1.2518,
+      "step": 9048
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.883956687312465e-05,
+      "loss": 1.5456,
+      "step": 9049
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.881766076220137e-05,
+      "loss": 1.3256,
+      "step": 9050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.879575934283313e-05,
+      "loss": 1.0071,
+      "step": 9051
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.877386261669942e-05,
+      "loss": 1.2143,
+      "step": 9052
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.875197058547929e-05,
+      "loss": 1.0239,
+      "step": 9053
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.873008325085148e-05,
+      "loss": 0.9546,
+      "step": 9054
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.870820061449436e-05,
+      "loss": 1.0214,
+      "step": 9055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.868632267808586e-05,
+      "loss": 1.0768,
+      "step": 9056
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.866444944330366e-05,
+      "loss": 2.7174,
+      "step": 9057
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.864258091182501e-05,
+      "loss": 1.4419,
+      "step": 9058
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.862071708532684e-05,
+      "loss": 1.0191,
+      "step": 9059
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.859885796548567e-05,
+      "loss": 1.2187,
+      "step": 9060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.857700355397778e-05,
+      "loss": 1.4383,
+      "step": 9061
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.85551538524789e-05,
+      "loss": 0.9844,
+      "step": 9062
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.853330886266454e-05,
+      "loss": 1.3545,
+      "step": 9063
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8511468586209796e-05,
+      "loss": 1.3875,
+      "step": 9064
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.848963302478943e-05,
+      "loss": 1.1849,
+      "step": 9065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.846780218007785e-05,
+      "loss": 1.386,
+      "step": 9066
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8445976053749e-05,
+      "loss": 1.3905,
+      "step": 9067
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.84241546474766e-05,
+      "loss": 0.9249,
+      "step": 9068
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.840233796293392e-05,
+      "loss": 1.4153,
+      "step": 9069
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8380526001793915e-05,
+      "loss": 1.0081,
+      "step": 9070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.835871876572916e-05,
+      "loss": 0.9469,
+      "step": 9071
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.833691625641186e-05,
+      "loss": 1.0705,
+      "step": 9072
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8315118475513866e-05,
+      "loss": 1.0983,
+      "step": 9073
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.829332542470667e-05,
+      "loss": 1.3193,
+      "step": 9074
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.827153710566139e-05,
+      "loss": 1.7347,
+      "step": 9075
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8249753520048826e-05,
+      "loss": 0.9346,
+      "step": 9076
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8227974669539303e-05,
+      "loss": 1.1004,
+      "step": 9077
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8206200555802895e-05,
+      "loss": 1.4881,
+      "step": 9078
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.818443118050927e-05,
+      "loss": 1.0958,
+      "step": 9079
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.816266654532775e-05,
+      "loss": 1.4992,
+      "step": 9080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8140906651927265e-05,
+      "loss": 0.9989,
+      "step": 9081
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8119151501976436e-05,
+      "loss": 1.3047,
+      "step": 9082
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8097401097143385e-05,
+      "loss": 1.1941,
+      "step": 9083
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.807565543909606e-05,
+      "loss": 1.1568,
+      "step": 9084
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.805391452950191e-05,
+      "loss": 0.9725,
+      "step": 9085
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.803217837002808e-05,
+      "loss": 1.3041,
+      "step": 9086
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.801044696234136e-05,
+      "loss": 1.1251,
+      "step": 9087
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.798872030810807e-05,
+      "loss": 1.2002,
+      "step": 9088
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.796699840899429e-05,
+      "loss": 1.537,
+      "step": 9089
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7945281266665676e-05,
+      "loss": 1.4597,
+      "step": 9090
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.792356888278754e-05,
+      "loss": 1.2348,
+      "step": 9091
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7901861259024854e-05,
+      "loss": 1.2406,
+      "step": 9092
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.788015839704212e-05,
+      "loss": 1.7512,
+      "step": 9093
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7858460298503586e-05,
+      "loss": 1.4778,
+      "step": 9094
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.783676696507308e-05,
+      "loss": 1.0879,
+      "step": 9095
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7815078398414106e-05,
+      "loss": 0.8613,
+      "step": 9096
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.779339460018971e-05,
+      "loss": 1.1959,
+      "step": 9097
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.777171557206278e-05,
+      "loss": 1.1655,
+      "step": 9098
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7750041315695564e-05,
+      "loss": 1.1764,
+      "step": 9099
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.772837183275012e-05,
+      "loss": 1.4937,
+      "step": 9100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.77067071248881e-05,
+      "loss": 1.4617,
+      "step": 9101
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7685047193770785e-05,
+      "loss": 1.5686,
+      "step": 9102
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.766339204105912e-05,
+      "loss": 1.0315,
+      "step": 9103
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.764174166841359e-05,
+      "loss": 1.2985,
+      "step": 9104
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.762009607749441e-05,
+      "loss": 1.5207,
+      "step": 9105
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.759845526996142e-05,
+      "loss": 1.489,
+      "step": 9106
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7576819247474046e-05,
+      "loss": 1.0891,
+      "step": 9107
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7555188011691375e-05,
+      "loss": 0.8185,
+      "step": 9108
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.753356156427212e-05,
+      "loss": 1.278,
+      "step": 9109
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7511939906874635e-05,
+      "loss": 1.6314,
+      "step": 9110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7490323041156916e-05,
+      "loss": 1.3563,
+      "step": 9111
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7468710968776556e-05,
+      "loss": 1.4814,
+      "step": 9112
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.74471036913908e-05,
+      "loss": 1.577,
+      "step": 9113
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.742550121065658e-05,
+      "loss": 1.1327,
+      "step": 9114
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.740390352823032e-05,
+      "loss": 1.0119,
+      "step": 9115
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.738231064576821e-05,
+      "loss": 1.2882,
+      "step": 9116
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.736072256492602e-05,
+      "loss": 1.0716,
+      "step": 9117
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.733913928735917e-05,
+      "loss": 0.7897,
+      "step": 9118
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.731756081472271e-05,
+      "loss": 1.3057,
+      "step": 9119
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.729598714867122e-05,
+      "loss": 1.3873,
+      "step": 9120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7274418290859106e-05,
+      "loss": 0.9805,
+      "step": 9121
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.725285424294025e-05,
+      "loss": 1.1377,
+      "step": 9122
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.723129500656824e-05,
+      "loss": 1.1423,
+      "step": 9123
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.72097405833963e-05,
+      "loss": 0.9782,
+      "step": 9124
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.718819097507717e-05,
+      "loss": 1.6028,
+      "step": 9125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.716664618326337e-05,
+      "loss": 1.1381,
+      "step": 9126
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7145106209606964e-05,
+      "loss": 1.164,
+      "step": 9127
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7123571055759676e-05,
+      "loss": 0.9789,
+      "step": 9128
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.710204072337289e-05,
+      "loss": 0.7719,
+      "step": 9129
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.708051521409751e-05,
+      "loss": 1.2648,
+      "step": 9130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.70589945295842e-05,
+      "loss": 1.1606,
+      "step": 9131
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.703747867148317e-05,
+      "loss": 1.3897,
+      "step": 9132
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7015967641444304e-05,
+      "loss": 1.4407,
+      "step": 9133
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.699446144111711e-05,
+      "loss": 1.4966,
+      "step": 9134
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.69729600721507e-05,
+      "loss": 1.6966,
+      "step": 9135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6951463536193844e-05,
+      "loss": 1.7767,
+      "step": 9136
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6929971834894905e-05,
+      "loss": 1.0821,
+      "step": 9137
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.690848496990194e-05,
+      "loss": 1.2546,
+      "step": 9138
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.688700294286255e-05,
+      "loss": 1.4113,
+      "step": 9139
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.686552575542408e-05,
+      "loss": 1.4021,
+      "step": 9140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.684405340923335e-05,
+      "loss": 1.1034,
+      "step": 9141
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6822585905936914e-05,
+      "loss": 1.0856,
+      "step": 9142
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.680112324718096e-05,
+      "loss": 0.7982,
+      "step": 9143
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.677966543461126e-05,
+      "loss": 1.4972,
+      "step": 9144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6758212469873234e-05,
+      "loss": 1.5528,
+      "step": 9145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.673676435461193e-05,
+      "loss": 1.1035,
+      "step": 9146
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6715321090472024e-05,
+      "loss": 1.2793,
+      "step": 9147
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6693882679097805e-05,
+      "loss": 1.3778,
+      "step": 9148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6672449122133225e-05,
+      "loss": 1.1383,
+      "step": 9149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.665102042122183e-05,
+      "loss": 1.007,
+      "step": 9150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.662959657800683e-05,
+      "loss": 1.4205,
+      "step": 9151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.660817759413099e-05,
+      "loss": 1.4838,
+      "step": 9152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6586763471236764e-05,
+      "loss": 1.902,
+      "step": 9153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.656535421096624e-05,
+      "loss": 1.4851,
+      "step": 9154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.65439498149611e-05,
+      "loss": 0.9566,
+      "step": 9155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.65225502848627e-05,
+      "loss": 1.0652,
+      "step": 9156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.650115562231192e-05,
+      "loss": 1.0617,
+      "step": 9157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.647976582894935e-05,
+      "loss": 1.4729,
+      "step": 9158
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6458380906415244e-05,
+      "loss": 1.3782,
+      "step": 9159
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.64370008563494e-05,
+      "loss": 1.1798,
+      "step": 9160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.641562568039131e-05,
+      "loss": 1.4004,
+      "step": 9161
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.639425538017997e-05,
+      "loss": 1.1252,
+      "step": 9162
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.637288995735415e-05,
+      "loss": 1.5791,
+      "step": 9163
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6351529413552166e-05,
+      "loss": 1.18,
+      "step": 9164
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.633017375041198e-05,
+      "loss": 1.2856,
+      "step": 9165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.630882296957121e-05,
+      "loss": 1.1575,
+      "step": 9166
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6287477072666995e-05,
+      "loss": 1.0508,
+      "step": 9167
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.626613606133622e-05,
+      "loss": 1.1686,
+      "step": 9168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.624479993721533e-05,
+      "loss": 1.1635,
+      "step": 9169
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.622346870194042e-05,
+      "loss": 1.333,
+      "step": 9170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.620214235714719e-05,
+      "loss": 1.866,
+      "step": 9171
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6180820904471e-05,
+      "loss": 1.4923,
+      "step": 9172
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.615950434554679e-05,
+      "loss": 1.1129,
+      "step": 9173
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6138192682009154e-05,
+      "loss": 1.2071,
+      "step": 9174
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6116885915492304e-05,
+      "loss": 0.9455,
+      "step": 9175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.609558404763007e-05,
+      "loss": 1.858,
+      "step": 9176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.607428708005595e-05,
+      "loss": 0.8632,
+      "step": 9177
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6052995014402957e-05,
+      "loss": 1.2555,
+      "step": 9178
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6031707852303845e-05,
+      "loss": 1.1096,
+      "step": 9179
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.601042559539093e-05,
+      "loss": 1.6555,
+      "step": 9180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.598914824529617e-05,
+      "loss": 0.8052,
+      "step": 9181
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5967875803651155e-05,
+      "loss": 1.1669,
+      "step": 9182
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.594660827208708e-05,
+      "loss": 0.8769,
+      "step": 9183
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5925345652234766e-05,
+      "loss": 1.3318,
+      "step": 9184
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5904087945724685e-05,
+      "loss": 1.2681,
+      "step": 9185
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.588283515418689e-05,
+      "loss": 1.1624,
+      "step": 9186
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.586158727925108e-05,
+      "loss": 1.0751,
+      "step": 9187
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.584034432254661e-05,
+      "loss": 1.2331,
+      "step": 9188
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.581910628570238e-05,
+      "loss": 1.4524,
+      "step": 9189
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.579787317034694e-05,
+      "loss": 1.0976,
+      "step": 9190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.577664497810851e-05,
+      "loss": 0.9176,
+      "step": 9191
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.57554217106149e-05,
+      "loss": 1.2439,
+      "step": 9192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5734203369493555e-05,
+      "loss": 1.1102,
+      "step": 9193
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.571298995637149e-05,
+      "loss": 1.7024,
+      "step": 9194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.569178147287537e-05,
+      "loss": 1.5959,
+      "step": 9195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.567057792063155e-05,
+      "loss": 1.2587,
+      "step": 9196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.564937930126594e-05,
+      "loss": 0.8905,
+      "step": 9197
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5628185616404095e-05,
+      "loss": 1.4832,
+      "step": 9198
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5606996867671124e-05,
+      "loss": 1.0564,
+      "step": 9199
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5585813056691844e-05,
+      "loss": 0.7344,
+      "step": 9200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.556463418509066e-05,
+      "loss": 1.3869,
+      "step": 9201
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.55434602544916e-05,
+      "loss": 1.4566,
+      "step": 9202
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5522291266518306e-05,
+      "loss": 1.071,
+      "step": 9203
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.550112722279409e-05,
+      "loss": 1.4708,
+      "step": 9204
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.547996812494179e-05,
+      "loss": 0.8929,
+      "step": 9205
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.545881397458393e-05,
+      "loss": 1.4349,
+      "step": 9206
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.543766477334265e-05,
+      "loss": 1.5437,
+      "step": 9207
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.541652052283969e-05,
+      "loss": 1.4166,
+      "step": 9208
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.539538122469645e-05,
+      "loss": 1.366,
+      "step": 9209
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.537424688053391e-05,
+      "loss": 1.3199,
+      "step": 9210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.535311749197268e-05,
+      "loss": 1.2223,
+      "step": 9211
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5331993060633e-05,
+      "loss": 1.1347,
+      "step": 9212
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5310873588134705e-05,
+      "loss": 0.8037,
+      "step": 9213
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.528975907609734e-05,
+      "loss": 1.3347,
+      "step": 9214
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.52686495261399e-05,
+      "loss": 0.8912,
+      "step": 9215
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.524754493988114e-05,
+      "loss": 1.0601,
+      "step": 9216
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.522644531893938e-05,
+      "loss": 1.5933,
+      "step": 9217
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5205350664932603e-05,
+      "loss": 1.0824,
+      "step": 9218
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5184260979478344e-05,
+      "loss": 1.6695,
+      "step": 9219
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.516317626419381e-05,
+      "loss": 1.4783,
+      "step": 9220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.514209652069581e-05,
+      "loss": 0.9559,
+      "step": 9221
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5121021750600756e-05,
+      "loss": 0.7065,
+      "step": 9222
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5099951955524714e-05,
+      "loss": 1.6118,
+      "step": 9223
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5078887137083336e-05,
+      "loss": 1.1453,
+      "step": 9224
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.505782729689194e-05,
+      "loss": 0.9189,
+      "step": 9225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.503677243656537e-05,
+      "loss": 1.1275,
+      "step": 9226
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5015722557718156e-05,
+      "loss": 1.5157,
+      "step": 9227
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.499467766196446e-05,
+      "loss": 1.2958,
+      "step": 9228
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4973637750918006e-05,
+      "loss": 1.0793,
+      "step": 9229
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.495260282619224e-05,
+      "loss": 1.2819,
+      "step": 9230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4931572889400044e-05,
+      "loss": 0.7997,
+      "step": 9231
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.491054794215408e-05,
+      "loss": 1.1274,
+      "step": 9232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.488952798606655e-05,
+      "loss": 1.4679,
+      "step": 9233
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4868513022749335e-05,
+      "loss": 0.8694,
+      "step": 9234
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4847503053813924e-05,
+      "loss": 1.5336,
+      "step": 9235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.482649808087131e-05,
+      "loss": 1.3221,
+      "step": 9236
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.480549810553222e-05,
+      "loss": 1.1006,
+      "step": 9237
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.478450312940696e-05,
+      "loss": 1.6255,
+      "step": 9238
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.476351315410548e-05,
+      "loss": 1.7033,
+      "step": 9239
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.47425281812373e-05,
+      "loss": 1.4315,
+      "step": 9240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.472154821241161e-05,
+      "loss": 0.7776,
+      "step": 9241
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4700573249237145e-05,
+      "loss": 0.978,
+      "step": 9242
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4679603293322306e-05,
+      "loss": 0.9655,
+      "step": 9243
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.465863834627512e-05,
+      "loss": 1.1461,
+      "step": 9244
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.463767840970319e-05,
+      "loss": 0.9745,
+      "step": 9245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4616723485213764e-05,
+      "loss": 1.2132,
+      "step": 9246
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.459577357441372e-05,
+      "loss": 0.874,
+      "step": 9247
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.457482867890949e-05,
+      "loss": 1.9015,
+      "step": 9248
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.455388880030719e-05,
+      "loss": 0.6943,
+      "step": 9249
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4532953940212495e-05,
+      "loss": 1.5862,
+      "step": 9250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.451202410023079e-05,
+      "loss": 0.892,
+      "step": 9251
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4491099281966924e-05,
+      "loss": 0.8069,
+      "step": 9252
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.447017948702547e-05,
+      "loss": 1.9185,
+      "step": 9253
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.44492647170106e-05,
+      "loss": 1.0158,
+      "step": 9254
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.442835497352608e-05,
+      "loss": 1.4762,
+      "step": 9255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.440745025817531e-05,
+      "loss": 1.1262,
+      "step": 9256
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4386550572561296e-05,
+      "loss": 1.0413,
+      "step": 9257
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.436565591828665e-05,
+      "loss": 0.9541,
+      "step": 9258
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.434476629695362e-05,
+      "loss": 1.1964,
+      "step": 9259
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4323881710164044e-05,
+      "loss": 1.4786,
+      "step": 9260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.430300215951937e-05,
+      "loss": 1.1822,
+      "step": 9261
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 0.8796582818031311,
+      "eval_runtime": 20.5455,
+      "eval_samples_per_second": 138.863,
+      "eval_steps_per_second": 138.863,
+      "step": 9261
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.428212764662073e-05,
+      "loss": 1.078,
+      "step": 9262
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.426125817306874e-05,
+      "loss": 1.2076,
+      "step": 9263
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.424039374046374e-05,
+      "loss": 1.1078,
+      "step": 9264
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.421953435040564e-05,
+      "loss": 1.4306,
+      "step": 9265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4198680004493964e-05,
+      "loss": 1.563,
+      "step": 9266
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.41778307043279e-05,
+      "loss": 0.8982,
+      "step": 9267
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.415698645150612e-05,
+      "loss": 1.3312,
+      "step": 9268
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.413614724762706e-05,
+      "loss": 1.2314,
+      "step": 9269
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4115313094288635e-05,
+      "loss": 1.2663,
+      "step": 9270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4094483993088536e-05,
+      "loss": 1.5621,
+      "step": 9271
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4073659945623937e-05,
+      "loss": 1.3203,
+      "step": 9272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4052840953491615e-05,
+      "loss": 0.8273,
+      "step": 9273
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.403202701828803e-05,
+      "loss": 1.2855,
+      "step": 9274
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.401121814160921e-05,
+      "loss": 1.1598,
+      "step": 9275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.399041432505084e-05,
+      "loss": 1.4837,
+      "step": 9276
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.396961557020817e-05,
+      "loss": 0.9725,
+      "step": 9277
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.394882187867612e-05,
+      "loss": 1.0938,
+      "step": 9278
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.392803325204911e-05,
+      "loss": 1.269,
+      "step": 9279
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3907249691921286e-05,
+      "loss": 0.9725,
+      "step": 9280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.388647119988635e-05,
+      "loss": 1.4771,
+      "step": 9281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.386569777753764e-05,
+      "loss": 1.2422,
+      "step": 9282
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.384492942646808e-05,
+      "loss": 1.2733,
+      "step": 9283
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3824166148270245e-05,
+      "loss": 0.9816,
+      "step": 9284
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3803407944536267e-05,
+      "loss": 1.2674,
+      "step": 9285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.378265481685794e-05,
+      "loss": 0.898,
+      "step": 9286
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.376190676682662e-05,
+      "loss": 0.998,
+      "step": 9287
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.374116379603335e-05,
+      "loss": 1.5429,
+      "step": 9288
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3720425906068655e-05,
+      "loss": 1.1744,
+      "step": 9289
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.36996930985228e-05,
+      "loss": 1.2959,
+      "step": 9290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3678965374985584e-05,
+      "loss": 1.5216,
+      "step": 9291
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.365824273704646e-05,
+      "loss": 1.3012,
+      "step": 9292
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3637525186294474e-05,
+      "loss": 1.379,
+      "step": 9293
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.361681272431825e-05,
+      "loss": 1.0235,
+      "step": 9294
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.359610535270609e-05,
+      "loss": 1.3973,
+      "step": 9295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.357540307304583e-05,
+      "loss": 1.6582,
+      "step": 9296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.355470588692497e-05,
+      "loss": 1.3315,
+      "step": 9297
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.353401379593062e-05,
+      "loss": 0.9487,
+      "step": 9298
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.351332680164948e-05,
+      "loss": 1.1425,
+      "step": 9299
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3492644905667826e-05,
+      "loss": 1.4014,
+      "step": 9300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3471968109571586e-05,
+      "loss": 1.3119,
+      "step": 9301
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.345129641494629e-05,
+      "loss": 1.1782,
+      "step": 9302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.343062982337709e-05,
+      "loss": 1.1054,
+      "step": 9303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.340996833644875e-05,
+      "loss": 1.7405,
+      "step": 9304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.338931195574557e-05,
+      "loss": 1.2358,
+      "step": 9305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3368660682851536e-05,
+      "loss": 0.8153,
+      "step": 9306
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3348014519350233e-05,
+      "loss": 0.9893,
+      "step": 9307
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.332737346682478e-05,
+      "loss": 0.3824,
+      "step": 9308
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3306737526858104e-05,
+      "loss": 0.8015,
+      "step": 9309
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.328610670103247e-05,
+      "loss": 1.3322,
+      "step": 9310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.326548099092992e-05,
+      "loss": 1.4386,
+      "step": 9311
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.324486039813207e-05,
+      "loss": 1.3749,
+      "step": 9312
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.322424492422014e-05,
+      "loss": 1.689,
+      "step": 9313
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.320363457077494e-05,
+      "loss": 1.6515,
+      "step": 9314
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.318302933937697e-05,
+      "loss": 1.04,
+      "step": 9315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.316242923160616e-05,
+      "loss": 1.5816,
+      "step": 9316
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3141834249042237e-05,
+      "loss": 0.7861,
+      "step": 9317
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.312124439326442e-05,
+      "loss": 1.0745,
+      "step": 9318
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3100659665851584e-05,
+      "loss": 1.7062,
+      "step": 9319
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.308008006838219e-05,
+      "loss": 0.9816,
+      "step": 9320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3059505602434326e-05,
+      "loss": 1.3519,
+      "step": 9321
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3038936269585676e-05,
+      "loss": 1.615,
+      "step": 9322
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.30183720714135e-05,
+      "loss": 1.3897,
+      "step": 9323
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2997813009494724e-05,
+      "loss": 1.2866,
+      "step": 9324
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2977259085405866e-05,
+      "loss": 0.9946,
+      "step": 9325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.295671030072297e-05,
+      "loss": 1.2729,
+      "step": 9326
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2936166657021785e-05,
+      "loss": 0.7781,
+      "step": 9327
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.291562815587761e-05,
+      "loss": 1.294,
+      "step": 9328
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.289509479886539e-05,
+      "loss": 1.4183,
+      "step": 9329
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.287456658755964e-05,
+      "loss": 1.2579,
+      "step": 9330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.285404352353451e-05,
+      "loss": 1.0547,
+      "step": 9331
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.283352560836371e-05,
+      "loss": 1.1599,
+      "step": 9332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.281301284362063e-05,
+      "loss": 1.8374,
+      "step": 9333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.279250523087818e-05,
+      "loss": 1.3786,
+      "step": 9334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2772002771708946e-05,
+      "loss": 1.1508,
+      "step": 9335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2751505467685095e-05,
+      "loss": 1.4211,
+      "step": 9336
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.273101332037833e-05,
+      "loss": 1.0243,
+      "step": 9337
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.271052633136007e-05,
+      "loss": 1.0722,
+      "step": 9338
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.269004450220127e-05,
+      "loss": 1.0409,
+      "step": 9339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2669567834472526e-05,
+      "loss": 0.909,
+      "step": 9340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.264909632974403e-05,
+      "loss": 1.0701,
+      "step": 9341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.262862998958552e-05,
+      "loss": 1.3196,
+      "step": 9342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.260816881556642e-05,
+      "loss": 1.4795,
+      "step": 9343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.258771280925571e-05,
+      "loss": 1.3758,
+      "step": 9344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2567261972221966e-05,
+      "loss": 0.9536,
+      "step": 9345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.254681630603349e-05,
+      "loss": 1.6328,
+      "step": 9346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.252637581225797e-05,
+      "loss": 1.4814,
+      "step": 9347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2505940492462875e-05,
+      "loss": 1.1861,
+      "step": 9348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.24855103482152e-05,
+      "loss": 0.9866,
+      "step": 9349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2465085381081554e-05,
+      "loss": 1.3858,
+      "step": 9350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.244466559262816e-05,
+      "loss": 0.9556,
+      "step": 9351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2424250984420875e-05,
+      "loss": 0.9528,
+      "step": 9352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2403841558025064e-05,
+      "loss": 0.9916,
+      "step": 9353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.238343731500577e-05,
+      "loss": 1.5162,
+      "step": 9354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2363038256927636e-05,
+      "loss": 1.042,
+      "step": 9355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.234264438535487e-05,
+      "loss": 1.0775,
+      "step": 9356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2322255701851335e-05,
+      "loss": 1.245,
+      "step": 9357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.230187220798044e-05,
+      "loss": 1.2579,
+      "step": 9358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.228149390530525e-05,
+      "loss": 1.2649,
+      "step": 9359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2261120795388376e-05,
+      "loss": 1.0498,
+      "step": 9360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.224075287979207e-05,
+      "loss": 0.9505,
+      "step": 9361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2220390160078216e-05,
+      "loss": 1.1365,
+      "step": 9362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2200032637808184e-05,
+      "loss": 1.0342,
+      "step": 9363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.217968031454305e-05,
+      "loss": 1.3425,
+      "step": 9364
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2159333191843475e-05,
+      "loss": 1.7654,
+      "step": 9365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2138991271269695e-05,
+      "loss": 1.0204,
+      "step": 9366
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2118654554381555e-05,
+      "loss": 1.4026,
+      "step": 9367
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.209832304273855e-05,
+      "loss": 0.6488,
+      "step": 9368
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.207799673789964e-05,
+      "loss": 1.4151,
+      "step": 9369
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2057675641423554e-05,
+      "loss": 1.6375,
+      "step": 9370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2037359754868526e-05,
+      "loss": 1.0888,
+      "step": 9371
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.20170490797924e-05,
+      "loss": 1.0612,
+      "step": 9372
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1996743617752664e-05,
+      "loss": 1.3934,
+      "step": 9373
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1976443370306316e-05,
+      "loss": 1.4693,
+      "step": 9374
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.195614833901005e-05,
+      "loss": 1.2372,
+      "step": 9375
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.193585852542009e-05,
+      "loss": 1.2747,
+      "step": 9376
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.191557393109231e-05,
+      "loss": 0.9455,
+      "step": 9377
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.18952945575822e-05,
+      "loss": 1.5365,
+      "step": 9378
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.187502040644473e-05,
+      "loss": 1.0619,
+      "step": 9379
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1854751479234604e-05,
+      "loss": 1.2524,
+      "step": 9380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.183448777750606e-05,
+      "loss": 1.2275,
+      "step": 9381
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.181422930281296e-05,
+      "loss": 0.9842,
+      "step": 9382
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1793976056708715e-05,
+      "loss": 0.9354,
+      "step": 9383
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1773728040746486e-05,
+      "loss": 1.6654,
+      "step": 9384
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.17534852564788e-05,
+      "loss": 0.5046,
+      "step": 9385
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.173324770545796e-05,
+      "loss": 1.558,
+      "step": 9386
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1713015389235804e-05,
+      "loss": 1.0636,
+      "step": 9387
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.169278830936378e-05,
+      "loss": 1.4429,
+      "step": 9388
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.167256646739294e-05,
+      "loss": 1.0872,
+      "step": 9389
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1652349864873896e-05,
+      "loss": 1.1069,
+      "step": 9390
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.16321385033569e-05,
+      "loss": 0.5978,
+      "step": 9391
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1611932384391794e-05,
+      "loss": 0.8804,
+      "step": 9392
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1591731509528024e-05,
+      "loss": 1.3471,
+      "step": 9393
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.157153588031461e-05,
+      "loss": 0.9243,
+      "step": 9394
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1551345498300186e-05,
+      "loss": 0.5519,
+      "step": 9395
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.153116036503299e-05,
+      "loss": 1.3472,
+      "step": 9396
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.151098048206084e-05,
+      "loss": 0.9988,
+      "step": 9397
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.149080585093116e-05,
+      "loss": 1.3437,
+      "step": 9398
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.147063647319101e-05,
+      "loss": 1.5571,
+      "step": 9399
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.145047235038694e-05,
+      "loss": 1.4429,
+      "step": 9400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1430313484065186e-05,
+      "loss": 0.9867,
+      "step": 9401
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.14101598757716e-05,
+      "loss": 1.3373,
+      "step": 9402
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1390011527051546e-05,
+      "loss": 1.5732,
+      "step": 9403
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.136986843945004e-05,
+      "loss": 1.3956,
+      "step": 9404
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.134973061451174e-05,
+      "loss": 1.4725,
+      "step": 9405
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.132959805378074e-05,
+      "loss": 1.1119,
+      "step": 9406
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.130947075880092e-05,
+      "loss": 1.2297,
+      "step": 9407
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1289348731115645e-05,
+      "loss": 1.1576,
+      "step": 9408
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.12692319722679e-05,
+      "loss": 0.3482,
+      "step": 9409
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.124912048380031e-05,
+      "loss": 1.2127,
+      "step": 9410
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.122901426725499e-05,
+      "loss": 1.1001,
+      "step": 9411
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1208913324173726e-05,
+      "loss": 0.7946,
+      "step": 9412
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.118881765609791e-05,
+      "loss": 1.1814,
+      "step": 9413
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.116872726456851e-05,
+      "loss": 1.596,
+      "step": 9414
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.114864215112612e-05,
+      "loss": 1.3243,
+      "step": 9415
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1128562317310815e-05,
+      "loss": 1.0148,
+      "step": 9416
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.110848776466239e-05,
+      "loss": 0.7653,
+      "step": 9417
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.108841849472019e-05,
+      "loss": 1.4172,
+      "step": 9418
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.106835450902318e-05,
+      "loss": 0.7643,
+      "step": 9419
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.104829580910983e-05,
+      "loss": 0.7853,
+      "step": 9420
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.102824239651839e-05,
+      "loss": 1.0956,
+      "step": 9421
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.100819427278648e-05,
+      "loss": 1.0423,
+      "step": 9422
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.098815143945145e-05,
+      "loss": 0.8706,
+      "step": 9423
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.096811389805023e-05,
+      "loss": 1.0688,
+      "step": 9424
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.094808165011932e-05,
+      "loss": 2.0484,
+      "step": 9425
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0928054697194866e-05,
+      "loss": 1.078,
+      "step": 9426
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0908033040812466e-05,
+      "loss": 1.4027,
+      "step": 9427
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0888016682507494e-05,
+      "loss": 0.7705,
+      "step": 9428
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0868005623814785e-05,
+      "loss": 1.3694,
+      "step": 9429
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.084799986626886e-05,
+      "loss": 1.6099,
+      "step": 9430
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.082799941140376e-05,
+      "loss": 1.144,
+      "step": 9431
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0808004260753175e-05,
+      "loss": 0.936,
+      "step": 9432
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0788014415850353e-05,
+      "loss": 1.693,
+      "step": 9433
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0768029878228134e-05,
+      "loss": 1.2119,
+      "step": 9434
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.074805064941898e-05,
+      "loss": 0.9342,
+      "step": 9435
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0728076730954945e-05,
+      "loss": 1.1428,
+      "step": 9436
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.070810812436762e-05,
+      "loss": 0.7622,
+      "step": 9437
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.068814483118823e-05,
+      "loss": 1.4317,
+      "step": 9438
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.066818685294761e-05,
+      "loss": 0.8822,
+      "step": 9439
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0648234191176165e-05,
+      "loss": 1.5948,
+      "step": 9440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0628286847403895e-05,
+      "loss": 1.2912,
+      "step": 9441
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.060834482316042e-05,
+      "loss": 1.4427,
+      "step": 9442
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.058840811997487e-05,
+      "loss": 1.5854,
+      "step": 9443
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.056847673937602e-05,
+      "loss": 1.1585,
+      "step": 9444
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0548550682892284e-05,
+      "loss": 1.1402,
+      "step": 9445
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0528629952051626e-05,
+      "loss": 1.3096,
+      "step": 9446
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.05087145483816e-05,
+      "loss": 1.0516,
+      "step": 9447
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0488804473409292e-05,
+      "loss": 1.4235,
+      "step": 9448
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.046889972866147e-05,
+      "loss": 1.4191,
+      "step": 9449
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0449000315664466e-05,
+      "loss": 1.0393,
+      "step": 9450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0429106235944205e-05,
+      "loss": 0.8691,
+      "step": 9451
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.04092174910262e-05,
+      "loss": 1.0722,
+      "step": 9452
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0389334082435517e-05,
+      "loss": 1.1766,
+      "step": 9453
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.036945601169685e-05,
+      "loss": 1.8997,
+      "step": 9454
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0349583280334504e-05,
+      "loss": 0.8719,
+      "step": 9455
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0329715889872333e-05,
+      "loss": 0.9038,
+      "step": 9456
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0309853841833802e-05,
+      "loss": 1.4723,
+      "step": 9457
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0289997137741975e-05,
+      "loss": 1.1224,
+      "step": 9458
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0270145779119485e-05,
+      "loss": 1.0383,
+      "step": 9459
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0250299767488555e-05,
+      "loss": 1.3698,
+      "step": 9460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0230459104371033e-05,
+      "loss": 1.1665,
+      "step": 9461
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0210623791288308e-05,
+      "loss": 0.9901,
+      "step": 9462
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.019079382976142e-05,
+      "loss": 1.0749,
+      "step": 9463
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0170969221310896e-05,
+      "loss": 1.2186,
+      "step": 9464
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.015114996745696e-05,
+      "loss": 1.8671,
+      "step": 9465
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0131336069719372e-05,
+      "loss": 1.1941,
+      "step": 9466
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.01115275296175e-05,
+      "loss": 1.2083,
+      "step": 9467
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0091724348670282e-05,
+      "loss": 0.8946,
+      "step": 9468
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0071926528396266e-05,
+      "loss": 0.9513,
+      "step": 9469
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0052134070313563e-05,
+      "loss": 1.3742,
+      "step": 9470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0032346975939916e-05,
+      "loss": 2.2821,
+      "step": 9471
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0012565246792603e-05,
+      "loss": 1.0046,
+      "step": 9472
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9992788884388532e-05,
+      "loss": 0.9334,
+      "step": 9473
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9973017890244213e-05,
+      "loss": 0.694,
+      "step": 9474
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9953252265875665e-05,
+      "loss": 1.5846,
+      "step": 9475
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9933492012798548e-05,
+      "loss": 1.3827,
+      "step": 9476
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.991373713252813e-05,
+      "loss": 1.1997,
+      "step": 9477
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9893987626579246e-05,
+      "loss": 1.06,
+      "step": 9478
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9874243496466336e-05,
+      "loss": 1.2644,
+      "step": 9479
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.985450474370336e-05,
+      "loss": 1.7234,
+      "step": 9480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9834771369803914e-05,
+      "loss": 1.4286,
+      "step": 9481
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9815043376281237e-05,
+      "loss": 1.3272,
+      "step": 9482
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9795320764648084e-05,
+      "loss": 1.6312,
+      "step": 9483
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.977560353641684e-05,
+      "loss": 1.4793,
+      "step": 9484
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.975589169309938e-05,
+      "loss": 1.1245,
+      "step": 9485
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.973618523620729e-05,
+      "loss": 1.0414,
+      "step": 9486
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.971648416725167e-05,
+      "loss": 1.6069,
+      "step": 9487
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9696788487743242e-05,
+      "loss": 1.3884,
+      "step": 9488
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9677098199192322e-05,
+      "loss": 0.8936,
+      "step": 9489
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.965741330310874e-05,
+      "loss": 1.204,
+      "step": 9490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9637733801001978e-05,
+      "loss": 1.6215,
+      "step": 9491
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9618059694381118e-05,
+      "loss": 1.244,
+      "step": 9492
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9598390984754765e-05,
+      "loss": 1.5894,
+      "step": 9493
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.957872767363118e-05,
+      "loss": 1.067,
+      "step": 9494
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9559069762518156e-05,
+      "loss": 1.7441,
+      "step": 9495
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9539417252923096e-05,
+      "loss": 1.546,
+      "step": 9496
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9519770146352978e-05,
+      "loss": 1.1555,
+      "step": 9497
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.950012844431439e-05,
+      "loss": 1.0626,
+      "step": 9498
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9480492148313465e-05,
+      "loss": 1.5365,
+      "step": 9499
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9460861259855987e-05,
+      "loss": 1.2107,
+      "step": 9500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.944123578044723e-05,
+      "loss": 1.2598,
+      "step": 9501
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9421615711592122e-05,
+      "loss": 1.0735,
+      "step": 9502
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9402001054795158e-05,
+      "loss": 1.1323,
+      "step": 9503
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9382391811560428e-05,
+      "loss": 1.1385,
+      "step": 9504
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.936278798339159e-05,
+      "loss": 1.8569,
+      "step": 9505
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.934318957179192e-05,
+      "loss": 1.3197,
+      "step": 9506
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9323596578264223e-05,
+      "loss": 1.0404,
+      "step": 9507
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.930400900431094e-05,
+      "loss": 1.0887,
+      "step": 9508
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9284426851434054e-05,
+      "loss": 1.2282,
+      "step": 9509
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.926485012113519e-05,
+      "loss": 1.1838,
+      "step": 9510
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9245278814915524e-05,
+      "loss": 1.5572,
+      "step": 9511
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.922571293427575e-05,
+      "loss": 1.8334,
+      "step": 9512
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9206152480716253e-05,
+      "loss": 0.9375,
+      "step": 9513
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9186597455736964e-05,
+      "loss": 1.4168,
+      "step": 9514
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.916704786083738e-05,
+      "loss": 0.488,
+      "step": 9515
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.914750369751662e-05,
+      "loss": 2.0866,
+      "step": 9516
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9127964967273314e-05,
+      "loss": 0.8168,
+      "step": 9517
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.910843167160575e-05,
+      "loss": 0.8955,
+      "step": 9518
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9088903812011726e-05,
+      "loss": 1.1723,
+      "step": 9519
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9069381389988736e-05,
+      "loss": 1.0736,
+      "step": 9520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.90498644070338e-05,
+      "loss": 1.2779,
+      "step": 9521
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.903035286464344e-05,
+      "loss": 1.0674,
+      "step": 9522
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.901084676431386e-05,
+      "loss": 1.2465,
+      "step": 9523
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8991346107540818e-05,
+      "loss": 1.4644,
+      "step": 9524
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8971850895819663e-05,
+      "loss": 1.1543,
+      "step": 9525
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8952361130645333e-05,
+      "loss": 1.4287,
+      "step": 9526
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.893287681351228e-05,
+      "loss": 1.1886,
+      "step": 9527
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8913397945914623e-05,
+      "loss": 1.6374,
+      "step": 9528
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8893924529346027e-05,
+      "loss": 1.4141,
+      "step": 9529
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8874456565299746e-05,
+      "loss": 1.0588,
+      "step": 9530
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8854994055268613e-05,
+      "loss": 1.2868,
+      "step": 9531
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.883553700074505e-05,
+      "loss": 1.1514,
+      "step": 9532
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8816085403221038e-05,
+      "loss": 1.1681,
+      "step": 9533
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8796639264188165e-05,
+      "loss": 1.5636,
+      "step": 9534
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8777198585137598e-05,
+      "loss": 1.6613,
+      "step": 9535
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8757763367560054e-05,
+      "loss": 1.0428,
+      "step": 9536
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.873833361294591e-05,
+      "loss": 1.1948,
+      "step": 9537
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8718909322785003e-05,
+      "loss": 1.0815,
+      "step": 9538
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.869949049856684e-05,
+      "loss": 1.4909,
+      "step": 9539
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.868007714178048e-05,
+      "loss": 1.542,
+      "step": 9540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8660669253914596e-05,
+      "loss": 1.2456,
+      "step": 9541
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8641266836457382e-05,
+      "loss": 1.0532,
+      "step": 9542
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.862186989089667e-05,
+      "loss": 0.855,
+      "step": 9543
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8602478418719837e-05,
+      "loss": 1.1188,
+      "step": 9544
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.858309242141385e-05,
+      "loss": 1.2508,
+      "step": 9545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.856371190046525e-05,
+      "loss": 1.5391,
+      "step": 9546
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8544336857360175e-05,
+      "loss": 0.8755,
+      "step": 9547
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.852496729358437e-05,
+      "loss": 1.238,
+      "step": 9548
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8505603210623045e-05,
+      "loss": 1.2223,
+      "step": 9549
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.848624460996111e-05,
+      "loss": 1.6637,
+      "step": 9550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8466891493083004e-05,
+      "loss": 1.469,
+      "step": 9551
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8447543861472758e-05,
+      "loss": 1.2162,
+      "step": 9552
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.842820171661401e-05,
+      "loss": 1.5818,
+      "step": 9553
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8408865059989888e-05,
+      "loss": 0.8869,
+      "step": 9554
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8389533893083164e-05,
+      "loss": 1.5804,
+      "step": 9555
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8370208217376183e-05,
+      "loss": 1.5655,
+      "step": 9556
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8350888034350896e-05,
+      "loss": 1.4121,
+      "step": 9557
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8331573345488836e-05,
+      "loss": 1.0069,
+      "step": 9558
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8312264152271006e-05,
+      "loss": 1.3024,
+      "step": 9559
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8292960456178087e-05,
+      "loss": 1.1208,
+      "step": 9560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8273662258690336e-05,
+      "loss": 1.3336,
+      "step": 9561
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.825436956128754e-05,
+      "loss": 0.9737,
+      "step": 9562
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8235082365449127e-05,
+      "loss": 1.0829,
+      "step": 9563
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.821580067265407e-05,
+      "loss": 1.4077,
+      "step": 9564
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8196524484380872e-05,
+      "loss": 1.6681,
+      "step": 9565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8177253802107695e-05,
+      "loss": 1.6409,
+      "step": 9566
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8157988627312226e-05,
+      "loss": 1.1516,
+      "step": 9567
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.813872896147176e-05,
+      "loss": 1.921,
+      "step": 9568
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8119474806063163e-05,
+      "loss": 1.2141,
+      "step": 9569
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8100226162562858e-05,
+      "loss": 1.2463,
+      "step": 9570
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8080983032446882e-05,
+      "loss": 1.0837,
+      "step": 9571
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.806174541719081e-05,
+      "loss": 1.2982,
+      "step": 9572
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8042513318269815e-05,
+      "loss": 1.527,
+      "step": 9573
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.802328673715867e-05,
+      "loss": 1.244,
+      "step": 9574
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8004065675331647e-05,
+      "loss": 1.1087,
+      "step": 9575
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7984850134262674e-05,
+      "loss": 0.9213,
+      "step": 9576
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.796564011542523e-05,
+      "loss": 1.2068,
+      "step": 9577
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7946435620292354e-05,
+      "loss": 0.9661,
+      "step": 9578
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7927236650336695e-05,
+      "loss": 1.2292,
+      "step": 9579
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7908043207030453e-05,
+      "loss": 1.0632,
+      "step": 9580
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7888855291845405e-05,
+      "loss": 0.7741,
+      "step": 9581
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.786967290625292e-05,
+      "loss": 0.9187,
+      "step": 9582
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7850496051723917e-05,
+      "loss": 1.7758,
+      "step": 9583
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7831324729728924e-05,
+      "loss": 2.1028,
+      "step": 9584
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7812158941738054e-05,
+      "loss": 1.2423,
+      "step": 9585
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7792998689220895e-05,
+      "loss": 1.0363,
+      "step": 9586
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7773843973646717e-05,
+      "loss": 1.0342,
+      "step": 9587
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.775469479648435e-05,
+      "loss": 0.9563,
+      "step": 9588
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7735551159202165e-05,
+      "loss": 1.0739,
+      "step": 9589
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7716413063268177e-05,
+      "loss": 1.5937,
+      "step": 9590
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7697280510149836e-05,
+      "loss": 1.8633,
+      "step": 9591
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7678153501314295e-05,
+      "loss": 0.8541,
+      "step": 9592
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.765903203822825e-05,
+      "loss": 1.1958,
+      "step": 9593
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7639916122357924e-05,
+      "loss": 1.382,
+      "step": 9594
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7620805755169255e-05,
+      "loss": 1.0797,
+      "step": 9595
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.760170093812756e-05,
+      "loss": 1.0353,
+      "step": 9596
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.758260167269786e-05,
+      "loss": 1.1974,
+      "step": 9597
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7563507960344702e-05,
+      "loss": 1.3824,
+      "step": 9598
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.754441980253224e-05,
+      "loss": 1.214,
+      "step": 9599
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7525337200724166e-05,
+      "loss": 1.0908,
+      "step": 9600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7506260156383802e-05,
+      "loss": 0.9574,
+      "step": 9601
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.748718867097394e-05,
+      "loss": 1.7534,
+      "step": 9602
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7468122745957047e-05,
+      "loss": 1.3044,
+      "step": 9603
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.744906238279513e-05,
+      "loss": 1.6427,
+      "step": 9604
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7430007582949757e-05,
+      "loss": 0.783,
+      "step": 9605
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7410958347882077e-05,
+      "loss": 1.1895,
+      "step": 9606
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7391914679052832e-05,
+      "loss": 1.0653,
+      "step": 9607
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.73728765779223e-05,
+      "loss": 1.1416,
+      "step": 9608
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.735384404595036e-05,
+      "loss": 1.4421,
+      "step": 9609
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7334817084596454e-05,
+      "loss": 1.5712,
+      "step": 9610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7315795695319636e-05,
+      "loss": 1.2051,
+      "step": 9611
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7296779879578415e-05,
+      "loss": 1.2244,
+      "step": 9612
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.727776963883101e-05,
+      "loss": 1.2602,
+      "step": 9613
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7258764974535135e-05,
+      "loss": 0.9585,
+      "step": 9614
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7239765888148107e-05,
+      "loss": 1.457,
+      "step": 9615
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7220772381126824e-05,
+      "loss": 0.8793,
+      "step": 9616
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7201784454927658e-05,
+      "loss": 1.141,
+      "step": 9617
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7182802111006723e-05,
+      "loss": 1.081,
+      "step": 9618
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7163825350819584e-05,
+      "loss": 1.448,
+      "step": 9619
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.714485417582139e-05,
+      "loss": 1.3219,
+      "step": 9620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7125888587466898e-05,
+      "loss": 1.5837,
+      "step": 9621
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7106928587210445e-05,
+      "loss": 1.4702,
+      "step": 9622
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7087974176505858e-05,
+      "loss": 1.0355,
+      "step": 9623
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7069025356806598e-05,
+      "loss": 1.4056,
+      "step": 9624
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.705008212956571e-05,
+      "loss": 1.7463,
+      "step": 9625
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7031144496235794e-05,
+      "loss": 0.9169,
+      "step": 9626
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7012212458269027e-05,
+      "loss": 1.5279,
+      "step": 9627
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6993286017117102e-05,
+      "loss": 1.0877,
+      "step": 9628
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.697436517423134e-05,
+      "loss": 0.9843,
+      "step": 9629
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6955449931062648e-05,
+      "loss": 0.7689,
+      "step": 9630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.693654028906142e-05,
+      "loss": 1.6645,
+      "step": 9631
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.691763624967778e-05,
+      "loss": 1.2621,
+      "step": 9632
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6898737814361218e-05,
+      "loss": 1.1813,
+      "step": 9633
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.687984498456093e-05,
+      "loss": 1.6011,
+      "step": 9634
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.686095776172566e-05,
+      "loss": 2.0792,
+      "step": 9635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6842076147303684e-05,
+      "loss": 1.0802,
+      "step": 9636
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6823200142742888e-05,
+      "loss": 1.7432,
+      "step": 9637
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6804329749490754e-05,
+      "loss": 1.0094,
+      "step": 9638
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.67854649689942e-05,
+      "loss": 1.6042,
+      "step": 9639
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6766605802699872e-05,
+      "loss": 2.0216,
+      "step": 9640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.674775225205389e-05,
+      "loss": 2.0067,
+      "step": 9641
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6728904318502e-05,
+      "loss": 0.7692,
+      "step": 9642
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6710062003489467e-05,
+      "loss": 1.1239,
+      "step": 9643
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6691225308461154e-05,
+      "loss": 1.4989,
+      "step": 9644
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6672394234861497e-05,
+      "loss": 1.2507,
+      "step": 9645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6653568784134497e-05,
+      "loss": 1.3521,
+      "step": 9646
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6634748957723686e-05,
+      "loss": 1.2313,
+      "step": 9647
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6615934757072268e-05,
+      "loss": 1.2026,
+      "step": 9648
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6597126183622856e-05,
+      "loss": 1.4561,
+      "step": 9649
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.657832323881776e-05,
+      "loss": 0.8466,
+      "step": 9650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6559525924098817e-05,
+      "loss": 1.6954,
+      "step": 9651
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6540734240907427e-05,
+      "loss": 1.1978,
+      "step": 9652
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6521948190684574e-05,
+      "loss": 1.3281,
+      "step": 9653
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6503167774870806e-05,
+      "loss": 2.3847,
+      "step": 9654
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6484392994906226e-05,
+      "loss": 0.7834,
+      "step": 9655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6465623852230513e-05,
+      "loss": 1.5831,
+      "step": 9656
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.644686034828291e-05,
+      "loss": 0.9046,
+      "step": 9657
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6428102484502237e-05,
+      "loss": 1.1532,
+      "step": 9658
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6409350262326915e-05,
+      "loss": 1.4087,
+      "step": 9659
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6390603683194816e-05,
+      "loss": 1.3122,
+      "step": 9660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6371862748543487e-05,
+      "loss": 1.1383,
+      "step": 9661
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6353127459810013e-05,
+      "loss": 1.3535,
+      "step": 9662
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6334397818431055e-05,
+      "loss": 1.1686,
+      "step": 9663
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6315673825842845e-05,
+      "loss": 0.9888,
+      "step": 9664
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6296955483481134e-05,
+      "loss": 0.7932,
+      "step": 9665
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.627824279278127e-05,
+      "loss": 0.9731,
+      "step": 9666
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6259535755178188e-05,
+      "loss": 1.1753,
+      "step": 9667
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6240834372106372e-05,
+      "loss": 1.3327,
+      "step": 9668
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6222138644999873e-05,
+      "loss": 1.2056,
+      "step": 9669
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.62034485752923e-05,
+      "loss": 1.8802,
+      "step": 9670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6184764164416842e-05,
+      "loss": 0.83,
+      "step": 9671
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6166085413806253e-05,
+      "loss": 1.2452,
+      "step": 9672
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.614741232489284e-05,
+      "loss": 1.1978,
+      "step": 9673
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6128744899108482e-05,
+      "loss": 1.1129,
+      "step": 9674
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6110083137884668e-05,
+      "loss": 1.0575,
+      "step": 9675
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6091427042652327e-05,
+      "loss": 1.3305,
+      "step": 9676
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6072776614842097e-05,
+      "loss": 1.2073,
+      "step": 9677
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.60541318558841e-05,
+      "loss": 1.5734,
+      "step": 9678
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6035492767208057e-05,
+      "loss": 1.2496,
+      "step": 9679
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6016859350243227e-05,
+      "loss": 1.3566,
+      "step": 9680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5998231606418466e-05,
+      "loss": 0.9132,
+      "step": 9681
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5979609537162165e-05,
+      "loss": 1.329,
+      "step": 9682
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.59609931439023e-05,
+      "loss": 1.1918,
+      "step": 9683
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5942382428066404e-05,
+      "loss": 1.173,
+      "step": 9684
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5923777391081594e-05,
+      "loss": 1.3416,
+      "step": 9685
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5905178034374488e-05,
+      "loss": 1.5463,
+      "step": 9686
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5886584359371336e-05,
+      "loss": 2.1501,
+      "step": 9687
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5867996367497925e-05,
+      "loss": 1.015,
+      "step": 9688
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5849414060179623e-05,
+      "loss": 1.5981,
+      "step": 9689
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5830837438841337e-05,
+      "loss": 1.2453,
+      "step": 9690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.581226650490758e-05,
+      "loss": 0.9992,
+      "step": 9691
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5793701259802328e-05,
+      "loss": 0.9269,
+      "step": 9692
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5775141704949258e-05,
+      "loss": 0.7715,
+      "step": 9693
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.575658784177153e-05,
+      "loss": 1.0852,
+      "step": 9694
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5738039671691882e-05,
+      "loss": 1.008,
+      "step": 9695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5719497196132636e-05,
+      "loss": 1.2406,
+      "step": 9696
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5700960416515608e-05,
+      "loss": 1.7061,
+      "step": 9697
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.568242933426226e-05,
+      "loss": 1.1366,
+      "step": 9698
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5663903950793567e-05,
+      "loss": 1.754,
+      "step": 9699
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.564538426753009e-05,
+      "loss": 1.7262,
+      "step": 9700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.562687028589199e-05,
+      "loss": 1.0958,
+      "step": 9701
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.560836200729887e-05,
+      "loss": 0.6939,
+      "step": 9702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5589859433170015e-05,
+      "loss": 1.4838,
+      "step": 9703
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5571362564924216e-05,
+      "loss": 1.0212,
+      "step": 9704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.555287140397986e-05,
+      "loss": 1.49,
+      "step": 9705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.553438595175487e-05,
+      "loss": 1.3886,
+      "step": 9706
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.551590620966673e-05,
+      "loss": 0.9599,
+      "step": 9707
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5497432179132508e-05,
+      "loss": 1.4284,
+      "step": 9708
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5478963861568805e-05,
+      "loss": 1.2082,
+      "step": 9709
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5460501258391813e-05,
+      "loss": 1.3449,
+      "step": 9710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.544204437101727e-05,
+      "loss": 1.0391,
+      "step": 9711
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5423593200860507e-05,
+      "loss": 0.9887,
+      "step": 9712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5405147749336323e-05,
+      "loss": 0.8596,
+      "step": 9713
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5386708017859185e-05,
+      "loss": 1.0731,
+      "step": 9714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5368274007843072e-05,
+      "loss": 1.1853,
+      "step": 9715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5349845720701536e-05,
+      "loss": 1.2867,
+      "step": 9716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.533142315784768e-05,
+      "loss": 1.1204,
+      "step": 9717
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5313006320694164e-05,
+      "loss": 1.7263,
+      "step": 9718
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5294595210653248e-05,
+      "loss": 1.1553,
+      "step": 9719
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.527618982913671e-05,
+      "loss": 1.8228,
+      "step": 9720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5257790177555897e-05,
+      "loss": 1.0276,
+      "step": 9721
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5239396257321755e-05,
+      "loss": 0.933,
+      "step": 9722
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5221008069844697e-05,
+      "loss": 1.2753,
+      "step": 9723
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.520262561653479e-05,
+      "loss": 1.5725,
+      "step": 9724
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5184248898801632e-05,
+      "loss": 1.0926,
+      "step": 9725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.516587791805437e-05,
+      "loss": 1.0715,
+      "step": 9726
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.514751267570172e-05,
+      "loss": 1.0905,
+      "step": 9727
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5129153173151998e-05,
+      "loss": 1.1041,
+      "step": 9728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5110799411812946e-05,
+      "loss": 1.3737,
+      "step": 9729
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5092451393092033e-05,
+      "loss": 0.9202,
+      "step": 9730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5074109118396194e-05,
+      "loss": 1.834,
+      "step": 9731
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5055772589131942e-05,
+      "loss": 1.1907,
+      "step": 9732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5037441806705374e-05,
+      "loss": 1.0277,
+      "step": 9733
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5019116772522067e-05,
+      "loss": 2.4838,
+      "step": 9734
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.500079748798725e-05,
+      "loss": 1.5146,
+      "step": 9735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4982483954505664e-05,
+      "loss": 0.7916,
+      "step": 9736
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4964176173481623e-05,
+      "loss": 1.0705,
+      "step": 9737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4945874146319026e-05,
+      "loss": 1.1785,
+      "step": 9738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4927577874421238e-05,
+      "loss": 1.1202,
+      "step": 9739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.490928735919127e-05,
+      "loss": 1.7172,
+      "step": 9740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4891002602031677e-05,
+      "loss": 0.7673,
+      "step": 9741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.487272360434456e-05,
+      "loss": 1.6164,
+      "step": 9742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4854450367531567e-05,
+      "loss": 0.9531,
+      "step": 9743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4836182892993943e-05,
+      "loss": 0.8117,
+      "step": 9744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4817921182132442e-05,
+      "loss": 0.7646,
+      "step": 9745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4799665236347403e-05,
+      "loss": 2.057,
+      "step": 9746
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.478141505703874e-05,
+      "loss": 1.1236,
+      "step": 9747
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.476317064560588e-05,
+      "loss": 0.7736,
+      "step": 9748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4744932003447873e-05,
+      "loss": 0.9794,
+      "step": 9749
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4726699131963226e-05,
+      "loss": 0.9158,
+      "step": 9750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.47084720325501e-05,
+      "loss": 0.9759,
+      "step": 9751
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4690250706606167e-05,
+      "loss": 1.1229,
+      "step": 9752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4672035155528662e-05,
+      "loss": 1.0724,
+      "step": 9753
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.46538253807144e-05,
+      "loss": 1.0328,
+      "step": 9754
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4635621383559716e-05,
+      "loss": 0.8814,
+      "step": 9755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.461742316546052e-05,
+      "loss": 1.54,
+      "step": 9756
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4599230727812296e-05,
+      "loss": 1.0012,
+      "step": 9757
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.458104407201005e-05,
+      "loss": 1.3295,
+      "step": 9758
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4562863199448406e-05,
+      "loss": 1.2821,
+      "step": 9759
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.454468811152143e-05,
+      "loss": 1.0996,
+      "step": 9760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4526518809622845e-05,
+      "loss": 2.2801,
+      "step": 9761
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4508355295145924e-05,
+      "loss": 1.4942,
+      "step": 9762
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4490197569483442e-05,
+      "loss": 1.2558,
+      "step": 9763
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4472045634027775e-05,
+      "loss": 1.09,
+      "step": 9764
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4453899490170872e-05,
+      "loss": 1.1534,
+      "step": 9765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.443575913930415e-05,
+      "loss": 1.0885,
+      "step": 9766
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.441762458281863e-05,
+      "loss": 0.9275,
+      "step": 9767
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4399495822104956e-05,
+      "loss": 1.5649,
+      "step": 9768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4381372858553253e-05,
+      "loss": 1.6827,
+      "step": 9769
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4363255693553245e-05,
+      "loss": 0.8609,
+      "step": 9770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4345144328494107e-05,
+      "loss": 1.208,
+      "step": 9771
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4327038764764686e-05,
+      "loss": 0.9515,
+      "step": 9772
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.430893900375335e-05,
+      "loss": 1.4049,
+      "step": 9773
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.429084504684801e-05,
+      "loss": 1.3748,
+      "step": 9774
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4272756895436176e-05,
+      "loss": 0.9227,
+      "step": 9775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.42546745509048e-05,
+      "loss": 1.4572,
+      "step": 9776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.423659801464051e-05,
+      "loss": 1.0342,
+      "step": 9777
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4218527288029436e-05,
+      "loss": 1.0233,
+      "step": 9778
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4200462372457266e-05,
+      "loss": 1.1407,
+      "step": 9779
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.418240326930924e-05,
+      "loss": 0.8179,
+      "step": 9780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.416434997997017e-05,
+      "loss": 0.8907,
+      "step": 9781
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.41463025058244e-05,
+      "loss": 1.2447,
+      "step": 9782
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4128260848255857e-05,
+      "loss": 1.5635,
+      "step": 9783
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.411022500864798e-05,
+      "loss": 1.5596,
+      "step": 9784
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4092194988383786e-05,
+      "loss": 1.1357,
+      "step": 9785
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4074170788845885e-05,
+      "loss": 1.2853,
+      "step": 9786
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4056152411416344e-05,
+      "loss": 1.2931,
+      "step": 9787
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.403813985747685e-05,
+      "loss": 1.5567,
+      "step": 9788
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.402013312840865e-05,
+      "loss": 1.2382,
+      "step": 9789
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4002132225592534e-05,
+      "loss": 1.1978,
+      "step": 9790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3984137150408814e-05,
+      "loss": 1.4904,
+      "step": 9791
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.396614790423739e-05,
+      "loss": 0.5546,
+      "step": 9792
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3948164488457714e-05,
+      "loss": 1.4056,
+      "step": 9793
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3930186904448783e-05,
+      "loss": 1.3891,
+      "step": 9794
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3912215153589136e-05,
+      "loss": 1.2161,
+      "step": 9795
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3894249237256893e-05,
+      "loss": 1.3876,
+      "step": 9796
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.387628915682967e-05,
+      "loss": 1.5144,
+      "step": 9797
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.38583349136847e-05,
+      "loss": 1.0122,
+      "step": 9798
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.384038650919873e-05,
+      "loss": 0.821,
+      "step": 9799
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3822443944748086e-05,
+      "loss": 1.2925,
+      "step": 9800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3804507221708605e-05,
+      "loss": 2.1991,
+      "step": 9801
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3786576341455767e-05,
+      "loss": 1.2241,
+      "step": 9802
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3768651305364452e-05,
+      "loss": 1.0742,
+      "step": 9803
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3750732114809194e-05,
+      "loss": 1.1964,
+      "step": 9804
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.373281877116412e-05,
+      "loss": 1.0696,
+      "step": 9805
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.371491127580282e-05,
+      "loss": 1.6678,
+      "step": 9806
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.369700963009851e-05,
+      "loss": 1.0281,
+      "step": 9807
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3679113835423826e-05,
+      "loss": 1.0724,
+      "step": 9808
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.366122389315111e-05,
+      "loss": 1.586,
+      "step": 9809
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3643339804652166e-05,
+      "loss": 0.9096,
+      "step": 9810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.362546157129839e-05,
+      "loss": 1.8605,
+      "step": 9811
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3607589194460724e-05,
+      "loss": 1.2144,
+      "step": 9812
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3589722675509608e-05,
+      "loss": 1.1397,
+      "step": 9813
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3571862015815084e-05,
+      "loss": 1.4988,
+      "step": 9814
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3554007216746754e-05,
+      "loss": 0.8254,
+      "step": 9815
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3536158279673737e-05,
+      "loss": 1.6709,
+      "step": 9816
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3518315205964724e-05,
+      "loss": 1.764,
+      "step": 9817
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3500477996987947e-05,
+      "loss": 1.2341,
+      "step": 9818
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3482646654111195e-05,
+      "loss": 0.8302,
+      "step": 9819
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.346482117870179e-05,
+      "loss": 1.4005,
+      "step": 9820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3447001572126626e-05,
+      "loss": 0.8694,
+      "step": 9821
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3429187835752142e-05,
+      "loss": 1.2436,
+      "step": 9822
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3411379970944347e-05,
+      "loss": 0.9187,
+      "step": 9823
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3393577979068714e-05,
+      "loss": 1.1845,
+      "step": 9824
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3375781861490355e-05,
+      "loss": 0.956,
+      "step": 9825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3357991619573905e-05,
+      "loss": 1.3139,
+      "step": 9826
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3340207254683556e-05,
+      "loss": 1.254,
+      "step": 9827
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3322428768183025e-05,
+      "loss": 1.0792,
+      "step": 9828
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3304656161435613e-05,
+      "loss": 1.137,
+      "step": 9829
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3286889435804125e-05,
+      "loss": 1.179,
+      "step": 9830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3269128592650956e-05,
+      "loss": 0.8027,
+      "step": 9831
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3251373633338046e-05,
+      "loss": 1.0247,
+      "step": 9832
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3233624559226852e-05,
+      "loss": 0.8486,
+      "step": 9833
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3215881371678427e-05,
+      "loss": 1.1733,
+      "step": 9834
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3198144072053306e-05,
+      "loss": 0.8868,
+      "step": 9835
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3180412661711637e-05,
+      "loss": 1.2759,
+      "step": 9836
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3162687142013072e-05,
+      "loss": 1.3066,
+      "step": 9837
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3144967514316863e-05,
+      "loss": 1.2614,
+      "step": 9838
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.312725377998177e-05,
+      "loss": 1.2623,
+      "step": 9839
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3109545940366084e-05,
+      "loss": 1.6625,
+      "step": 9840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.309184399682769e-05,
+      "loss": 1.0135,
+      "step": 9841
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3074147950723958e-05,
+      "loss": 0.7664,
+      "step": 9842
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3056457803411914e-05,
+      "loss": 1.0938,
+      "step": 9843
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.303877355624806e-05,
+      "loss": 0.9773,
+      "step": 9844
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3021095210588405e-05,
+      "loss": 0.5268,
+      "step": 9845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3003422767788574e-05,
+      "loss": 1.0294,
+      "step": 9846
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2985756229203713e-05,
+      "loss": 1.0732,
+      "step": 9847
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2968095596188532e-05,
+      "loss": 1.19,
+      "step": 9848
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.295044087009729e-05,
+      "loss": 1.0456,
+      "step": 9849
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2932792052283725e-05,
+      "loss": 1.1408,
+      "step": 9850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.29151491441012e-05,
+      "loss": 1.4231,
+      "step": 9851
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.289751214690261e-05,
+      "loss": 0.8054,
+      "step": 9852
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.287988106204039e-05,
+      "loss": 0.982,
+      "step": 9853
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2862255890866492e-05,
+      "loss": 1.2256,
+      "step": 9854
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2844636634732474e-05,
+      "loss": 1.0409,
+      "step": 9855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.282702329498938e-05,
+      "loss": 0.9803,
+      "step": 9856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.280941587298785e-05,
+      "loss": 1.4007,
+      "step": 9857
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2791814370078035e-05,
+      "loss": 1.0112,
+      "step": 9858
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2774218787609648e-05,
+      "loss": 1.0453,
+      "step": 9859
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.275662912693197e-05,
+      "loss": 2.0168,
+      "step": 9860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2739045389393765e-05,
+      "loss": 1.2558,
+      "step": 9861
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.272146757634338e-05,
+      "loss": 1.2443,
+      "step": 9862
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.270389568912873e-05,
+      "loss": 1.158,
+      "step": 9863
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2686329729097246e-05,
+      "loss": 1.3674,
+      "step": 9864
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2668769697595916e-05,
+      "loss": 1.9001,
+      "step": 9865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2651215595971252e-05,
+      "loss": 0.8433,
+      "step": 9866
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.263366742556936e-05,
+      "loss": 1.115,
+      "step": 9867
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2616125187735847e-05,
+      "loss": 1.4074,
+      "step": 9868
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2598588883815873e-05,
+      "loss": 1.8307,
+      "step": 9869
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.258105851515415e-05,
+      "loss": 0.875,
+      "step": 9870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.256353408309496e-05,
+      "loss": 1.4422,
+      "step": 9871
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.254601558898206e-05,
+      "loss": 1.5501,
+      "step": 9872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.252850303415881e-05,
+      "loss": 0.7905,
+      "step": 9873
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.25109964199681e-05,
+      "loss": 0.9464,
+      "step": 9874
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.249349574775237e-05,
+      "loss": 1.5723,
+      "step": 9875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.247600101885362e-05,
+      "loss": 1.3623,
+      "step": 9876
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.245851223461333e-05,
+      "loss": 1.344,
+      "step": 9877
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.244102939637258e-05,
+      "loss": 1.3679,
+      "step": 9878
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2423552505471955e-05,
+      "loss": 1.7426,
+      "step": 9879
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.240608156325167e-05,
+      "loss": 2.0855,
+      "step": 9880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2388616571051414e-05,
+      "loss": 1.2639,
+      "step": 9881
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2371157530210386e-05,
+      "loss": 1.7164,
+      "step": 9882
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.235370444206739e-05,
+      "loss": 1.2634,
+      "step": 9883
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2336257307960774e-05,
+      "loss": 1.0375,
+      "step": 9884
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2318816129228382e-05,
+      "loss": 1.2262,
+      "step": 9885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.230138090720768e-05,
+      "loss": 1.0405,
+      "step": 9886
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2283951643235568e-05,
+      "loss": 0.8227,
+      "step": 9887
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.226652833864856e-05,
+      "loss": 1.1859,
+      "step": 9888
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2249110994782728e-05,
+      "loss": 1.3755,
+      "step": 9889
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2231699612973634e-05,
+      "loss": 1.2136,
+      "step": 9890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.221429419455643e-05,
+      "loss": 1.0538,
+      "step": 9891
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.219689474086578e-05,
+      "loss": 1.4425,
+      "step": 9892
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.217950125323591e-05,
+      "loss": 0.7489,
+      "step": 9893
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2162113733000566e-05,
+      "loss": 1.0598,
+      "step": 9894
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2144732181493054e-05,
+      "loss": 0.6179,
+      "step": 9895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2127356600046225e-05,
+      "loss": 0.8756,
+      "step": 9896
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2109986989992483e-05,
+      "loss": 1.2952,
+      "step": 9897
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.209262335266371e-05,
+      "loss": 0.7116,
+      "step": 9898
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2075265689391387e-05,
+      "loss": 1.2303,
+      "step": 9899
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2057914001506552e-05,
+      "loss": 1.029,
+      "step": 9900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2040568290339736e-05,
+      "loss": 1.6444,
+      "step": 9901
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2023228557221078e-05,
+      "loss": 1.9259,
+      "step": 9902
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.200589480348012e-05,
+      "loss": 0.8504,
+      "step": 9903
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1988567030446126e-05,
+      "loss": 1.1569,
+      "step": 9904
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1971245239447802e-05,
+      "loss": 1.4649,
+      "step": 9905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.195392943181339e-05,
+      "loss": 1.2244,
+      "step": 9906
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1936619608870713e-05,
+      "loss": 1.2578,
+      "step": 9907
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1919315771947125e-05,
+      "loss": 1.0605,
+      "step": 9908
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1902017922369455e-05,
+      "loss": 1.5321,
+      "step": 9909
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.188472606146417e-05,
+      "loss": 1.2248,
+      "step": 9910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1867440190557222e-05,
+      "loss": 1.2945,
+      "step": 9911
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.185016031097412e-05,
+      "loss": 1.0842,
+      "step": 9912
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1832886424039954e-05,
+      "loss": 1.2506,
+      "step": 9913
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.181561853107924e-05,
+      "loss": 1.232,
+      "step": 9914
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1798356633416138e-05,
+      "loss": 1.0785,
+      "step": 9915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1781100732374326e-05,
+      "loss": 1.0838,
+      "step": 9916
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.176385082927698e-05,
+      "loss": 0.827,
+      "step": 9917
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1746606925446923e-05,
+      "loss": 2.1081,
+      "step": 9918
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.172936902220637e-05,
+      "loss": 1.0073,
+      "step": 9919
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1712137120877175e-05,
+      "loss": 1.1245,
+      "step": 9920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1694911222780712e-05,
+      "loss": 1.8673,
+      "step": 9921
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.167769132923787e-05,
+      "loss": 1.8933,
+      "step": 9922
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.166047744156913e-05,
+      "loss": 1.7597,
+      "step": 9923
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.164326956109447e-05,
+      "loss": 0.7224,
+      "step": 9924
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1626067689133388e-05,
+      "loss": 1.4022,
+      "step": 9925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.160887182700496e-05,
+      "loss": 1.1573,
+      "step": 9926
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.15916819760278e-05,
+      "loss": 1.4457,
+      "step": 9927
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.157449813752006e-05,
+      "loss": 1.0392,
+      "step": 9928
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.15573203127994e-05,
+      "loss": 1.0746,
+      "step": 9929
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1540148503183065e-05,
+      "loss": 1.209,
+      "step": 9930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1522982709987805e-05,
+      "loss": 1.7513,
+      "step": 9931
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.150582293452992e-05,
+      "loss": 1.1521,
+      "step": 9932
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1488669178125254e-05,
+      "loss": 0.9778,
+      "step": 9933
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1471521442089192e-05,
+      "loss": 0.9582,
+      "step": 9934
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.145437972773662e-05,
+      "loss": 1.6474,
+      "step": 9935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1437244036382e-05,
+      "loss": 1.605,
+      "step": 9936
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1420114369339318e-05,
+      "loss": 0.8211,
+      "step": 9937
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1402990727922125e-05,
+      "loss": 1.517,
+      "step": 9938
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.138587311344349e-05,
+      "loss": 1.0945,
+      "step": 9939
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1368761527215964e-05,
+      "loss": 1.1002,
+      "step": 9940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.135165597055174e-05,
+      "loss": 1.4622,
+      "step": 9941
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.13345564447625e-05,
+      "loss": 1.566,
+      "step": 9942
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.131746295115944e-05,
+      "loss": 0.9462,
+      "step": 9943
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1300375491053326e-05,
+      "loss": 1.5945,
+      "step": 9944
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1283294065754477e-05,
+      "loss": 1.6786,
+      "step": 9945
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1266218676572668e-05,
+      "loss": 1.212,
+      "step": 9946
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1249149324817298e-05,
+      "loss": 1.242,
+      "step": 9947
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.123208601179725e-05,
+      "loss": 1.1267,
+      "step": 9948
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1215028738820984e-05,
+      "loss": 1.3397,
+      "step": 9949
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1197977507196508e-05,
+      "loss": 1.5981,
+      "step": 9950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.118093231823127e-05,
+      "loss": 0.3826,
+      "step": 9951
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1163893173232363e-05,
+      "loss": 0.8108,
+      "step": 9952
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1146860073506346e-05,
+      "loss": 1.6712,
+      "step": 9953
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1129833020359347e-05,
+      "loss": 1.0454,
+      "step": 9954
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1112812015097093e-05,
+      "loss": 0.966,
+      "step": 9955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1095797059024702e-05,
+      "loss": 1.0087,
+      "step": 9956
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1078788153446936e-05,
+      "loss": 1.0178,
+      "step": 9957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1061785299668057e-05,
+      "loss": 1.472,
+      "step": 9958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1044788498991876e-05,
+      "loss": 0.841,
+      "step": 9959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1027797752721735e-05,
+      "loss": 1.3256,
+      "step": 9960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1010813062160527e-05,
+      "loss": 1.3115,
+      "step": 9961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0993834428610616e-05,
+      "loss": 1.5629,
+      "step": 9962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0976861853373987e-05,
+      "loss": 1.4765,
+      "step": 9963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.095989533775211e-05,
+      "loss": 1.7606,
+      "step": 9964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0942934883046005e-05,
+      "loss": 1.4229,
+      "step": 9965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.092598049055622e-05,
+      "loss": 1.0106,
+      "step": 9966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0909032161582865e-05,
+      "loss": 1.3517,
+      "step": 9967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.089208989742554e-05,
+      "loss": 1.1666,
+      "step": 9968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.087515369938342e-05,
+      "loss": 0.7974,
+      "step": 9969
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0858223568755186e-05,
+      "loss": 1.5998,
+      "step": 9970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0841299506839095e-05,
+      "loss": 1.1005,
+      "step": 9971
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.082438151493287e-05,
+      "loss": 1.4079,
+      "step": 9972
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.080746959433383e-05,
+      "loss": 1.1748,
+      "step": 9973
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0790563746338788e-05,
+      "loss": 1.5131,
+      "step": 9974
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.077366397224413e-05,
+      "loss": 1.2837,
+      "step": 9975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.075677027334578e-05,
+      "loss": 1.1017,
+      "step": 9976
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0739882650939113e-05,
+      "loss": 0.7955,
+      "step": 9977
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0723001106319105e-05,
+      "loss": 1.2429,
+      "step": 9978
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0706125640780306e-05,
+      "loss": 1.0332,
+      "step": 9979
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.068925625561673e-05,
+      "loss": 0.9137,
+      "step": 9980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0672392952121943e-05,
+      "loss": 1.4205,
+      "step": 9981
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0655535731589083e-05,
+      "loss": 0.8364,
+      "step": 9982
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.063868459531072e-05,
+      "loss": 1.2882,
+      "step": 9983
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.062183954457907e-05,
+      "loss": 0.9203,
+      "step": 9984
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.060500058068582e-05,
+      "loss": 1.6832,
+      "step": 9985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0588167704922222e-05,
+      "loss": 1.3932,
+      "step": 9986
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0571340918579063e-05,
+      "loss": 1.2615,
+      "step": 9987
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0554520222946593e-05,
+      "loss": 1.6049,
+      "step": 9988
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0537705619314684e-05,
+      "loss": 1.4283,
+      "step": 9989
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0520897108972703e-05,
+      "loss": 1.1921,
+      "step": 9990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.050409469320954e-05,
+      "loss": 1.5625,
+      "step": 9991
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0487298373313645e-05,
+      "loss": 1.0404,
+      "step": 9992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0470508150572976e-05,
+      "loss": 1.1447,
+      "step": 9993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0453724026275046e-05,
+      "loss": 1.3511,
+      "step": 9994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.043694600170687e-05,
+      "loss": 0.7948,
+      "step": 9995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.042017407815502e-05,
+      "loss": 0.7656,
+      "step": 9996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0403408256905598e-05,
+      "loss": 1.4643,
+      "step": 9997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0386648539244248e-05,
+      "loss": 1.3684,
+      "step": 9998
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.036989492645609e-05,
+      "loss": 1.3118,
+      "step": 9999
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0353147419825834e-05,
+      "loss": 1.3242,
+      "step": 10000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.033640602063771e-05,
+      "loss": 1.1654,
+      "step": 10001
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0319670730175468e-05,
+      "loss": 1.4192,
+      "step": 10002
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0302941549722398e-05,
+      "loss": 1.0566,
+      "step": 10003
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0286218480561325e-05,
+      "loss": 1.2425,
+      "step": 10004
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.02695015239746e-05,
+      "loss": 1.524,
+      "step": 10005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0252790681244093e-05,
+      "loss": 1.1317,
+      "step": 10006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0236085953651217e-05,
+      "loss": 1.0875,
+      "step": 10007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0219387342476958e-05,
+      "loss": 1.5078,
+      "step": 10008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0202694849001725e-05,
+      "loss": 1.3297,
+      "step": 10009
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0186008474505558e-05,
+      "loss": 0.8611,
+      "step": 10010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.016932822026798e-05,
+      "loss": 1.2461,
+      "step": 10011
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.015265408756808e-05,
+      "loss": 1.5451,
+      "step": 10012
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.013598607768443e-05,
+      "loss": 1.4266,
+      "step": 10013
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0119324191895217e-05,
+      "loss": 1.1017,
+      "step": 10014
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.010266843147799e-05,
+      "loss": 1.4296,
+      "step": 10015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0086018797710025e-05,
+      "loss": 1.6537,
+      "step": 10016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.006937529186803e-05,
+      "loss": 1.1645,
+      "step": 10017
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0052737915228247e-05,
+      "loss": 1.1094,
+      "step": 10018
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.003610666906647e-05,
+      "loss": 1.2416,
+      "step": 10019
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0019481554657972e-05,
+      "loss": 1.052,
+      "step": 10020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.00028625732776e-05,
+      "loss": 0.7143,
+      "step": 10021
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9986249726199746e-05,
+      "loss": 1.2317,
+      "step": 10022
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9969643014698292e-05,
+      "loss": 1.3953,
+      "step": 10023
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9953042440046686e-05,
+      "loss": 0.8863,
+      "step": 10024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.993644800351786e-05,
+      "loss": 0.8949,
+      "step": 10025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9919859706384302e-05,
+      "loss": 1.1252,
+      "step": 10026
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.990327754991803e-05,
+      "loss": 1.1814,
+      "step": 10027
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9886701535390607e-05,
+      "loss": 0.7372,
+      "step": 10028
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9870131664073088e-05,
+      "loss": 1.3171,
+      "step": 10029
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9853567937236094e-05,
+      "loss": 1.1414,
+      "step": 10030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.983701035614973e-05,
+      "loss": 1.5004,
+      "step": 10031
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9820458922083684e-05,
+      "loss": 1.3688,
+      "step": 10032
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9803913636307136e-05,
+      "loss": 0.7679,
+      "step": 10033
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9787374500088796e-05,
+      "loss": 1.478,
+      "step": 10034
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9770841514696937e-05,
+      "loss": 0.9913,
+      "step": 10035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.97543146813993e-05,
+      "loss": 1.3963,
+      "step": 10036
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9737794001463185e-05,
+      "loss": 1.2041,
+      "step": 10037
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.972127947615545e-05,
+      "loss": 1.0021,
+      "step": 10038
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9704771106742437e-05,
+      "loss": 1.2302,
+      "step": 10039
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9688268894490037e-05,
+      "loss": 1.9016,
+      "step": 10040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.967177284066367e-05,
+      "loss": 1.7056,
+      "step": 10041
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9655282946528265e-05,
+      "loss": 2.0918,
+      "step": 10042
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9638799213348314e-05,
+      "loss": 1.5239,
+      "step": 10043
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9622321642387796e-05,
+      "loss": 1.3292,
+      "step": 10044
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.960585023491027e-05,
+      "loss": 1.4087,
+      "step": 10045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.958938499217874e-05,
+      "loss": 1.3562,
+      "step": 10046
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9572925915455796e-05,
+      "loss": 1.1109,
+      "step": 10047
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.955647300600356e-05,
+      "loss": 0.0381,
+      "step": 10048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9540026265083667e-05,
+      "loss": 0.5043,
+      "step": 10049
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9523585693957268e-05,
+      "loss": 1.5053,
+      "step": 10050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9507151293885096e-05,
+      "loss": 1.456,
+      "step": 10051
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9490723066127292e-05,
+      "loss": 1.167,
+      "step": 10052
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.947430101194361e-05,
+      "loss": 1.1938,
+      "step": 10053
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9457885132593356e-05,
+      "loss": 1.0161,
+      "step": 10054
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.944147542933532e-05,
+      "loss": 1.6148,
+      "step": 10055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9425071903427827e-05,
+      "loss": 1.6316,
+      "step": 10056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9408674556128692e-05,
+      "loss": 1.0808,
+      "step": 10057
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.939228338869531e-05,
+      "loss": 1.3249,
+      "step": 10058
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9375898402384573e-05,
+      "loss": 1.2352,
+      "step": 10059
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9359519598452903e-05,
+      "loss": 0.5078,
+      "step": 10060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.934314697815629e-05,
+      "loss": 1.2581,
+      "step": 10061
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9326780542750166e-05,
+      "loss": 1.0264,
+      "step": 10062
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.931042029348954e-05,
+      "loss": 1.3324,
+      "step": 10063
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9294066231628948e-05,
+      "loss": 0.7485,
+      "step": 10064
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.927771835842245e-05,
+      "loss": 1.6417,
+      "step": 10065
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9261376675123612e-05,
+      "loss": 1.0786,
+      "step": 10066
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9245041182985556e-05,
+      "loss": 0.9229,
+      "step": 10067
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9228711883260907e-05,
+      "loss": 0.6964,
+      "step": 10068
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9212388777201816e-05,
+      "loss": 1.0118,
+      "step": 10069
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9196071866059972e-05,
+      "loss": 1.2755,
+      "step": 10070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9179761151086584e-05,
+      "loss": 1.3677,
+      "step": 10071
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9163456633532384e-05,
+      "loss": 1.0237,
+      "step": 10072
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9147158314647594e-05,
+      "loss": 0.8578,
+      "step": 10073
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.913086619568203e-05,
+      "loss": 1.5715,
+      "step": 10074
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9114580277884974e-05,
+      "loss": 0.8503,
+      "step": 10075
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9098300562505266e-05,
+      "loss": 1.0334,
+      "step": 10076
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.908202705079125e-05,
+      "loss": 0.9994,
+      "step": 10077
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906575974399083e-05,
+      "loss": 1.1604,
+      "step": 10078
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9049498643351382e-05,
+      "loss": 1.5923,
+      "step": 10079
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9033243750119832e-05,
+      "loss": 1.5765,
+      "step": 10080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.901699506554264e-05,
+      "loss": 0.5943,
+      "step": 10081
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.900075259086581e-05,
+      "loss": 1.2102,
+      "step": 10082
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.898451632733479e-05,
+      "loss": 1.1902,
+      "step": 10083
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8968286276194602e-05,
+      "loss": 1.1542,
+      "step": 10084
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8952062438689822e-05,
+      "loss": 0.7522,
+      "step": 10085
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8935844816064497e-05,
+      "loss": 1.6475,
+      "step": 10086
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.891963340956222e-05,
+      "loss": 1.4902,
+      "step": 10087
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8903428220426155e-05,
+      "loss": 1.1076,
+      "step": 10088
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8887229249898864e-05,
+      "loss": 1.4674,
+      "step": 10089
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8871036499222518e-05,
+      "loss": 1.3309,
+      "step": 10090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8854849969638854e-05,
+      "loss": 1.0248,
+      "step": 10091
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8838669662389053e-05,
+      "loss": 1.1028,
+      "step": 10092
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8822495578713873e-05,
+      "loss": 1.2958,
+      "step": 10093
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8806327719853512e-05,
+      "loss": 1.1452,
+      "step": 10094
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.879016608704778e-05,
+      "loss": 0.8518,
+      "step": 10095
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8774010681535958e-05,
+      "loss": 1.0732,
+      "step": 10096
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8757861504556885e-05,
+      "loss": 1.4575,
+      "step": 10097
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.874171855734893e-05,
+      "loss": 1.6372,
+      "step": 10098
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8725581841149897e-05,
+      "loss": 1.4249,
+      "step": 10099
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8709451357197206e-05,
+      "loss": 1.279,
+      "step": 10100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8693327106727754e-05,
+      "loss": 1.6798,
+      "step": 10101
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8677209090978e-05,
+      "loss": 1.8859,
+      "step": 10102
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8661097311183874e-05,
+      "loss": 1.3726,
+      "step": 10103
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8644991768580865e-05,
+      "loss": 1.147,
+      "step": 10104
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.862889246440397e-05,
+      "loss": 1.1966,
+      "step": 10105
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8612799399887693e-05,
+      "loss": 1.0735,
+      "step": 10106
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8596712576266096e-05,
+      "loss": 0.9328,
+      "step": 10107
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8580631994772723e-05,
+      "loss": 1.4463,
+      "step": 10108
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.85645576566407e-05,
+      "loss": 0.8365,
+      "step": 10109
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8548489563102567e-05,
+      "loss": 1.1462,
+      "step": 10110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8532427715390478e-05,
+      "loss": 0.6448,
+      "step": 10111
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.851637211473608e-05,
+      "loss": 1.1799,
+      "step": 10112
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8500322762370547e-05,
+      "loss": 1.2241,
+      "step": 10113
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8484279659524574e-05,
+      "loss": 1.159,
+      "step": 10114
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.846824280742835e-05,
+      "loss": 1.3138,
+      "step": 10115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8452212207311614e-05,
+      "loss": 1.6756,
+      "step": 10116
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8436187860403633e-05,
+      "loss": 1.319,
+      "step": 10117
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.842016976793315e-05,
+      "loss": 0.7468,
+      "step": 10118
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8404157931128497e-05,
+      "loss": 0.9298,
+      "step": 10119
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8388152351217434e-05,
+      "loss": 1.4727,
+      "step": 10120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.837215302942733e-05,
+      "loss": 1.6725,
+      "step": 10121
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.835615996698502e-05,
+      "loss": 1.2913,
+      "step": 10122
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.834017316511688e-05,
+      "loss": 1.1526,
+      "step": 10123
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8324192625048808e-05,
+      "loss": 1.065,
+      "step": 10124
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8308218348006246e-05,
+      "loss": 1.2617,
+      "step": 10125
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8292250335214056e-05,
+      "loss": 1.5006,
+      "step": 10126
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.827628858789673e-05,
+      "loss": 1.4044,
+      "step": 10127
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8260333107278216e-05,
+      "loss": 1.3122,
+      "step": 10128
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.824438389458205e-05,
+      "loss": 0.9714,
+      "step": 10129
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8228440951031245e-05,
+      "loss": 0.7885,
+      "step": 10130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8212504277848264e-05,
+      "loss": 0.9524,
+      "step": 10131
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.819657387625521e-05,
+      "loss": 0.8946,
+      "step": 10132
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.818064974747362e-05,
+      "loss": 0.9717,
+      "step": 10133
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8164731892724595e-05,
+      "loss": 1.5908,
+      "step": 10134
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8148820313228775e-05,
+      "loss": 0.8835,
+      "step": 10135
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.813291501020622e-05,
+      "loss": 1.3536,
+      "step": 10136
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8117015984876596e-05,
+      "loss": 1.361,
+      "step": 10137
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8101123238459085e-05,
+      "loss": 1.1683,
+      "step": 10138
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.808523677217233e-05,
+      "loss": 1.2627,
+      "step": 10139
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8069356587234566e-05,
+      "loss": 0.8781,
+      "step": 10140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.80534826848635e-05,
+      "loss": 0.6132,
+      "step": 10141
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.803761506627636e-05,
+      "loss": 0.7669,
+      "step": 10142
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8021753732689895e-05,
+      "loss": 1.1157,
+      "step": 10143
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.800589868532039e-05,
+      "loss": 1.1238,
+      "step": 10144
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7990049925383623e-05,
+      "loss": 1.3534,
+      "step": 10145
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7974207454094926e-05,
+      "loss": 1.0241,
+      "step": 10146
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7958371272669082e-05,
+      "loss": 0.9756,
+      "step": 10147
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7942541382320466e-05,
+      "loss": 1.489,
+      "step": 10148
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7926717784262916e-05,
+      "loss": 1.2396,
+      "step": 10149
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7910900479709823e-05,
+      "loss": 1.2085,
+      "step": 10150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7895089469874073e-05,
+      "loss": 1.4483,
+      "step": 10151
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7879284755968096e-05,
+      "loss": 0.9195,
+      "step": 10152
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7863486339203804e-05,
+      "loss": 1.1463,
+      "step": 10153
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7847694220792655e-05,
+      "loss": 1.5078,
+      "step": 10154
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7831908401945607e-05,
+      "loss": 0.8173,
+      "step": 10155
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.781612888387316e-05,
+      "loss": 0.8484,
+      "step": 10156
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7800355667785275e-05,
+      "loss": 1.2125,
+      "step": 10157
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7784588754891483e-05,
+      "loss": 0.9364,
+      "step": 10158
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7768828146400817e-05,
+      "loss": 0.9414,
+      "step": 10159
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7753073843521817e-05,
+      "loss": 1.8237,
+      "step": 10160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.773732584746256e-05,
+      "loss": 1.8915,
+      "step": 10161
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7721584159430648e-05,
+      "loss": 1.6436,
+      "step": 10162
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7705848780633127e-05,
+      "loss": 1.1867,
+      "step": 10163
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.769011971227662e-05,
+      "loss": 1.6279,
+      "step": 10164
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.767439695556725e-05,
+      "loss": 0.9999,
+      "step": 10165
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7658680511710712e-05,
+      "loss": 1.1218,
+      "step": 10166
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.764297038191216e-05,
+      "loss": 1.1504,
+      "step": 10167
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.762726656737621e-05,
+      "loss": 1.3133,
+      "step": 10168
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.76115690693071e-05,
+      "loss": 0.9872,
+      "step": 10169
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.759587788890853e-05,
+      "loss": 1.7537,
+      "step": 10170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7580193027383717e-05,
+      "loss": 1.1815,
+      "step": 10171
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7564514485935435e-05,
+      "loss": 1.1282,
+      "step": 10172
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7548842265765876e-05,
+      "loss": 0.7011,
+      "step": 10173
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7533176368076854e-05,
+      "loss": 1.4314,
+      "step": 10174
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7517516794069634e-05,
+      "loss": 0.8896,
+      "step": 10175
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.750186354494503e-05,
+      "loss": 1.6005,
+      "step": 10176
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.748621662190334e-05,
+      "loss": 1.5436,
+      "step": 10177
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7470576026144413e-05,
+      "loss": 0.9744,
+      "step": 10178
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7454941758867592e-05,
+      "loss": 1.2767,
+      "step": 10179
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.743931382127173e-05,
+      "loss": 1.0763,
+      "step": 10180
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.74236922145552e-05,
+      "loss": 0.9468,
+      "step": 10181
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7408076939915885e-05,
+      "loss": 0.9056,
+      "step": 10182
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.739246799855123e-05,
+      "loss": 1.1737,
+      "step": 10183
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.737686539165809e-05,
+      "loss": 1.5549,
+      "step": 10184
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.736126912043292e-05,
+      "loss": 1.0589,
+      "step": 10185
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7345679186071683e-05,
+      "loss": 1.0562,
+      "step": 10186
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7330095589769812e-05,
+      "loss": 1.1306,
+      "step": 10187
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7314518332722308e-05,
+      "loss": 1.6675,
+      "step": 10188
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7298947416123634e-05,
+      "loss": 1.3026,
+      "step": 10189
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7283382841167815e-05,
+      "loss": 1.2635,
+      "step": 10190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7267824609048344e-05,
+      "loss": 1.1705,
+      "step": 10191
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7252272720958274e-05,
+      "loss": 1.3433,
+      "step": 10192
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7236727178090152e-05,
+      "loss": 1.2465,
+      "step": 10193
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7221187981635988e-05,
+      "loss": 1.8902,
+      "step": 10194
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.720565513278739e-05,
+      "loss": 1.0405,
+      "step": 10195
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7190128632735424e-05,
+      "loss": 0.9602,
+      "step": 10196
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7174608482670695e-05,
+      "loss": 1.3974,
+      "step": 10197
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7159094683783305e-05,
+      "loss": 1.2984,
+      "step": 10198
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7143587237262915e-05,
+      "loss": 1.4511,
+      "step": 10199
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7128086144298593e-05,
+      "loss": 1.1942,
+      "step": 10200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7112591406079014e-05,
+      "loss": 0.889,
+      "step": 10201
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.709710302379235e-05,
+      "loss": 1.2571,
+      "step": 10202
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7081620998626235e-05,
+      "loss": 1.1816,
+      "step": 10203
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7066145331767934e-05,
+      "loss": 2.0595,
+      "step": 10204
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7050676024404066e-05,
+      "loss": 0.936,
+      "step": 10205
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7035213077720878e-05,
+      "loss": 0.9443,
+      "step": 10206
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7019756492904094e-05,
+      "loss": 1.1982,
+      "step": 10207
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7004306271138925e-05,
+      "loss": 0.7269,
+      "step": 10208
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698886241361016e-05,
+      "loss": 1.018,
+      "step": 10209
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6973424921502012e-05,
+      "loss": 1.3823,
+      "step": 10210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695799379599826e-05,
+      "loss": 1.2443,
+      "step": 10211
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6942569038282208e-05,
+      "loss": 1.3593,
+      "step": 10212
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6927150649536627e-05,
+      "loss": 0.7496,
+      "step": 10213
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.691173863094383e-05,
+      "loss": 0.9443,
+      "step": 10214
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6896332983685647e-05,
+      "loss": 1.2809,
+      "step": 10215
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6880933708943392e-05,
+      "loss": 1.3313,
+      "step": 10216
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6865540807897917e-05,
+      "loss": 1.1179,
+      "step": 10217
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.685015428172956e-05,
+      "loss": 1.026,
+      "step": 10218
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6834774131618182e-05,
+      "loss": 1.0783,
+      "step": 10219
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6819400358743197e-05,
+      "loss": 0.8967,
+      "step": 10220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6804032964283423e-05,
+      "loss": 1.1975,
+      "step": 10221
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.67886719494173e-05,
+      "loss": 1.1087,
+      "step": 10222
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6773317315322713e-05,
+      "loss": 1.2334,
+      "step": 10223
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.675796906317708e-05,
+      "loss": 1.5481,
+      "step": 10224
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.674262719415738e-05,
+      "loss": 1.3655,
+      "step": 10225
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.672729170943994e-05,
+      "loss": 1.4232,
+      "step": 10226
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6711962610200815e-05,
+      "loss": 1.2464,
+      "step": 10227
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.669663989761542e-05,
+      "loss": 0.9794,
+      "step": 10228
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.668132357285872e-05,
+      "loss": 1.264,
+      "step": 10229
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6666013637105204e-05,
+      "loss": 0.9039,
+      "step": 10230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6650710091528886e-05,
+      "loss": 1.7159,
+      "step": 10231
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.663541293730322e-05,
+      "loss": 1.201,
+      "step": 10232
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6620122175601227e-05,
+      "loss": 1.0958,
+      "step": 10233
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.660483780759544e-05,
+      "loss": 1.4927,
+      "step": 10234
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6589559834457868e-05,
+      "loss": 2.3486,
+      "step": 10235
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6574288257360104e-05,
+      "loss": 1.1936,
+      "step": 10236
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.655902307747311e-05,
+      "loss": 1.9098,
+      "step": 10237
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6543764295967502e-05,
+      "loss": 1.091,
+      "step": 10238
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6528511914013324e-05,
+      "loss": 1.1672,
+      "step": 10239
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.651326593278013e-05,
+      "loss": 1.276,
+      "step": 10240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.649802635343709e-05,
+      "loss": 2.0897,
+      "step": 10241
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.648279317715272e-05,
+      "loss": 0.6982,
+      "step": 10242
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6467566405095135e-05,
+      "loss": 1.3262,
+      "step": 10243
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6452346038431976e-05,
+      "loss": 0.9849,
+      "step": 10244
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.643713207833033e-05,
+      "loss": 0.9643,
+      "step": 10245
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6421924525956867e-05,
+      "loss": 1.1223,
+      "step": 10246
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6406723382477683e-05,
+      "loss": 1.3253,
+      "step": 10247
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.639152864905844e-05,
+      "loss": 0.8105,
+      "step": 10248
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6376340326864302e-05,
+      "loss": 1.4722,
+      "step": 10249
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6361158417059918e-05,
+      "loss": 1.7191,
+      "step": 10250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6345982920809476e-05,
+      "loss": 1.7711,
+      "step": 10251
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6330813839276638e-05,
+      "loss": 1.4049,
+      "step": 10252
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6315651173624613e-05,
+      "loss": 1.9604,
+      "step": 10253
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6300494925016087e-05,
+      "loss": 1.8186,
+      "step": 10254
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.628534509461326e-05,
+      "loss": 0.9771,
+      "step": 10255
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6270201683577857e-05,
+      "loss": 0.9148,
+      "step": 10256
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.625506469307111e-05,
+      "loss": 0.8358,
+      "step": 10257
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.62399341242537e-05,
+      "loss": 1.569,
+      "step": 10258
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6224809978285882e-05,
+      "loss": 1.5308,
+      "step": 10259
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6209692256327413e-05,
+      "loss": 0.8116,
+      "step": 10260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.619458095953753e-05,
+      "loss": 1.2454,
+      "step": 10261
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.617947608907503e-05,
+      "loss": 0.8473,
+      "step": 10262
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6164377646098085e-05,
+      "loss": 1.0454,
+      "step": 10263
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6149285631764545e-05,
+      "loss": 0.7809,
+      "step": 10264
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6134200047231674e-05,
+      "loss": 1.2183,
+      "step": 10265
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6119120893656247e-05,
+      "loss": 1.3579,
+      "step": 10266
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6104048172194573e-05,
+      "loss": 1.5055,
+      "step": 10267
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6088981884002452e-05,
+      "loss": 0.9768,
+      "step": 10268
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6073922030235155e-05,
+      "loss": 2.065,
+      "step": 10269
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6058868612047517e-05,
+      "loss": 1.253,
+      "step": 10270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.604382163059386e-05,
+      "loss": 1.421,
+      "step": 10271
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.602878108702801e-05,
+      "loss": 1.3557,
+      "step": 10272
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.601374698250331e-05,
+      "loss": 0.9386,
+      "step": 10273
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.599871931817256e-05,
+      "loss": 1.3182,
+      "step": 10274
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5983698095188138e-05,
+      "loss": 1.0403,
+      "step": 10275
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.596868331470187e-05,
+      "loss": 1.2125,
+      "step": 10276
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.595367497786514e-05,
+      "loss": 0.7535,
+      "step": 10277
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.593867308582878e-05,
+      "loss": 0.5054,
+      "step": 10278
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5923677639743184e-05,
+      "loss": 0.7849,
+      "step": 10279
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5908688640758218e-05,
+      "loss": 0.8291,
+      "step": 10280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5893706090023264e-05,
+      "loss": 1.3375,
+      "step": 10281
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.587872998868719e-05,
+      "loss": 1.2188,
+      "step": 10282
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5863760337898427e-05,
+      "loss": 1.0281,
+      "step": 10283
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5848797138804815e-05,
+      "loss": 1.7824,
+      "step": 10284
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.583384039255379e-05,
+      "loss": 1.3642,
+      "step": 10285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.581889010029224e-05,
+      "loss": 0.8472,
+      "step": 10286
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5803946263166593e-05,
+      "loss": 1.1587,
+      "step": 10287
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5789008882322755e-05,
+      "loss": 1.3646,
+      "step": 10288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5774077958906154e-05,
+      "loss": 1.4834,
+      "step": 10289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5759153494061705e-05,
+      "loss": 1.4682,
+      "step": 10290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5744235488933855e-05,
+      "loss": 1.4012,
+      "step": 10291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.572932394466653e-05,
+      "loss": 0.971,
+      "step": 10292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.571441886240317e-05,
+      "loss": 1.1239,
+      "step": 10293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5699520243286737e-05,
+      "loss": 1.2784,
+      "step": 10294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.568462808845964e-05,
+      "loss": 1.1975,
+      "step": 10295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5669742399063857e-05,
+      "loss": 1.0166,
+      "step": 10296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.565486317624085e-05,
+      "loss": 1.7662,
+      "step": 10297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5639990421131566e-05,
+      "loss": 1.382,
+      "step": 10298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.562512413487651e-05,
+      "loss": 1.3463,
+      "step": 10299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5610264318615597e-05,
+      "loss": 1.2676,
+      "step": 10300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.559541097348829e-05,
+      "loss": 0.8405,
+      "step": 10301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5580564100633632e-05,
+      "loss": 0.9085,
+      "step": 10302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.556572370119006e-05,
+      "loss": 1.4078,
+      "step": 10303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5550889776295586e-05,
+      "loss": 1.1305,
+      "step": 10304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5536062327087686e-05,
+      "loss": 1.2831,
+      "step": 10305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5521241354703343e-05,
+      "loss": 1.2269,
+      "step": 10306
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5506426860279044e-05,
+      "loss": 0.9372,
+      "step": 10307
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5491618844950807e-05,
+      "loss": 0.8772,
+      "step": 10308
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.547681730985412e-05,
+      "loss": 1.5133,
+      "step": 10309
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.546202225612401e-05,
+      "loss": 0.8634,
+      "step": 10310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5447233684894948e-05,
+      "loss": 1.3765,
+      "step": 10311
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5432451597300967e-05,
+      "loss": 1.7337,
+      "step": 10312
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5417675994475577e-05,
+      "loss": 1.144,
+      "step": 10313
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5402906877551782e-05,
+      "loss": 1.0064,
+      "step": 10314
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5388144247662117e-05,
+      "loss": 1.3138,
+      "step": 10315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5373388105938602e-05,
+      "loss": 1.0168,
+      "step": 10316
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5358638453512753e-05,
+      "loss": 0.9263,
+      "step": 10317
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5343895291515598e-05,
+      "loss": 1.1358,
+      "step": 10318
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5329158621077666e-05,
+      "loss": 1.2665,
+      "step": 10319
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5314428443328988e-05,
+      "loss": 1.5228,
+      "step": 10320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5299704759399115e-05,
+      "loss": 1.5075,
+      "step": 10321
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5284987570417043e-05,
+      "loss": 0.8654,
+      "step": 10322
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5270276877511323e-05,
+      "loss": 1.0848,
+      "step": 10323
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5255572681810004e-05,
+      "loss": 1.5392,
+      "step": 10324
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5240874984440611e-05,
+      "loss": 1.0875,
+      "step": 10325
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5226183786530202e-05,
+      "loss": 1.9674,
+      "step": 10326
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5211499089205316e-05,
+      "loss": 0.9858,
+      "step": 10327
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.519682089359199e-05,
+      "loss": 0.6177,
+      "step": 10328
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5182149200815776e-05,
+      "loss": 1.1699,
+      "step": 10329
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.516748401200172e-05,
+      "loss": 1.2458,
+      "step": 10330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.51528253282744e-05,
+      "loss": 1.8797,
+      "step": 10331
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5138173150757807e-05,
+      "loss": 1.5558,
+      "step": 10332
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.512352748057554e-05,
+      "loss": 1.1273,
+      "step": 10333
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5108888318850622e-05,
+      "loss": 1.5186,
+      "step": 10334
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5094255666705625e-05,
+      "loss": 1.5815,
+      "step": 10335
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5079629525262618e-05,
+      "loss": 1.3311,
+      "step": 10336
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5065009895643111e-05,
+      "loss": 1.7065,
+      "step": 10337
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5050396778968168e-05,
+      "loss": 1.1115,
+      "step": 10338
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5035790176358377e-05,
+      "loss": 0.7833,
+      "step": 10339
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.502119008893378e-05,
+      "loss": 2.1828,
+      "step": 10340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5006596517813941e-05,
+      "loss": 1.1192,
+      "step": 10341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4992009464117918e-05,
+      "loss": 2.1039,
+      "step": 10342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4977428928964244e-05,
+      "loss": 1.3247,
+      "step": 10343
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4962854913470992e-05,
+      "loss": 1.3584,
+      "step": 10344
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4948287418755725e-05,
+      "loss": 1.1734,
+      "step": 10345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4933726445935493e-05,
+      "loss": 1.0097,
+      "step": 10346
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4919171996126879e-05,
+      "loss": 1.0955,
+      "step": 10347
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4904624070445894e-05,
+      "loss": 0.9639,
+      "step": 10348
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.489008267000811e-05,
+      "loss": 1.0549,
+      "step": 10349
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4875547795928602e-05,
+      "loss": 0.8888,
+      "step": 10350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4861019449321923e-05,
+      "loss": 1.2807,
+      "step": 10351
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4846497631302113e-05,
+      "loss": 1.2009,
+      "step": 10352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4831982342982731e-05,
+      "loss": 1.3012,
+      "step": 10353
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4817473585476848e-05,
+      "loss": 1.2888,
+      "step": 10354
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4802971359897e-05,
+      "loss": 0.823,
+      "step": 10355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.478847566735525e-05,
+      "loss": 1.3968,
+      "step": 10356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4773986508963156e-05,
+      "loss": 1.0058,
+      "step": 10357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.475950388583177e-05,
+      "loss": 0.8195,
+      "step": 10358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.474502779907161e-05,
+      "loss": 1.3428,
+      "step": 10359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4730558249792748e-05,
+      "loss": 1.2075,
+      "step": 10360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4716095239104722e-05,
+      "loss": 1.7106,
+      "step": 10361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4701638768116587e-05,
+      "loss": 1.2446,
+      "step": 10362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4687188837936894e-05,
+      "loss": 0.6639,
+      "step": 10363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4672745449673663e-05,
+      "loss": 2.0907,
+      "step": 10364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4658308604434456e-05,
+      "loss": 1.2669,
+      "step": 10365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4643878303326297e-05,
+      "loss": 0.7524,
+      "step": 10366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4629454547455734e-05,
+      "loss": 1.7179,
+      "step": 10367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4615037337928827e-05,
+      "loss": 0.9294,
+      "step": 10368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.460062667585106e-05,
+      "loss": 1.6438,
+      "step": 10369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4586222562327479e-05,
+      "loss": 1.4426,
+      "step": 10370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4571824998462625e-05,
+      "loss": 1.2997,
+      "step": 10371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4557433985360525e-05,
+      "loss": 1.2786,
+      "step": 10372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.454304952412472e-05,
+      "loss": 1.3613,
+      "step": 10373
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4528671615858192e-05,
+      "loss": 1.0729,
+      "step": 10374
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.451430026166347e-05,
+      "loss": 1.0893,
+      "step": 10375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4499935462642566e-05,
+      "loss": 1.6856,
+      "step": 10376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.448557721989704e-05,
+      "loss": 0.8091,
+      "step": 10377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4471225534527876e-05,
+      "loss": 1.1778,
+      "step": 10378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4456880407635598e-05,
+      "loss": 1.5303,
+      "step": 10379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4442541840320179e-05,
+      "loss": 0.884,
+      "step": 10380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.442820983368114e-05,
+      "loss": 0.9185,
+      "step": 10381
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4413884388817478e-05,
+      "loss": 1.293,
+      "step": 10382
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4399565506827684e-05,
+      "loss": 1.5992,
+      "step": 10383
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.43852531888098e-05,
+      "loss": 1.2051,
+      "step": 10384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4370947435861249e-05,
+      "loss": 1.3622,
+      "step": 10385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.435664824907904e-05,
+      "loss": 0.9241,
+      "step": 10386
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4342355629559668e-05,
+      "loss": 1.5953,
+      "step": 10387
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4328069578399105e-05,
+      "loss": 1.2314,
+      "step": 10388
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4313790096692835e-05,
+      "loss": 0.9973,
+      "step": 10389
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4299517185535815e-05,
+      "loss": 1.2481,
+      "step": 10390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4285250846022535e-05,
+      "loss": 1.4487,
+      "step": 10391
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4270991079246942e-05,
+      "loss": 1.0246,
+      "step": 10392
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4256737886302507e-05,
+      "loss": 1.6733,
+      "step": 10393
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4242491268282177e-05,
+      "loss": 1.201,
+      "step": 10394
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4228251226278433e-05,
+      "loss": 1.0397,
+      "step": 10395
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4214017761383181e-05,
+      "loss": 1.4183,
+      "step": 10396
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4199790874687879e-05,
+      "loss": 1.38,
+      "step": 10397
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4185570567283479e-05,
+      "loss": 1.1623,
+      "step": 10398
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4171356840260408e-05,
+      "loss": 0.6892,
+      "step": 10399
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4157149694708594e-05,
+      "loss": 1.9823,
+      "step": 10400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4142949131717454e-05,
+      "loss": 1.1414,
+      "step": 10401
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4128755152375928e-05,
+      "loss": 1.0041,
+      "step": 10402
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4114567757772434e-05,
+      "loss": 1.2545,
+      "step": 10403
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.410038694899486e-05,
+      "loss": 1.1667,
+      "step": 10404
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4086212727130655e-05,
+      "loss": 1.0859,
+      "step": 10405
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4072045093266662e-05,
+      "loss": 1.2054,
+      "step": 10406
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.40578840484893e-05,
+      "loss": 0.8168,
+      "step": 10407
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.404372959388447e-05,
+      "loss": 0.8939,
+      "step": 10408
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4029581730537555e-05,
+      "loss": 1.2091,
+      "step": 10409
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4015440459533425e-05,
+      "loss": 1.8371,
+      "step": 10410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4001305781956486e-05,
+      "loss": 1.033,
+      "step": 10411
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3987177698890552e-05,
+      "loss": 1.0295,
+      "step": 10412
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3973056211419e-05,
+      "loss": 1.0904,
+      "step": 10413
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3958941320624718e-05,
+      "loss": 1.1915,
+      "step": 10414
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3944833027590043e-05,
+      "loss": 1.0356,
+      "step": 10415
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3930731333396852e-05,
+      "loss": 0.9793,
+      "step": 10416
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3916636239126413e-05,
+      "loss": 1.0258,
+      "step": 10417
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3902547745859606e-05,
+      "loss": 1.0581,
+      "step": 10418
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3888465854676747e-05,
+      "loss": 1.0631,
+      "step": 10419
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3874390566657669e-05,
+      "loss": 1.0059,
+      "step": 10420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3860321882881688e-05,
+      "loss": 1.0227,
+      "step": 10421
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3846259804427586e-05,
+      "loss": 1.3683,
+      "step": 10422
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3832204332373677e-05,
+      "loss": 1.5544,
+      "step": 10423
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3818155467797755e-05,
+      "loss": 1.3104,
+      "step": 10424
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3804113211777115e-05,
+      "loss": 0.8293,
+      "step": 10425
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3790077565388537e-05,
+      "loss": 1.3662,
+      "step": 10426
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3776048529708296e-05,
+      "loss": 1.605,
+      "step": 10427
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3762026105812153e-05,
+      "loss": 0.9902,
+      "step": 10428
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.374801029477537e-05,
+      "loss": 1.1882,
+      "step": 10429
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3734001097672721e-05,
+      "loss": 0.7387,
+      "step": 10430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3719998515578425e-05,
+      "loss": 1.3353,
+      "step": 10431
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3706002549566266e-05,
+      "loss": 1.194,
+      "step": 10432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3692013200709408e-05,
+      "loss": 1.3448,
+      "step": 10433
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3678030470080616e-05,
+      "loss": 1.2862,
+      "step": 10434
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3664054358752098e-05,
+      "loss": 0.8809,
+      "step": 10435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3650084867795576e-05,
+      "loss": 0.6503,
+      "step": 10436
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3636121998282236e-05,
+      "loss": 1.2026,
+      "step": 10437
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.362216575128279e-05,
+      "loss": 0.7016,
+      "step": 10438
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3608216127867401e-05,
+      "loss": 0.9398,
+      "step": 10439
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3594273129105762e-05,
+      "loss": 1.8342,
+      "step": 10440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3580336756067046e-05,
+      "loss": 0.8187,
+      "step": 10441
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3566407009819938e-05,
+      "loss": 1.3514,
+      "step": 10442
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3552483891432532e-05,
+      "loss": 1.0165,
+      "step": 10443
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3538567401972513e-05,
+      "loss": 1.2338,
+      "step": 10444
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3524657542507002e-05,
+      "loss": 1.475,
+      "step": 10445
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3510754314102658e-05,
+      "loss": 1.4707,
+      "step": 10446
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3496857717825563e-05,
+      "loss": 1.1863,
+      "step": 10447
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3482967754741382e-05,
+      "loss": 1.4148,
+      "step": 10448
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3469084425915168e-05,
+      "loss": 0.8905,
+      "step": 10449
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3455207732411523e-05,
+      "loss": 1.0075,
+      "step": 10450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.344133767529452e-05,
+      "loss": 1.4484,
+      "step": 10451
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3427474255627793e-05,
+      "loss": 1.2635,
+      "step": 10452
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3413617474474382e-05,
+      "loss": 1.1968,
+      "step": 10453
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3399767332896828e-05,
+      "loss": 0.5922,
+      "step": 10454
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3385923831957192e-05,
+      "loss": 1.1457,
+      "step": 10455
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3372086972717024e-05,
+      "loss": 1.5378,
+      "step": 10456
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3358256756237341e-05,
+      "loss": 1.3928,
+      "step": 10457
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3344433183578697e-05,
+      "loss": 1.5251,
+      "step": 10458
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3330616255801054e-05,
+      "loss": 0.8625,
+      "step": 10459
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3316805973963941e-05,
+      "loss": 1.1681,
+      "step": 10460
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3303002339126347e-05,
+      "loss": 1.5354,
+      "step": 10461
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3289205352346768e-05,
+      "loss": 0.7814,
+      "step": 10462
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3275415014683168e-05,
+      "loss": 1.2441,
+      "step": 10463
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3261631327193013e-05,
+      "loss": 0.9416,
+      "step": 10464
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3247854290933249e-05,
+      "loss": 1.1364,
+      "step": 10465
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3234083906960338e-05,
+      "loss": 1.3534,
+      "step": 10466
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3220320176330204e-05,
+      "loss": 1.1667,
+      "step": 10467
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3206563100098268e-05,
+      "loss": 0.9489,
+      "step": 10468
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3192812679319466e-05,
+      "loss": 1.2743,
+      "step": 10469
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3179068915048165e-05,
+      "loss": 1.0476,
+      "step": 10470
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3165331808338265e-05,
+      "loss": 1.4153,
+      "step": 10471
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.315160136024317e-05,
+      "loss": 1.2783,
+      "step": 10472
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3137877571815738e-05,
+      "loss": 1.42,
+      "step": 10473
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3124160444108335e-05,
+      "loss": 1.2283,
+      "step": 10474
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.31104499781728e-05,
+      "loss": 0.9677,
+      "step": 10475
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3096746175060482e-05,
+      "loss": 1.1702,
+      "step": 10476
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3083049035822215e-05,
+      "loss": 1.3588,
+      "step": 10477
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3069358561508294e-05,
+      "loss": 0.815,
+      "step": 10478
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3055674753168578e-05,
+      "loss": 0.8941,
+      "step": 10479
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3041997611852296e-05,
+      "loss": 0.9654,
+      "step": 10480
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3028327138608266e-05,
+      "loss": 1.1003,
+      "step": 10481
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3014663334484756e-05,
+      "loss": 0.8147,
+      "step": 10482
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.300100620052953e-05,
+      "loss": 1.4065,
+      "step": 10483
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.298735573778983e-05,
+      "loss": 1.1614,
+      "step": 10484
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2973711947312416e-05,
+      "loss": 1.0329,
+      "step": 10485
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2960074830143487e-05,
+      "loss": 1.8675,
+      "step": 10486
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.294644438732876e-05,
+      "loss": 1.3291,
+      "step": 10487
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2932820619913433e-05,
+      "loss": 0.8649,
+      "step": 10488
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2919203528942236e-05,
+      "loss": 1.641,
+      "step": 10489
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2905593115459335e-05,
+      "loss": 1.5342,
+      "step": 10490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2891989380508374e-05,
+      "loss": 1.2022,
+      "step": 10491
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2878392325132516e-05,
+      "loss": 1.5819,
+      "step": 10492
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2864801950374416e-05,
+      "loss": 1.5925,
+      "step": 10493
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2851218257276187e-05,
+      "loss": 1.2182,
+      "step": 10494
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2837641246879483e-05,
+      "loss": 2.821,
+      "step": 10495
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2824070920225362e-05,
+      "loss": 1.0325,
+      "step": 10496
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2810507278354445e-05,
+      "loss": 1.3986,
+      "step": 10497
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.27969503223068e-05,
+      "loss": 1.1308,
+      "step": 10498
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2783400053121997e-05,
+      "loss": 1.0218,
+      "step": 10499
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2769856471839103e-05,
+      "loss": 1.2467,
+      "step": 10500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2756319579496656e-05,
+      "loss": 1.6478,
+      "step": 10501
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.274278937713268e-05,
+      "loss": 0.8785,
+      "step": 10502
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2729265865784701e-05,
+      "loss": 2.5348,
+      "step": 10503
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2715749046489712e-05,
+      "loss": 1.039,
+      "step": 10504
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2702238920284215e-05,
+      "loss": 1.0586,
+      "step": 10505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2688735488204195e-05,
+      "loss": 1.4107,
+      "step": 10506
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2675238751285079e-05,
+      "loss": 0.9959,
+      "step": 10507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2661748710561839e-05,
+      "loss": 1.2639,
+      "step": 10508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2648265367068912e-05,
+      "loss": 2.0275,
+      "step": 10509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.263478872184023e-05,
+      "loss": 1.2837,
+      "step": 10510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2621318775909208e-05,
+      "loss": 1.2864,
+      "step": 10511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2607855530308687e-05,
+      "loss": 1.2743,
+      "step": 10512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.259439898607111e-05,
+      "loss": 1.1235,
+      "step": 10513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2580949144228328e-05,
+      "loss": 1.0771,
+      "step": 10514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.256750600581169e-05,
+      "loss": 0.9939,
+      "step": 10515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2554069571852055e-05,
+      "loss": 0.7061,
+      "step": 10516
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.254063984337972e-05,
+      "loss": 0.5022,
+      "step": 10517
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2527216821424515e-05,
+      "loss": 1.2479,
+      "step": 10518
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2513800507015727e-05,
+      "loss": 2.5621,
+      "step": 10519
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2500390901182147e-05,
+      "loss": 1.3374,
+      "step": 10520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2486988004952049e-05,
+      "loss": 1.042,
+      "step": 10521
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2473591819353192e-05,
+      "loss": 1.586,
+      "step": 10522
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2460202345412797e-05,
+      "loss": 1.154,
+      "step": 10523
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2446819584157588e-05,
+      "loss": 0.9024,
+      "step": 10524
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2433443536613786e-05,
+      "loss": 1.3167,
+      "step": 10525
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2420074203807075e-05,
+      "loss": 1.3908,
+      "step": 10526
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2406711586762677e-05,
+      "loss": 1.3932,
+      "step": 10527
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2393355686505216e-05,
+      "loss": 0.8709,
+      "step": 10528
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2380006504058839e-05,
+      "loss": 1.727,
+      "step": 10529
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2366664040447217e-05,
+      "loss": 0.8068,
+      "step": 10530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2353328296693435e-05,
+      "loss": 1.4651,
+      "step": 10531
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2339999273820147e-05,
+      "loss": 1.0011,
+      "step": 10532
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2326676972849383e-05,
+      "loss": 1.6482,
+      "step": 10533
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2313361394802736e-05,
+      "loss": 0.8428,
+      "step": 10534
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.230005254070129e-05,
+      "loss": 1.3136,
+      "step": 10535
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2286750411565563e-05,
+      "loss": 1.3197,
+      "step": 10536
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.227345500841558e-05,
+      "loss": 1.6081,
+      "step": 10537
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2260166332270873e-05,
+      "loss": 1.0786,
+      "step": 10538
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2246884384150425e-05,
+      "loss": 1.7074,
+      "step": 10539
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2233609165072724e-05,
+      "loss": 1.1338,
+      "step": 10540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.222034067605573e-05,
+      "loss": 1.4052,
+      "step": 10541
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2207078918116888e-05,
+      "loss": 1.2733,
+      "step": 10542
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.219382389227316e-05,
+      "loss": 1.596,
+      "step": 10543
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2180575599540899e-05,
+      "loss": 0.7578,
+      "step": 10544
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2167334040936051e-05,
+      "loss": 1.4664,
+      "step": 10545
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.215409921747398e-05,
+      "loss": 1.2483,
+      "step": 10546
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2140871130169562e-05,
+      "loss": 1.5265,
+      "step": 10547
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2127649780037164e-05,
+      "loss": 0.9146,
+      "step": 10548
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2114435168090566e-05,
+      "loss": 1.0629,
+      "step": 10549
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2101227295343132e-05,
+      "loss": 1.183,
+      "step": 10550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2088026162807642e-05,
+      "loss": 1.295,
+      "step": 10551
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2074831771496397e-05,
+      "loss": 1.4728,
+      "step": 10552
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2061644122421167e-05,
+      "loss": 1.5941,
+      "step": 10553
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2048463216593164e-05,
+      "loss": 1.2763,
+      "step": 10554
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2035289055023136e-05,
+      "loss": 1.0755,
+      "step": 10555
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2022121638721318e-05,
+      "loss": 1.3131,
+      "step": 10556
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.200896096869738e-05,
+      "loss": 0.7987,
+      "step": 10557
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1995807045960516e-05,
+      "loss": 1.3185,
+      "step": 10558
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1982659871519409e-05,
+      "loss": 1.9705,
+      "step": 10559
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.196951944638216e-05,
+      "loss": 0.7328,
+      "step": 10560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1956385771556422e-05,
+      "loss": 1.2822,
+      "step": 10561
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.194325884804931e-05,
+      "loss": 1.2174,
+      "step": 10562
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1930138676867376e-05,
+      "loss": 1.055,
+      "step": 10563
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1917025259016778e-05,
+      "loss": 1.2107,
+      "step": 10564
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1903918595503006e-05,
+      "loss": 1.1613,
+      "step": 10565
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1890818687331106e-05,
+      "loss": 1.4519,
+      "step": 10566
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1877725535505613e-05,
+      "loss": 1.4579,
+      "step": 10567
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1864639141030532e-05,
+      "loss": 1.1228,
+      "step": 10568
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.185155950490936e-05,
+      "loss": 1.5658,
+      "step": 10569
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1838486628145018e-05,
+      "loss": 1.4091,
+      "step": 10570
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1825420511739981e-05,
+      "loss": 1.0323,
+      "step": 10571
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1812361156696183e-05,
+      "loss": 1.0468,
+      "step": 10572
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1799308564015032e-05,
+      "loss": 1.0977,
+      "step": 10573
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1786262734697429e-05,
+      "loss": 1.3355,
+      "step": 10574
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1773223669743728e-05,
+      "loss": 1.064,
+      "step": 10575
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1760191370153794e-05,
+      "loss": 1.437,
+      "step": 10576
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1747165836926966e-05,
+      "loss": 1.4985,
+      "step": 10577
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1734147071062063e-05,
+      "loss": 1.4886,
+      "step": 10578
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1721135073557377e-05,
+      "loss": 1.3243,
+      "step": 10579
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.170812984541072e-05,
+      "loss": 0.994,
+      "step": 10580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1695131387619306e-05,
+      "loss": 1.2396,
+      "step": 10581
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1682139701179884e-05,
+      "loss": 1.2246,
+      "step": 10582
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.16691547870887e-05,
+      "loss": 1.2024,
+      "step": 10583
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1656176646341433e-05,
+      "loss": 0.8752,
+      "step": 10584
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.164320527993331e-05,
+      "loss": 0.955,
+      "step": 10585
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.163024068885894e-05,
+      "loss": 0.8828,
+      "step": 10586
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1617282874112467e-05,
+      "loss": 1.6002,
+      "step": 10587
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1604331836687564e-05,
+      "loss": 1.0829,
+      "step": 10588
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1591387577577318e-05,
+      "loss": 1.0115,
+      "step": 10589
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1578450097774307e-05,
+      "loss": 1.3331,
+      "step": 10590
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.156551939827062e-05,
+      "loss": 0.9323,
+      "step": 10591
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1552595480057749e-05,
+      "loss": 1.1725,
+      "step": 10592
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.153967834412677e-05,
+      "loss": 1.1031,
+      "step": 10593
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1526767991468158e-05,
+      "loss": 1.5513,
+      "step": 10594
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1513864423071918e-05,
+      "loss": 1.5419,
+      "step": 10595
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1500967639927529e-05,
+      "loss": 1.3523,
+      "step": 10596
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1488077643023886e-05,
+      "loss": 1.411,
+      "step": 10597
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1475194433349457e-05,
+      "loss": 0.8841,
+      "step": 10598
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1462318011892125e-05,
+      "loss": 1.6587,
+      "step": 10599
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1449448379639283e-05,
+      "loss": 0.7716,
+      "step": 10600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1436585537577794e-05,
+      "loss": 0.9939,
+      "step": 10601
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1423729486694001e-05,
+      "loss": 0.8605,
+      "step": 10602
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1410880227973718e-05,
+      "loss": 0.8122,
+      "step": 10603
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1398037762402258e-05,
+      "loss": 1.3156,
+      "step": 10604
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1385202090964386e-05,
+      "loss": 0.9464,
+      "step": 10605
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1372373214644394e-05,
+      "loss": 1.4751,
+      "step": 10606
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1359551134425972e-05,
+      "loss": 1.3417,
+      "step": 10607
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1346735851292368e-05,
+      "loss": 1.4572,
+      "step": 10608
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.133392736622626e-05,
+      "loss": 1.4235,
+      "step": 10609
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1321125680209833e-05,
+      "loss": 0.8859,
+      "step": 10610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1308330794224731e-05,
+      "loss": 1.1205,
+      "step": 10611
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1295542709252093e-05,
+      "loss": 0.9023,
+      "step": 10612
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1282761426272537e-05,
+      "loss": 0.9015,
+      "step": 10613
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1269986946266142e-05,
+      "loss": 0.9908,
+      "step": 10614
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1257219270212472e-05,
+      "loss": 0.9385,
+      "step": 10615
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1244458399090573e-05,
+      "loss": 1.3851,
+      "step": 10616
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1231704333878989e-05,
+      "loss": 0.7388,
+      "step": 10617
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.121895707555568e-05,
+      "loss": 1.5774,
+      "step": 10618
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1206216625098153e-05,
+      "loss": 1.2685,
+      "step": 10619
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1193482983483361e-05,
+      "loss": 1.0988,
+      "step": 10620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1180756151687732e-05,
+      "loss": 1.5523,
+      "step": 10621
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.116803613068721e-05,
+      "loss": 1.0145,
+      "step": 10622
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1155322921457135e-05,
+      "loss": 1.6278,
+      "step": 10623
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1142616524972372e-05,
+      "loss": 1.4669,
+      "step": 10624
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.112991694220733e-05,
+      "loss": 0.8098,
+      "step": 10625
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1117224174135788e-05,
+      "loss": 0.9955,
+      "step": 10626
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1104538221731065e-05,
+      "loss": 1.3993,
+      "step": 10627
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1091859085965939e-05,
+      "loss": 1.1401,
+      "step": 10628
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1079186767812643e-05,
+      "loss": 1.7183,
+      "step": 10629
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1066521268242924e-05,
+      "loss": 1.3217,
+      "step": 10630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1053862588227992e-05,
+      "loss": 0.9175,
+      "step": 10631
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1041210728738527e-05,
+      "loss": 1.0242,
+      "step": 10632
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.102856569074473e-05,
+      "loss": 0.874,
+      "step": 10633
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1015927475216181e-05,
+      "loss": 0.9743,
+      "step": 10634
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1003296083122027e-05,
+      "loss": 1.0247,
+      "step": 10635
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0990671515430872e-05,
+      "loss": 1.0773,
+      "step": 10636
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0978053773110775e-05,
+      "loss": 1.2159,
+      "step": 10637
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0965442857129282e-05,
+      "loss": 1.1413,
+      "step": 10638
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0952838768453433e-05,
+      "loss": 1.2485,
+      "step": 10639
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0940241508049709e-05,
+      "loss": 1.3272,
+      "step": 10640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0927651076884093e-05,
+      "loss": 1.2797,
+      "step": 10641
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0915067475922048e-05,
+      "loss": 1.0409,
+      "step": 10642
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.090249070612851e-05,
+      "loss": 1.2136,
+      "step": 10643
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0889920768467853e-05,
+      "loss": 1.4015,
+      "step": 10644
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0877357663903986e-05,
+      "loss": 1.1236,
+      "step": 10645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0864801393400259e-05,
+      "loss": 1.516,
+      "step": 10646
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0852251957919501e-05,
+      "loss": 1.51,
+      "step": 10647
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0839709358424021e-05,
+      "loss": 1.525,
+      "step": 10648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0827173595875616e-05,
+      "loss": 1.2155,
+      "step": 10649
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0814644671235552e-05,
+      "loss": 1.3852,
+      "step": 10650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0802122585464546e-05,
+      "loss": 1.3292,
+      "step": 10651
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0789607339522823e-05,
+      "loss": 1.7102,
+      "step": 10652
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0777098934370067e-05,
+      "loss": 0.9532,
+      "step": 10653
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0764597370965457e-05,
+      "loss": 1.7026,
+      "step": 10654
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0752102650267614e-05,
+      "loss": 0.8302,
+      "step": 10655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.073961477323464e-05,
+      "loss": 0.9817,
+      "step": 10656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0727133740824146e-05,
+      "loss": 0.9224,
+      "step": 10657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.07146595539932e-05,
+      "loss": 1.2359,
+      "step": 10658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0702192213698348e-05,
+      "loss": 1.4214,
+      "step": 10659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0689731720895569e-05,
+      "loss": 1.1578,
+      "step": 10660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0677278076540376e-05,
+      "loss": 1.2999,
+      "step": 10661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0664831281587706e-05,
+      "loss": 0.6653,
+      "step": 10662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0652391336992052e-05,
+      "loss": 1.1621,
+      "step": 10663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0639958243707293e-05,
+      "loss": 0.9867,
+      "step": 10664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0627532002686835e-05,
+      "loss": 1.3114,
+      "step": 10665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0615112614883526e-05,
+      "loss": 1.1303,
+      "step": 10666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0602700081249695e-05,
+      "loss": 1.5997,
+      "step": 10667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0590294402737178e-05,
+      "loss": 0.866,
+      "step": 10668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0577895580297236e-05,
+      "loss": 1.1591,
+      "step": 10669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0565503614880667e-05,
+      "loss": 1.1118,
+      "step": 10670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0553118507437665e-05,
+      "loss": 1.1696,
+      "step": 10671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0540740258917958e-05,
+      "loss": 1.0338,
+      "step": 10672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0528368870270734e-05,
+      "loss": 1.029,
+      "step": 10673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.051600434244463e-05,
+      "loss": 0.9767,
+      "step": 10674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.050364667638779e-05,
+      "loss": 0.9533,
+      "step": 10675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0491295873047825e-05,
+      "loss": 1.1489,
+      "step": 10676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0478951933371805e-05,
+      "loss": 1.0074,
+      "step": 10677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0466614858306289e-05,
+      "loss": 2.351,
+      "step": 10678
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0454284648797285e-05,
+      "loss": 0.9311,
+      "step": 10679
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0441961305790315e-05,
+      "loss": 1.396,
+      "step": 10680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0429644830230368e-05,
+      "loss": 1.1276,
+      "step": 10681
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.041733522306183e-05,
+      "loss": 1.7083,
+      "step": 10682
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.040503248522866e-05,
+      "loss": 1.013,
+      "step": 10683
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0392736617674248e-05,
+      "loss": 0.855,
+      "step": 10684
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0380447621341449e-05,
+      "loss": 0.7141,
+      "step": 10685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0368165497172622e-05,
+      "loss": 0.8025,
+      "step": 10686
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0355890246109556e-05,
+      "loss": 1.4518,
+      "step": 10687
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0343621869093556e-05,
+      "loss": 1.0406,
+      "step": 10688
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0331360367065358e-05,
+      "loss": 0.8316,
+      "step": 10689
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0319105740965218e-05,
+      "loss": 1.0848,
+      "step": 10690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0306857991732843e-05,
+      "loss": 1.2014,
+      "step": 10691
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.029461712030737e-05,
+      "loss": 1.0512,
+      "step": 10692
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.028238312762747e-05,
+      "loss": 1.8081,
+      "step": 10693
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0270156014631271e-05,
+      "loss": 1.1039,
+      "step": 10694
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0257935782256356e-05,
+      "loss": 0.8012,
+      "step": 10695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0245722431439819e-05,
+      "loss": 1.2118,
+      "step": 10696
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0233515963118156e-05,
+      "loss": 0.8658,
+      "step": 10697
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0221316378227398e-05,
+      "loss": 1.0148,
+      "step": 10698
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0209123677702992e-05,
+      "loss": 1.2088,
+      "step": 10699
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.019693786247996e-05,
+      "loss": 1.1645,
+      "step": 10700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.01847589334927e-05,
+      "loss": 1.3466,
+      "step": 10701
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0172586891675117e-05,
+      "loss": 1.5791,
+      "step": 10702
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0160421737960556e-05,
+      "loss": 1.1369,
+      "step": 10703
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0148263473281871e-05,
+      "loss": 1.6535,
+      "step": 10704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0136112098571381e-05,
+      "loss": 1.3375,
+      "step": 10705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0123967614760877e-05,
+      "loss": 1.29,
+      "step": 10706
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0111830022781633e-05,
+      "loss": 0.9855,
+      "step": 10707
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0099699323564327e-05,
+      "loss": 0.9403,
+      "step": 10708
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0087575518039194e-05,
+      "loss": 1.8283,
+      "step": 10709
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0075458607135902e-05,
+      "loss": 1.5498,
+      "step": 10710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0063348591783594e-05,
+      "loss": 1.2612,
+      "step": 10711
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0051245472910876e-05,
+      "loss": 1.4917,
+      "step": 10712
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0039149251445844e-05,
+      "loss": 0.7577,
+      "step": 10713
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0027059928316052e-05,
+      "loss": 1.0154,
+      "step": 10714
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.001497750444853e-05,
+      "loss": 1.3495,
+      "step": 10715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0002901980769775e-05,
+      "loss": 0.9057,
+      "step": 10716
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.990833358205754e-06,
+      "loss": 1.2652,
+      "step": 10717
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.978771637681928e-06,
+      "loss": 0.6764,
+      "step": 10718
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.966716820123157e-06,
+      "loss": 0.9775,
+      "step": 10719
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.954668906453868e-06,
+      "loss": 1.6731,
+      "step": 10720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.9426278975979e-06,
+      "loss": 1.2434,
+      "step": 10721
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.930593794478571e-06,
+      "loss": 1.6331,
+      "step": 10722
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.918566598018687e-06,
+      "loss": 1.5359,
+      "step": 10723
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.906546309140497e-06,
+      "loss": 1.0662,
+      "step": 10724
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.894532928765742e-06,
+      "loss": 1.5363,
+      "step": 10725
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.882526457815633e-06,
+      "loss": 1.257,
+      "step": 10726
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.870526897210841e-06,
+      "loss": 1.0842,
+      "step": 10727
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.85853424787152e-06,
+      "loss": 0.9315,
+      "step": 10728
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.846548510717258e-06,
+      "loss": 1.4827,
+      "step": 10729
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.83456968666715e-06,
+      "loss": 1.3279,
+      "step": 10730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.822597776639752e-06,
+      "loss": 1.5526,
+      "step": 10731
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.810632781553087e-06,
+      "loss": 1.3852,
+      "step": 10732
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.798674702324684e-06,
+      "loss": 1.4076,
+      "step": 10733
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.786723539871457e-06,
+      "loss": 0.9615,
+      "step": 10734
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.774779295109847e-06,
+      "loss": 1.2807,
+      "step": 10735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.762841968955772e-06,
+      "loss": 1.4775,
+      "step": 10736
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.750911562324583e-06,
+      "loss": 1.6364,
+      "step": 10737
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.738988076131162e-06,
+      "loss": 1.8566,
+      "step": 10738
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.72707151128981e-06,
+      "loss": 1.4648,
+      "step": 10739
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.715161868714273e-06,
+      "loss": 0.8383,
+      "step": 10740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.703259149317835e-06,
+      "loss": 0.9631,
+      "step": 10741
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.691363354013194e-06,
+      "loss": 1.3263,
+      "step": 10742
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.679474483712537e-06,
+      "loss": 0.8857,
+      "step": 10743
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.667592539327563e-06,
+      "loss": 1.355,
+      "step": 10744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.655717521769336e-06,
+      "loss": 1.3961,
+      "step": 10745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.643849431948481e-06,
+      "loss": 1.1775,
+      "step": 10746
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.63198827077506e-06,
+      "loss": 0.937,
+      "step": 10747
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.620134039158602e-06,
+      "loss": 1.0945,
+      "step": 10748
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.60828673800811e-06,
+      "loss": 0.9476,
+      "step": 10749
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.596446368232049e-06,
+      "loss": 1.0417,
+      "step": 10750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.584612930738357e-06,
+      "loss": 1.0882,
+      "step": 10751
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.572786426434454e-06,
+      "loss": 0.8084,
+      "step": 10752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.560966856227194e-06,
+      "loss": 0.8292,
+      "step": 10753
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.549154221022926e-06,
+      "loss": 1.2022,
+      "step": 10754
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.537348521727497e-06,
+      "loss": 1.6047,
+      "step": 10755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.525549759246122e-06,
+      "loss": 1.7368,
+      "step": 10756
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.51375793448358e-06,
+      "loss": 1.3526,
+      "step": 10757
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.501973048344093e-06,
+      "loss": 1.1803,
+      "step": 10758
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.490195101731326e-06,
+      "loss": 1.3328,
+      "step": 10759
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.478424095548443e-06,
+      "loss": 1.5393,
+      "step": 10760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.46666003069806e-06,
+      "loss": 1.0679,
+      "step": 10761
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.454902908082265e-06,
+      "loss": 1.0872,
+      "step": 10762
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.443152728602612e-06,
+      "loss": 1.1919,
+      "step": 10763
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.431409493160114e-06,
+      "loss": 1.3157,
+      "step": 10764
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.419673202655287e-06,
+      "loss": 1.6422,
+      "step": 10765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.407943857988056e-06,
+      "loss": 1.2731,
+      "step": 10766
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.396221460057852e-06,
+      "loss": 1.5391,
+      "step": 10767
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.384506009763572e-06,
+      "loss": 1.1822,
+      "step": 10768
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.372797508003573e-06,
+      "loss": 1.2561,
+      "step": 10769
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.36109595567568e-06,
+      "loss": 1.3251,
+      "step": 10770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.349401353677211e-06,
+      "loss": 1.4401,
+      "step": 10771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.337713702904881e-06,
+      "loss": 1.7,
+      "step": 10772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.326033004254942e-06,
+      "loss": 0.8481,
+      "step": 10773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.314359258623052e-06,
+      "loss": 1.1718,
+      "step": 10774
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.302692466904438e-06,
+      "loss": 1.5926,
+      "step": 10775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.291032629993701e-06,
+      "loss": 0.8974,
+      "step": 10776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.279379748784912e-06,
+      "loss": 1.6486,
+      "step": 10777
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.267733824171643e-06,
+      "loss": 0.7782,
+      "step": 10778
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.256094857046938e-06,
+      "loss": 0.8158,
+      "step": 10779
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.24446284830326e-06,
+      "loss": 1.2476,
+      "step": 10780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.232837798832616e-06,
+      "loss": 0.8895,
+      "step": 10781
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.221219709526385e-06,
+      "loss": 0.9571,
+      "step": 10782
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.209608581275475e-06,
+      "loss": 1.3282,
+      "step": 10783
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.198004414970252e-06,
+      "loss": 1.1328,
+      "step": 10784
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.186407211500537e-06,
+      "loss": 1.0315,
+      "step": 10785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.174816971755617e-06,
+      "loss": 0.8293,
+      "step": 10786
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.163233696624252e-06,
+      "loss": 1.1829,
+      "step": 10787
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.151657386994673e-06,
+      "loss": 1.2296,
+      "step": 10788
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.140088043754569e-06,
+      "loss": 1.6643,
+      "step": 10789
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.128525667791076e-06,
+      "loss": 1.3598,
+      "step": 10790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.116970259990831e-06,
+      "loss": 1.0735,
+      "step": 10791
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.105421821239945e-06,
+      "loss": 1.0242,
+      "step": 10792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.09388035242391e-06,
+      "loss": 1.1264,
+      "step": 10793
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.082345854427788e-06,
+      "loss": 1.0653,
+      "step": 10794
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.070818328136045e-06,
+      "loss": 0.8295,
+      "step": 10795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.059297774432628e-06,
+      "loss": 1.2982,
+      "step": 10796
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.04778419420096e-06,
+      "loss": 0.9328,
+      "step": 10797
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.036277588323915e-06,
+      "loss": 1.5255,
+      "step": 10798
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.024777957683838e-06,
+      "loss": 1.2865,
+      "step": 10799
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.013285303162544e-06,
+      "loss": 1.7021,
+      "step": 10800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.001799625641294e-06,
+      "loss": 1.1246,
+      "step": 10801
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.99032092600086e-06,
+      "loss": 1.2903,
+      "step": 10802
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.978849205121398e-06,
+      "loss": 0.9719,
+      "step": 10803
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.9673844638826e-06,
+      "loss": 1.6552,
+      "step": 10804
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.955926703163608e-06,
+      "loss": 1.2416,
+      "step": 10805
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.944475923842998e-06,
+      "loss": 1.2589,
+      "step": 10806
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.933032126798857e-06,
+      "loss": 1.6197,
+      "step": 10807
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.921595312908727e-06,
+      "loss": 1.5966,
+      "step": 10808
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.910165483049548e-06,
+      "loss": 1.0332,
+      "step": 10809
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.898742638097812e-06,
+      "loss": 1.3494,
+      "step": 10810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.887326778929438e-06,
+      "loss": 1.2824,
+      "step": 10811
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.875917906419783e-06,
+      "loss": 1.0354,
+      "step": 10812
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.864516021443759e-06,
+      "loss": 1.0616,
+      "step": 10813
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.853121124875619e-06,
+      "loss": 0.8053,
+      "step": 10814
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.841733217589165e-06,
+      "loss": 1.4528,
+      "step": 10815
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.830352300457645e-06,
+      "loss": 0.6882,
+      "step": 10816
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.81897837435376e-06,
+      "loss": 1.3358,
+      "step": 10817
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.807611440149689e-06,
+      "loss": 1.2693,
+      "step": 10818
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.796251498717035e-06,
+      "loss": 1.1279,
+      "step": 10819
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.784898550926923e-06,
+      "loss": 1.2197,
+      "step": 10820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.773552597649904e-06,
+      "loss": 1.2174,
+      "step": 10821
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.762213639756e-06,
+      "loss": 1.2244,
+      "step": 10822
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.750881678114719e-06,
+      "loss": 1.5481,
+      "step": 10823
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.739556713594987e-06,
+      "loss": 1.2949,
+      "step": 10824
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.728238747065242e-06,
+      "loss": 1.6175,
+      "step": 10825
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.716927779393346e-06,
+      "loss": 1.1106,
+      "step": 10826
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.70562381144665e-06,
+      "loss": 0.9768,
+      "step": 10827
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.69432684409196e-06,
+      "loss": 1.3648,
+      "step": 10828
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.68303687819556e-06,
+      "loss": 0.5457,
+      "step": 10829
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.671753914623148e-06,
+      "loss": 0.9061,
+      "step": 10830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.660477954239942e-06,
+      "loss": 0.8545,
+      "step": 10831
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.649208997910585e-06,
+      "loss": 0.6854,
+      "step": 10832
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.637947046499206e-06,
+      "loss": 1.2682,
+      "step": 10833
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.626692100869404e-06,
+      "loss": 1.4044,
+      "step": 10834
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.615444161884179e-06,
+      "loss": 1.2558,
+      "step": 10835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.604203230406094e-06,
+      "loss": 1.1434,
+      "step": 10836
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.592969307297095e-06,
+      "loss": 0.8404,
+      "step": 10837
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.581742393418612e-06,
+      "loss": 1.5558,
+      "step": 10838
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.570522489631572e-06,
+      "loss": 1.0242,
+      "step": 10839
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.559309596796295e-06,
+      "loss": 1.3426,
+      "step": 10840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.548103715772616e-06,
+      "loss": 0.9414,
+      "step": 10841
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.536904847419825e-06,
+      "loss": 1.1085,
+      "step": 10842
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.525712992596657e-06,
+      "loss": 0.8491,
+      "step": 10843
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.514528152161339e-06,
+      "loss": 1.1143,
+      "step": 10844
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.503350326971549e-06,
+      "loss": 1.7431,
+      "step": 10845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.49217951788438e-06,
+      "loss": 0.9776,
+      "step": 10846
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.481015725756447e-06,
+      "loss": 0.8263,
+      "step": 10847
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.46985895144382e-06,
+      "loss": 1.5909,
+      "step": 10848
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.458709195801973e-06,
+      "loss": 0.9014,
+      "step": 10849
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.447566459685963e-06,
+      "loss": 0.9285,
+      "step": 10850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.436430743950174e-06,
+      "loss": 1.3477,
+      "step": 10851
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.425302049448524e-06,
+      "loss": 1.0143,
+      "step": 10852
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.414180377034375e-06,
+      "loss": 0.922,
+      "step": 10853
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.403065727560554e-06,
+      "loss": 0.9793,
+      "step": 10854
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.391958101879372e-06,
+      "loss": 1.1176,
+      "step": 10855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.380857500842542e-06,
+      "loss": 1.0913,
+      "step": 10856
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.3697639253013e-06,
+      "loss": 0.871,
+      "step": 10857
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.358677376106294e-06,
+      "loss": 1.1896,
+      "step": 10858
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.347597854107681e-06,
+      "loss": 1.4044,
+      "step": 10859
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.336525360155056e-06,
+      "loss": 0.8517,
+      "step": 10860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.325459895097466e-06,
+      "loss": 0.9406,
+      "step": 10861
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.314401459783416e-06,
+      "loss": 0.8775,
+      "step": 10862
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.30335005506091e-06,
+      "loss": 1.1857,
+      "step": 10863
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.292305681777369e-06,
+      "loss": 0.9899,
+      "step": 10864
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.28126834077969e-06,
+      "loss": 1.3989,
+      "step": 10865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.270238032914269e-06,
+      "loss": 1.6467,
+      "step": 10866
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.259214759026878e-06,
+      "loss": 1.6222,
+      "step": 10867
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.248198519962812e-06,
+      "loss": 1.4643,
+      "step": 10868
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.237189316566819e-06,
+      "loss": 1.0545,
+      "step": 10869
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.226187149683107e-06,
+      "loss": 1.6087,
+      "step": 10870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.21519202015535e-06,
+      "loss": 1.1912,
+      "step": 10871
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.204203928826614e-06,
+      "loss": 1.155,
+      "step": 10872
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.193222876539542e-06,
+      "loss": 1.1494,
+      "step": 10873
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.182248864136166e-06,
+      "loss": 1.3742,
+      "step": 10874
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.171281892457983e-06,
+      "loss": 0.9934,
+      "step": 10875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.160321962345974e-06,
+      "loss": 1.041,
+      "step": 10876
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.149369074640523e-06,
+      "loss": 1.0786,
+      "step": 10877
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.138423230181535e-06,
+      "loss": 1.0689,
+      "step": 10878
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.127484429808362e-06,
+      "loss": 1.0622,
+      "step": 10879
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.116552674359812e-06,
+      "loss": 1.0725,
+      "step": 10880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.105627964674122e-06,
+      "loss": 1.3078,
+      "step": 10881
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.09471030158907e-06,
+      "loss": 0.7284,
+      "step": 10882
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.083799685941773e-06,
+      "loss": 1.2918,
+      "step": 10883
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.072896118568896e-06,
+      "loss": 1.5003,
+      "step": 10884
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.061999600306558e-06,
+      "loss": 1.0086,
+      "step": 10885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.051110131990302e-06,
+      "loss": 0.8631,
+      "step": 10886
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.040227714455163e-06,
+      "loss": 1.1898,
+      "step": 10887
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.029352348535613e-06,
+      "loss": 1.4884,
+      "step": 10888
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.0184840350656e-06,
+      "loss": 1.264,
+      "step": 10889
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.00762277487851e-06,
+      "loss": 1.7099,
+      "step": 10890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.996768568807212e-06,
+      "loss": 1.0522,
+      "step": 10891
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.985921417684039e-06,
+      "loss": 0.9603,
+      "step": 10892
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.975081322340727e-06,
+      "loss": 1.3734,
+      "step": 10893
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.964248283608522e-06,
+      "loss": 0.8313,
+      "step": 10894
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.953422302318137e-06,
+      "loss": 1.5753,
+      "step": 10895
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.942603379299718e-06,
+      "loss": 1.5515,
+      "step": 10896
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.93179151538287e-06,
+      "loss": 1.1281,
+      "step": 10897
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.920986711396661e-06,
+      "loss": 1.2238,
+      "step": 10898
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.910188968169641e-06,
+      "loss": 1.1497,
+      "step": 10899
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.89939828652977e-06,
+      "loss": 1.0358,
+      "step": 10900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.888614667304505e-06,
+      "loss": 1.0374,
+      "step": 10901
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.877838111320756e-06,
+      "loss": 1.3455,
+      "step": 10902
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.867068619404893e-06,
+      "loss": 0.9062,
+      "step": 10903
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.856306192382712e-06,
+      "loss": 1.1385,
+      "step": 10904
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.845550831079507e-06,
+      "loss": 1.3546,
+      "step": 10905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.834802536320007e-06,
+      "loss": 1.1091,
+      "step": 10906
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.82406130892842e-06,
+      "loss": 1.7487,
+      "step": 10907
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.81332714972841e-06,
+      "loss": 1.7335,
+      "step": 10908
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.802600059543064e-06,
+      "loss": 0.7595,
+      "step": 10909
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.791880039194933e-06,
+      "loss": 1.3014,
+      "step": 10910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.781167089506093e-06,
+      "loss": 2.149,
+      "step": 10911
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.770461211298008e-06,
+      "loss": 1.0239,
+      "step": 10912
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.759762405391635e-06,
+      "loss": 1.4937,
+      "step": 10913
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.74907067260735e-06,
+      "loss": 1.353,
+      "step": 10914
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.73838601376502e-06,
+      "loss": 1.219,
+      "step": 10915
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.727708429683967e-06,
+      "loss": 1.2871,
+      "step": 10916
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.717037921182957e-06,
+      "loss": 1.6979,
+      "step": 10917
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.706374489080226e-06,
+      "loss": 1.4667,
+      "step": 10918
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.695718134193474e-06,
+      "loss": 1.1902,
+      "step": 10919
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.685068857339816e-06,
+      "loss": 1.0886,
+      "step": 10920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.674426659335876e-06,
+      "loss": 1.3069,
+      "step": 10921
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.663791540997722e-06,
+      "loss": 1.7138,
+      "step": 10922
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.653163503140847e-06,
+      "loss": 1.0668,
+      "step": 10923
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.642542546580234e-06,
+      "loss": 1.3773,
+      "step": 10924
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.631928672130329e-06,
+      "loss": 1.1378,
+      "step": 10925
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.621321880605003e-06,
+      "loss": 1.1108,
+      "step": 10926
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.610722172817608e-06,
+      "loss": 1.3774,
+      "step": 10927
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.6001295495809365e-06,
+      "loss": 1.7325,
+      "step": 10928
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.589544011707284e-06,
+      "loss": 1.3598,
+      "step": 10929
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.578965560008311e-06,
+      "loss": 0.6633,
+      "step": 10930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.568394195295225e-06,
+      "loss": 1.5107,
+      "step": 10931
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.557829918378634e-06,
+      "loss": 1.4152,
+      "step": 10932
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.547272730068644e-06,
+      "loss": 1.3654,
+      "step": 10933
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.536722631174775e-06,
+      "loss": 0.7004,
+      "step": 10934
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.526179622506046e-06,
+      "loss": 1.269,
+      "step": 10935
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.5156437048708894e-06,
+      "loss": 0.9789,
+      "step": 10936
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.505114879077235e-06,
+      "loss": 1.4444,
+      "step": 10937
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.494593145932438e-06,
+      "loss": 0.9461,
+      "step": 10938
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.484078506243319e-06,
+      "loss": 1.0964,
+      "step": 10939
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.4735709608161785e-06,
+      "loss": 1.1414,
+      "step": 10940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.463070510456727e-06,
+      "loss": 1.0938,
+      "step": 10941
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.452577155970153e-06,
+      "loss": 1.6092,
+      "step": 10942
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.442090898161113e-06,
+      "loss": 1.4859,
+      "step": 10943
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.431611737833711e-06,
+      "loss": 1.7945,
+      "step": 10944
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.421139675791511e-06,
+      "loss": 0.9449,
+      "step": 10945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.410674712837506e-06,
+      "loss": 0.9377,
+      "step": 10946
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.4002168497741664e-06,
+      "loss": 1.4405,
+      "step": 10947
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.389766087403438e-06,
+      "loss": 0.8233,
+      "step": 10948
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.3793224265266915e-06,
+      "loss": 1.4409,
+      "step": 10949
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.368885867944753e-06,
+      "loss": 1.4616,
+      "step": 10950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.358456412457948e-06,
+      "loss": 1.0491,
+      "step": 10951
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.348034060865983e-06,
+      "loss": 1.3889,
+      "step": 10952
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.3376188139680725e-06,
+      "loss": 1.7278,
+      "step": 10953
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.3272106725628785e-06,
+      "loss": 1.1048,
+      "step": 10954
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.3168096374485074e-06,
+      "loss": 1.9612,
+      "step": 10955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.3064157094225536e-06,
+      "loss": 1.2939,
+      "step": 10956
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.296028889281992e-06,
+      "loss": 1.2274,
+      "step": 10957
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.285649177823317e-06,
+      "loss": 1.6693,
+      "step": 10958
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.2752765758424715e-06,
+      "loss": 1.9122,
+      "step": 10959
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.26491108413484e-06,
+      "loss": 1.6063,
+      "step": 10960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.254552703495265e-06,
+      "loss": 1.4368,
+      "step": 10961
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.244201434718034e-06,
+      "loss": 1.3006,
+      "step": 10962
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.2338572785969004e-06,
+      "loss": 1.154,
+      "step": 10963
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.2235202359250745e-06,
+      "loss": 1.3914,
+      "step": 10964
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.2131903074952215e-06,
+      "loss": 1.292,
+      "step": 10965
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.202867494099463e-06,
+      "loss": 1.2201,
+      "step": 10966
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.192551796529334e-06,
+      "loss": 0.8709,
+      "step": 10967
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.18224321557589e-06,
+      "loss": 1.1636,
+      "step": 10968
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.1719417520295874e-06,
+      "loss": 1.4707,
+      "step": 10969
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.161647406680372e-06,
+      "loss": 1.3103,
+      "step": 10970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.151360180317635e-06,
+      "loss": 0.8507,
+      "step": 10971
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.141080073730188e-06,
+      "loss": 1.2579,
+      "step": 10972
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.130807087706359e-06,
+      "loss": 1.276,
+      "step": 10973
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.120541223033883e-06,
+      "loss": 1.2583,
+      "step": 10974
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.110282480499952e-06,
+      "loss": 1.3702,
+      "step": 10975
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.100030860891238e-06,
+      "loss": 1.5236,
+      "step": 10976
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.0897863649938665e-06,
+      "loss": 0.9821,
+      "step": 10977
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.079548993593355e-06,
+      "loss": 1.5593,
+      "step": 10978
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.0693187474747405e-06,
+      "loss": 0.6659,
+      "step": 10979
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.059095627422507e-06,
+      "loss": 1.6884,
+      "step": 10980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.048879634220562e-06,
+      "loss": 1.5696,
+      "step": 10981
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.038670768652311e-06,
+      "loss": 0.9403,
+      "step": 10982
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.0284690315005504e-06,
+      "loss": 1.9134,
+      "step": 10983
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.0182744235475885e-06,
+      "loss": 1.8199,
+      "step": 10984
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.00808694557512e-06,
+      "loss": 1.1531,
+      "step": 10985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.997906598364401e-06,
+      "loss": 1.1852,
+      "step": 10986
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.987733382696049e-06,
+      "loss": 1.3209,
+      "step": 10987
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.9775672993501625e-06,
+      "loss": 0.8169,
+      "step": 10988
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.967408349106286e-06,
+      "loss": 1.268,
+      "step": 10989
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.957256532743417e-06,
+      "loss": 1.0898,
+      "step": 10990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.947111851040033e-06,
+      "loss": 1.2686,
+      "step": 10991
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.936974304774024e-06,
+      "loss": 1.4547,
+      "step": 10992
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.926843894722779e-06,
+      "loss": 0.9939,
+      "step": 10993
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.9167206216630755e-06,
+      "loss": 0.8063,
+      "step": 10994
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.906604486371204e-06,
+      "loss": 1.1479,
+      "step": 10995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.896495489622878e-06,
+      "loss": 0.9854,
+      "step": 10996
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.886393632193277e-06,
+      "loss": 1.2554,
+      "step": 10997
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.876298914857027e-06,
+      "loss": 1.4989,
+      "step": 10998
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.866211338388195e-06,
+      "loss": 0.6954,
+      "step": 10999
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.856130903560331e-06,
+      "loss": 1.1203,
+      "step": 11000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.8460576111464055e-06,
+      "loss": 1.3137,
+      "step": 11001
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.835991461918856e-06,
+      "loss": 0.9936,
+      "step": 11002
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.825932456649586e-06,
+      "loss": 0.8357,
+      "step": 11003
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.815880596109913e-06,
+      "loss": 1.1284,
+      "step": 11004
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.8058358810706215e-06,
+      "loss": 1.0687,
+      "step": 11005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.795798312301982e-06,
+      "loss": 1.8971,
+      "step": 11006
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.785767890573669e-06,
+      "loss": 0.7517,
+      "step": 11007
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.775744616654844e-06,
+      "loss": 1.4419,
+      "step": 11008
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.765728491314105e-06,
+      "loss": 1.1879,
+      "step": 11009
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.755719515319503e-06,
+      "loss": 0.7219,
+      "step": 11010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.745717689438536e-06,
+      "loss": 1.0519,
+      "step": 11011
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.735723014438166e-06,
+      "loss": 1.8161,
+      "step": 11012
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.725735491084795e-06,
+      "loss": 1.8074,
+      "step": 11013
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.715755120144307e-06,
+      "loss": 1.249,
+      "step": 11014
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.705781902381969e-06,
+      "loss": 1.6642,
+      "step": 11015
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.695815838562558e-06,
+      "loss": 1.3767,
+      "step": 11016
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.685856929450296e-06,
+      "loss": 0.9055,
+      "step": 11017
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.67590517580885e-06,
+      "loss": 1.1844,
+      "step": 11018
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.665960578401331e-06,
+      "loss": 0.8733,
+      "step": 11019
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.656023137990297e-06,
+      "loss": 0.9931,
+      "step": 11020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.646092855337771e-06,
+      "loss": 1.1239,
+      "step": 11021
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.63616973120521e-06,
+      "loss": 1.6208,
+      "step": 11022
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.626253766353563e-06,
+      "loss": 1.0936,
+      "step": 11023
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.616344961543175e-06,
+      "loss": 1.2889,
+      "step": 11024
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.6064433175339055e-06,
+      "loss": 1.4346,
+      "step": 11025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.596548835084993e-06,
+      "loss": 1.3242,
+      "step": 11026
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.586661514955162e-06,
+      "loss": 1.5965,
+      "step": 11027
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.576781357902595e-06,
+      "loss": 1.035,
+      "step": 11028
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.56690836468492e-06,
+      "loss": 1.3787,
+      "step": 11029
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.557042536059232e-06,
+      "loss": 1.5148,
+      "step": 11030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.547183872782014e-06,
+      "loss": 0.961,
+      "step": 11031
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.537332375609273e-06,
+      "loss": 0.7985,
+      "step": 11032
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.527488045296426e-06,
+      "loss": 1.087,
+      "step": 11033
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.517650882598347e-06,
+      "loss": 0.6746,
+      "step": 11034
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.507820888269389e-06,
+      "loss": 1.2489,
+      "step": 11035
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4979980630633044e-06,
+      "loss": 1.1863,
+      "step": 11036
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.488182407733334e-06,
+      "loss": 0.4495,
+      "step": 11037
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.478373923032166e-06,
+      "loss": 1.3072,
+      "step": 11038
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4685726097119094e-06,
+      "loss": 2.5981,
+      "step": 11039
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.458778468524163e-06,
+      "loss": 1.213,
+      "step": 11040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.448991500219959e-06,
+      "loss": 0.9879,
+      "step": 11041
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.439211705549753e-06,
+      "loss": 2.5989,
+      "step": 11042
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.429439085263489e-06,
+      "loss": 1.5244,
+      "step": 11043
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.419673640110546e-06,
+      "loss": 0.9315,
+      "step": 11044
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4099153708397475e-06,
+      "loss": 0.8185,
+      "step": 11045
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4001642781993835e-06,
+      "loss": 1.4658,
+      "step": 11046
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.390420362937177e-06,
+      "loss": 1.8077,
+      "step": 11047
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.380683625800299e-06,
+      "loss": 1.2135,
+      "step": 11048
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.370954067535395e-06,
+      "loss": 1.2159,
+      "step": 11049
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.361231688888525e-06,
+      "loss": 1.2053,
+      "step": 11050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.351516490605236e-06,
+      "loss": 1.5607,
+      "step": 11051
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.341808473430477e-06,
+      "loss": 1.3544,
+      "step": 11052
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.332107638108686e-06,
+      "loss": 0.8089,
+      "step": 11053
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.322413985383746e-06,
+      "loss": 0.9316,
+      "step": 11054
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.3127275159989615e-06,
+      "loss": 1.6167,
+      "step": 11055
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.303048230697151e-06,
+      "loss": 0.9351,
+      "step": 11056
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.293376130220474e-06,
+      "loss": 1.4707,
+      "step": 11057
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.283711215310639e-06,
+      "loss": 1.1423,
+      "step": 11058
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.274053486708764e-06,
+      "loss": 0.8037,
+      "step": 11059
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.2644029451553896e-06,
+      "loss": 1.1725,
+      "step": 11060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.25475959139058e-06,
+      "loss": 0.7549,
+      "step": 11061
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.245123426153787e-06,
+      "loss": 1.1863,
+      "step": 11062
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.235494450183921e-06,
+      "loss": 0.9088,
+      "step": 11063
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.225872664219334e-06,
+      "loss": 1.4479,
+      "step": 11064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.216258068997849e-06,
+      "loss": 1.1061,
+      "step": 11065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.206650665256741e-06,
+      "loss": 1.1052,
+      "step": 11066
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.1970504537327225e-06,
+      "loss": 1.5453,
+      "step": 11067
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.187457435161914e-06,
+      "loss": 0.9651,
+      "step": 11068
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.17787161027995e-06,
+      "loss": 1.2701,
+      "step": 11069
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.168292979821888e-06,
+      "loss": 1.3399,
+      "step": 11070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.158721544522228e-06,
+      "loss": 1.2271,
+      "step": 11071
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.149157305114928e-06,
+      "loss": 0.9367,
+      "step": 11072
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.1396002623333786e-06,
+      "loss": 0.9883,
+      "step": 11073
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.130050416910427e-06,
+      "loss": 1.2846,
+      "step": 11074
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.120507769578377e-06,
+      "loss": 1.5685,
+      "step": 11075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.110972321068975e-06,
+      "loss": 1.1751,
+      "step": 11076
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.1014440721134155e-06,
+      "loss": 1.2691,
+      "step": 11077
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.091923023442336e-06,
+      "loss": 1.5621,
+      "step": 11078
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.082409175785819e-06,
+      "loss": 1.582,
+      "step": 11079
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.072902529873403e-06,
+      "loss": 1.3195,
+      "step": 11080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.063403086434072e-06,
+      "loss": 1.3381,
+      "step": 11081
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.053910846196265e-06,
+      "loss": 0.8213,
+      "step": 11082
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.044425809887855e-06,
+      "loss": 1.3972,
+      "step": 11083
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.034947978236161e-06,
+      "loss": 1.5999,
+      "step": 11084
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.025477351967979e-06,
+      "loss": 1.4391,
+      "step": 11085
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.0160139318095165e-06,
+      "loss": 1.4679,
+      "step": 11086
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.006557718486439e-06,
+      "loss": 0.9969,
+      "step": 11087
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.997108712723898e-06,
+      "loss": 0.9836,
+      "step": 11088
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.987666915246403e-06,
+      "loss": 1.4407,
+      "step": 11089
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.978232326777999e-06,
+      "loss": 1.2619,
+      "step": 11090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.968804948042128e-06,
+      "loss": 1.3399,
+      "step": 11091
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.959384779761712e-06,
+      "loss": 1.0211,
+      "step": 11092
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.949971822659117e-06,
+      "loss": 1.2192,
+      "step": 11093
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.9405660774561e-06,
+      "loss": 1.3307,
+      "step": 11094
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.931167544873939e-06,
+      "loss": 1.252,
+      "step": 11095
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.921776225633313e-06,
+      "loss": 1.3146,
+      "step": 11096
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.9123921204543685e-06,
+      "loss": 0.797,
+      "step": 11097
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.903015230056697e-06,
+      "loss": 1.7046,
+      "step": 11098
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.893645555159344e-06,
+      "loss": 1.0374,
+      "step": 11099
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.884283096480769e-06,
+      "loss": 1.4967,
+      "step": 11100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.874927854738899e-06,
+      "loss": 1.3534,
+      "step": 11101
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.8655798306511245e-06,
+      "loss": 1.0,
+      "step": 11102
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.856239024934251e-06,
+      "loss": 1.1237,
+      "step": 11103
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.846905438304573e-06,
+      "loss": 1.2245,
+      "step": 11104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.837579071477761e-06,
+      "loss": 1.1931,
+      "step": 11105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.828259925168999e-06,
+      "loss": 1.2308,
+      "step": 11106
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.818948000092905e-06,
+      "loss": 1.4402,
+      "step": 11107
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.809643296963507e-06,
+      "loss": 0.8666,
+      "step": 11108
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.800345816494324e-06,
+      "loss": 1.2515,
+      "step": 11109
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.791055559398284e-06,
+      "loss": 1.0355,
+      "step": 11110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.781772526387785e-06,
+      "loss": 1.1222,
+      "step": 11111
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.772496718174681e-06,
+      "loss": 0.788,
+      "step": 11112
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.763228135470222e-06,
+      "loss": 1.243,
+      "step": 11113
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.753966778985164e-06,
+      "loss": 1.9402,
+      "step": 11114
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7447126494296954e-06,
+      "loss": 1.301,
+      "step": 11115
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.735465747513391e-06,
+      "loss": 1.1401,
+      "step": 11116
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.72622607394534e-06,
+      "loss": 0.7649,
+      "step": 11117
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.716993629434053e-06,
+      "loss": 1.281,
+      "step": 11118
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7077684146874975e-06,
+      "loss": 1.3426,
+      "step": 11119
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.698550430413074e-06,
+      "loss": 1.1367,
+      "step": 11120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.689339677317618e-06,
+      "loss": 1.0725,
+      "step": 11121
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.6801361561074405e-06,
+      "loss": 1.2653,
+      "step": 11122
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.670939867488267e-06,
+      "loss": 0.9004,
+      "step": 11123
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.661750812165301e-06,
+      "loss": 1.3164,
+      "step": 11124
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.652568990843166e-06,
+      "loss": 1.2985,
+      "step": 11125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.643394404225932e-06,
+      "loss": 1.0223,
+      "step": 11126
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.634227053017116e-06,
+      "loss": 0.905,
+      "step": 11127
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.625066937919688e-06,
+      "loss": 1.2371,
+      "step": 11128
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.615914059636074e-06,
+      "loss": 1.2908,
+      "step": 11129
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.606768418868114e-06,
+      "loss": 1.4481,
+      "step": 11130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.597630016317135e-06,
+      "loss": 1.3725,
+      "step": 11131
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.588498852683843e-06,
+      "loss": 1.6561,
+      "step": 11132
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.579374928668457e-06,
+      "loss": 2.5309,
+      "step": 11133
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.570258244970605e-06,
+      "loss": 0.903,
+      "step": 11134
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.561148802289351e-06,
+      "loss": 1.3763,
+      "step": 11135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.55204660132328e-06,
+      "loss": 1.1332,
+      "step": 11136
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5429516427703e-06,
+      "loss": 1.1022,
+      "step": 11137
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.533863927327854e-06,
+      "loss": 1.0572,
+      "step": 11138
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.524783455692795e-06,
+      "loss": 1.3488,
+      "step": 11139
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.515710228561433e-06,
+      "loss": 0.777,
+      "step": 11140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.506644246629533e-06,
+      "loss": 1.3818,
+      "step": 11141
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.49758551059224e-06,
+      "loss": 1.0308,
+      "step": 11142
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.488534021144232e-06,
+      "loss": 1.1785,
+      "step": 11143
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.479489778979574e-06,
+      "loss": 1.2349,
+      "step": 11144
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4704527847918015e-06,
+      "loss": 1.1302,
+      "step": 11145
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.461423039273883e-06,
+      "loss": 1.7692,
+      "step": 11146
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.45240054311823e-06,
+      "loss": 1.1666,
+      "step": 11147
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.443385297016701e-06,
+      "loss": 0.9333,
+      "step": 11148
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.434377301660598e-06,
+      "loss": 1.0521,
+      "step": 11149
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4253765577406805e-06,
+      "loss": 1.2301,
+      "step": 11150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.416383065947117e-06,
+      "loss": 1.0389,
+      "step": 11151
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.407396826969579e-06,
+      "loss": 1.6226,
+      "step": 11152
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3984178414971055e-06,
+      "loss": 1.1424,
+      "step": 11153
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.389446110218232e-06,
+      "loss": 1.2659,
+      "step": 11154
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.380481633820922e-06,
+      "loss": 1.4694,
+      "step": 11155
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.37152441299259e-06,
+      "loss": 1.1523,
+      "step": 11156
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.36257444842011e-06,
+      "loss": 0.7798,
+      "step": 11157
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.353631740789722e-06,
+      "loss": 1.8943,
+      "step": 11158
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.344696290787221e-06,
+      "loss": 1.1696,
+      "step": 11159
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.335768099097771e-06,
+      "loss": 1.1987,
+      "step": 11160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.326847166406001e-06,
+      "loss": 1.1896,
+      "step": 11161
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.317933493395999e-06,
+      "loss": 1.2409,
+      "step": 11162
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.30902708075125e-06,
+      "loss": 1.4332,
+      "step": 11163
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.300127929154708e-06,
+      "loss": 1.527,
+      "step": 11164
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.291236039288805e-06,
+      "loss": 1.59,
+      "step": 11165
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.282351411835351e-06,
+      "loss": 1.1846,
+      "step": 11166
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.273474047475668e-06,
+      "loss": 0.9718,
+      "step": 11167
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.264603946890467e-06,
+      "loss": 0.9701,
+      "step": 11168
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.255741110759915e-06,
+      "loss": 0.4637,
+      "step": 11169
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.246885539763624e-06,
+      "loss": 1.7034,
+      "step": 11170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.238037234580673e-06,
+      "loss": 1.4126,
+      "step": 11171
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.229196195889541e-06,
+      "loss": 1.7351,
+      "step": 11172
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.220362424368208e-06,
+      "loss": 0.9416,
+      "step": 11173
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.211535920694021e-06,
+      "loss": 1.0047,
+      "step": 11174
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2027166855438295e-06,
+      "loss": 1.4116,
+      "step": 11175
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.193904719593912e-06,
+      "loss": 0.8086,
+      "step": 11176
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.185100023519962e-06,
+      "loss": 1.5591,
+      "step": 11177
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.176302597997163e-06,
+      "loss": 1.5603,
+      "step": 11178
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.167512443700095e-06,
+      "loss": 0.9522,
+      "step": 11179
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.1587295613028e-06,
+      "loss": 0.8309,
+      "step": 11180
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.149953951478781e-06,
+      "loss": 1.1663,
+      "step": 11181
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.1411856149009454e-06,
+      "loss": 0.771,
+      "step": 11182
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.132424552241666e-06,
+      "loss": 1.9177,
+      "step": 11183
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.123670764172772e-06,
+      "loss": 1.3494,
+      "step": 11184
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.114924251365494e-06,
+      "loss": 1.5295,
+      "step": 11185
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.106185014490538e-06,
+      "loss": 1.024,
+      "step": 11186
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.097453054218048e-06,
+      "loss": 1.085,
+      "step": 11187
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.088728371217599e-06,
+      "loss": 2.5687,
+      "step": 11188
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.080010966158233e-06,
+      "loss": 1.5302,
+      "step": 11189
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.0713008397083705e-06,
+      "loss": 1.4821,
+      "step": 11190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.062597992535945e-06,
+      "loss": 0.8477,
+      "step": 11191
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.053902425308299e-06,
+      "loss": 1.0112,
+      "step": 11192
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.045214138692222e-06,
+      "loss": 1.1881,
+      "step": 11193
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.036533133353949e-06,
+      "loss": 1.7194,
+      "step": 11194
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.027859409959135e-06,
+      "loss": 1.3862,
+      "step": 11195
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.0191929691729145e-06,
+      "loss": 1.5346,
+      "step": 11196
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.010533811659845e-06,
+      "loss": 1.1698,
+      "step": 11197
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.001881938083908e-06,
+      "loss": 1.2563,
+      "step": 11198
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.993237349108571e-06,
+      "loss": 0.9141,
+      "step": 11199
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.984600045396681e-06,
+      "loss": 0.7328,
+      "step": 11200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.975970027610566e-06,
+      "loss": 0.8903,
+      "step": 11201
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.967347296412006e-06,
+      "loss": 1.6775,
+      "step": 11202
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.958731852462184e-06,
+      "loss": 2.1688,
+      "step": 11203
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.95012369642176e-06,
+      "loss": 1.1854,
+      "step": 11204
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.941522828950829e-06,
+      "loss": 1.4266,
+      "step": 11205
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.932929250708895e-06,
+      "loss": 0.9944,
+      "step": 11206
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.924342962354933e-06,
+      "loss": 1.2739,
+      "step": 11207
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.915763964547371e-06,
+      "loss": 1.1142,
+      "step": 11208
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.907192257944026e-06,
+      "loss": 1.1385,
+      "step": 11209
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.89862784320223e-06,
+      "loss": 1.73,
+      "step": 11210
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.890070720978679e-06,
+      "loss": 0.8202,
+      "step": 11211
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.88152089192957e-06,
+      "loss": 1.0094,
+      "step": 11212
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.872978356710511e-06,
+      "loss": 1.0462,
+      "step": 11213
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.864443115976547e-06,
+      "loss": 1.1024,
+      "step": 11214
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8559151703822085e-06,
+      "loss": 1.3483,
+      "step": 11215
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.847394520581372e-06,
+      "loss": 1.547,
+      "step": 11216
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.83888116722746e-06,
+      "loss": 0.9249,
+      "step": 11217
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.830375110973273e-06,
+      "loss": 0.7753,
+      "step": 11218
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8218763524710665e-06,
+      "loss": 1.7523,
+      "step": 11219
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.813384892372541e-06,
+      "loss": 1.1471,
+      "step": 11220
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.804900731328843e-06,
+      "loss": 1.9139,
+      "step": 11221
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.79642386999054e-06,
+      "loss": 1.1833,
+      "step": 11222
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.787954309007659e-06,
+      "loss": 1.3781,
+      "step": 11223
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.779492049029655e-06,
+      "loss": 1.3364,
+      "step": 11224
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.771037090705421e-06,
+      "loss": 1.2871,
+      "step": 11225
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.762589434683318e-06,
+      "loss": 1.6628,
+      "step": 11226
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.754149081611092e-06,
+      "loss": 1.2378,
+      "step": 11227
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7457160321359825e-06,
+      "loss": 1.5657,
+      "step": 11228
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.73729028690465e-06,
+      "loss": 1.0177,
+      "step": 11229
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.728871846563176e-06,
+      "loss": 1.0515,
+      "step": 11230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.720460711757135e-06,
+      "loss": 1.0092,
+      "step": 11231
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.712056883131466e-06,
+      "loss": 1.5173,
+      "step": 11232
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.703660361330586e-06,
+      "loss": 1.3256,
+      "step": 11233
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6952711469983926e-06,
+      "loss": 0.7594,
+      "step": 11234
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.686889240778158e-06,
+      "loss": 1.602,
+      "step": 11235
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.678514643312637e-06,
+      "loss": 1.2512,
+      "step": 11236
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.670147355243981e-06,
+      "loss": 0.8241,
+      "step": 11237
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.661787377213822e-06,
+      "loss": 1.2051,
+      "step": 11238
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.653434709863203e-06,
+      "loss": 0.6288,
+      "step": 11239
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.645089353832632e-06,
+      "loss": 1.0908,
+      "step": 11240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.636751309762044e-06,
+      "loss": 1.1409,
+      "step": 11241
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6284205782908266e-06,
+      "loss": 1.4373,
+      "step": 11242
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6200971600577455e-06,
+      "loss": 1.3395,
+      "step": 11243
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.611781055701104e-06,
+      "loss": 1.2243,
+      "step": 11244
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.603472265858555e-06,
+      "loss": 0.9177,
+      "step": 11245
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.595170791167258e-06,
+      "loss": 1.2738,
+      "step": 11246
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.58687663226377e-06,
+      "loss": 2.6046,
+      "step": 11247
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.578589789784093e-06,
+      "loss": 1.5361,
+      "step": 11248
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.570310264363697e-06,
+      "loss": 0.8649,
+      "step": 11249
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.562038056637441e-06,
+      "loss": 1.2583,
+      "step": 11250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.553773167239672e-06,
+      "loss": 1.1225,
+      "step": 11251
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.545515596804162e-06,
+      "loss": 1.3823,
+      "step": 11252
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.537265345964082e-06,
+      "loss": 1.4875,
+      "step": 11253
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.529022415352091e-06,
+      "loss": 1.4021,
+      "step": 11254
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.520786805600275e-06,
+      "loss": 1.4141,
+      "step": 11255
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.512558517340137e-06,
+      "loss": 1.3654,
+      "step": 11256
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.504337551202653e-06,
+      "loss": 1.6761,
+      "step": 11257
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.496123907818206e-06,
+      "loss": 1.7194,
+      "step": 11258
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4879175878166365e-06,
+      "loss": 0.9279,
+      "step": 11259
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.47971859182722e-06,
+      "loss": 1.4658,
+      "step": 11260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.471526920478664e-06,
+      "loss": 0.9302,
+      "step": 11261
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.46334257439911e-06,
+      "loss": 1.36,
+      "step": 11262
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.455165554216178e-06,
+      "loss": 1.208,
+      "step": 11263
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.446995860556857e-06,
+      "loss": 1.3747,
+      "step": 11264
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.438833494047623e-06,
+      "loss": 1.1917,
+      "step": 11265
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.430678455314397e-06,
+      "loss": 1.2225,
+      "step": 11266
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.422530744982489e-06,
+      "loss": 1.3083,
+      "step": 11267
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.414390363676713e-06,
+      "loss": 1.3914,
+      "step": 11268
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.406257312021245e-06,
+      "loss": 1.1638,
+      "step": 11269
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.398131590639765e-06,
+      "loss": 1.4763,
+      "step": 11270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.390013200155363e-06,
+      "loss": 1.6505,
+      "step": 11271
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.381902141190574e-06,
+      "loss": 0.9398,
+      "step": 11272
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.373798414367369e-06,
+      "loss": 0.9376,
+      "step": 11273
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.365702020307139e-06,
+      "loss": 1.0966,
+      "step": 11274
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.35761295963073e-06,
+      "loss": 1.5682,
+      "step": 11275
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.349531232958426e-06,
+      "loss": 0.6507,
+      "step": 11276
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.341456840909952e-06,
+      "loss": 1.015,
+      "step": 11277
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.333389784104469e-06,
+      "loss": 0.8837,
+      "step": 11278
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3253300631605706e-06,
+      "loss": 0.8872,
+      "step": 11279
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3172776786962635e-06,
+      "loss": 1.4718,
+      "step": 11280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.309232631329041e-06,
+      "loss": 0.9838,
+      "step": 11281
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.301194921675799e-06,
+      "loss": 1.3739,
+      "step": 11282
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.29316455035289e-06,
+      "loss": 1.4875,
+      "step": 11283
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.285141517976088e-06,
+      "loss": 0.8028,
+      "step": 11284
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.277125825160621e-06,
+      "loss": 0.9517,
+      "step": 11285
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.269117472521134e-06,
+      "loss": 0.6234,
+      "step": 11286
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2611164606717214e-06,
+      "loss": 1.2093,
+      "step": 11287
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.253122790225916e-06,
+      "loss": 1.4951,
+      "step": 11288
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.245136461796695e-06,
+      "loss": 0.6889,
+      "step": 11289
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.237157475996434e-06,
+      "loss": 1.3709,
+      "step": 11290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.229185833436999e-06,
+      "loss": 0.9773,
+      "step": 11291
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.221221534729658e-06,
+      "loss": 1.2066,
+      "step": 11292
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.213264580485132e-06,
+      "loss": 1.0642,
+      "step": 11293
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.205314971313557e-06,
+      "loss": 1.1644,
+      "step": 11294
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1973727078245315e-06,
+      "loss": 1.6173,
+      "step": 11295
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.189437790627093e-06,
+      "loss": 1.0148,
+      "step": 11296
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.181510220329676e-06,
+      "loss": 0.9306,
+      "step": 11297
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.173589997540206e-06,
+      "loss": 0.7058,
+      "step": 11298
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.165677122865996e-06,
+      "loss": 0.8685,
+      "step": 11299
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.157771596913839e-06,
+      "loss": 1.6432,
+      "step": 11300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.149873420289918e-06,
+      "loss": 1.1265,
+      "step": 11301
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.141982593599891e-06,
+      "loss": 1.183,
+      "step": 11302
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.134099117448842e-06,
+      "loss": 0.8976,
+      "step": 11303
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.126222992441275e-06,
+      "loss": 0.8026,
+      "step": 11304
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.1183542191811646e-06,
+      "loss": 1.3447,
+      "step": 11305
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.110492798271881e-06,
+      "loss": 1.0176,
+      "step": 11306
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.102638730316255e-06,
+      "loss": 1.1094,
+      "step": 11307
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.094792015916527e-06,
+      "loss": 0.7811,
+      "step": 11308
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.086952655674436e-06,
+      "loss": 1.2527,
+      "step": 11309
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.079120650191104e-06,
+      "loss": 1.0743,
+      "step": 11310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.071296000067093e-06,
+      "loss": 1.0027,
+      "step": 11311
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.063478705902412e-06,
+      "loss": 0.8405,
+      "step": 11312
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.055668768296495e-06,
+      "loss": 1.6501,
+      "step": 11313
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.047866187848226e-06,
+      "loss": 0.9875,
+      "step": 11314
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.040070965155918e-06,
+      "loss": 1.0404,
+      "step": 11315
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0322831008173356e-06,
+      "loss": 0.8898,
+      "step": 11316
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.024502595429647e-06,
+      "loss": 1.2039,
+      "step": 11317
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0167294495894625e-06,
+      "loss": 2.178,
+      "step": 11318
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.008963663892862e-06,
+      "loss": 1.4632,
+      "step": 11319
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.001205238935335e-06,
+      "loss": 0.8772,
+      "step": 11320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.993454175311795e-06,
+      "loss": 1.5599,
+      "step": 11321
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.98571047361661e-06,
+      "loss": 1.2295,
+      "step": 11322
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.977974134443585e-06,
+      "loss": 1.1192,
+      "step": 11323
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.970245158385954e-06,
+      "loss": 0.9079,
+      "step": 11324
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.962523546036379e-06,
+      "loss": 1.9502,
+      "step": 11325
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.954809297986994e-06,
+      "loss": 1.1838,
+      "step": 11326
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.947102414829296e-06,
+      "loss": 1.4252,
+      "step": 11327
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.939402897154277e-06,
+      "loss": 0.9396,
+      "step": 11328
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.931710745552353e-06,
+      "loss": 1.1057,
+      "step": 11329
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.924025960613376e-06,
+      "loss": 1.3411,
+      "step": 11330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.916348542926607e-06,
+      "loss": 0.8138,
+      "step": 11331
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.908678493080775e-06,
+      "loss": 0.9895,
+      "step": 11332
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.901015811664044e-06,
+      "loss": 1.2291,
+      "step": 11333
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.893360499263976e-06,
+      "loss": 1.178,
+      "step": 11334
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.885712556467602e-06,
+      "loss": 1.2253,
+      "step": 11335
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.878071983861376e-06,
+      "loss": 0.9742,
+      "step": 11336
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.870438782031205e-06,
+      "loss": 1.3783,
+      "step": 11337
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.862812951562389e-06,
+      "loss": 1.2544,
+      "step": 11338
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.855194493039704e-06,
+      "loss": 0.9108,
+      "step": 11339
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.84758340704734e-06,
+      "loss": 1.3208,
+      "step": 11340
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.839979694168927e-06,
+      "loss": 2.1368,
+      "step": 11341
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.832383354987534e-06,
+      "loss": 0.9928,
+      "step": 11342
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.824794390085651e-06,
+      "loss": 1.2774,
+      "step": 11343
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8172128000452116e-06,
+      "loss": 0.5285,
+      "step": 11344
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8096385854475837e-06,
+      "loss": 0.9032,
+      "step": 11345
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8020717468735813e-06,
+      "loss": 0.8001,
+      "step": 11346
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7945122849034288e-06,
+      "loss": 1.2992,
+      "step": 11347
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.786960200116829e-06,
+      "loss": 1.5403,
+      "step": 11348
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7794154930928415e-06,
+      "loss": 1.1757,
+      "step": 11349
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.771878164410014e-06,
+      "loss": 1.1764,
+      "step": 11350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7643482146463407e-06,
+      "loss": 0.6942,
+      "step": 11351
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7568256443792247e-06,
+      "loss": 1.1755,
+      "step": 11352
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7493104541855174e-06,
+      "loss": 1.3317,
+      "step": 11353
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7418026446414565e-06,
+      "loss": 1.3583,
+      "step": 11354
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7343022163227826e-06,
+      "loss": 1.4227,
+      "step": 11355
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.726809169804646e-06,
+      "loss": 1.5893,
+      "step": 11356
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.719323505661598e-06,
+      "loss": 1.6494,
+      "step": 11357
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7118452244676803e-06,
+      "loss": 1.0631,
+      "step": 11358
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7043743267963117e-06,
+      "loss": 1.4074,
+      "step": 11359
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.696910813220389e-06,
+      "loss": 1.3679,
+      "step": 11360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6894546843122324e-06,
+      "loss": 0.6384,
+      "step": 11361
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6820059406435625e-06,
+      "loss": 1.5039,
+      "step": 11362
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6745645827856e-06,
+      "loss": 1.3523,
+      "step": 11363
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.667130611308922e-06,
+      "loss": 0.9098,
+      "step": 11364
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6597040267835947e-06,
+      "loss": 1.7959,
+      "step": 11365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6522848297790955e-06,
+      "loss": 1.7223,
+      "step": 11366
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6448730208643366e-06,
+      "loss": 1.1702,
+      "step": 11367
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.637468600607674e-06,
+      "loss": 1.3769,
+      "step": 11368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6300715695768763e-06,
+      "loss": 1.148,
+      "step": 11369
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6226819283391776e-06,
+      "loss": 0.9631,
+      "step": 11370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.615299677461226e-06,
+      "loss": 1.3157,
+      "step": 11371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6079248175090896e-06,
+      "loss": 0.8992,
+      "step": 11372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6005573490482835e-06,
+      "loss": 1.3014,
+      "step": 11373
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.593197272643789e-06,
+      "loss": 1.1169,
+      "step": 11374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5858445888599434e-06,
+      "loss": 1.0651,
+      "step": 11375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5784992982605845e-06,
+      "loss": 0.8833,
+      "step": 11376
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.571161401408951e-06,
+      "loss": 1.1284,
+      "step": 11377
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.563830898867737e-06,
+      "loss": 1.1187,
+      "step": 11378
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5565077911990597e-06,
+      "loss": 0.9564,
+      "step": 11379
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.549192078964447e-06,
+      "loss": 1.4834,
+      "step": 11380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5418837627248845e-06,
+      "loss": 1.1064,
+      "step": 11381
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5345828430407903e-06,
+      "loss": 0.6746,
+      "step": 11382
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5272893204719938e-06,
+      "loss": 1.5434,
+      "step": 11383
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5200031955777925e-06,
+      "loss": 0.8958,
+      "step": 11384
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.512724468916917e-06,
+      "loss": 1.4714,
+      "step": 11385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5054531410474765e-06,
+      "loss": 1.5205,
+      "step": 11386
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4981892125270477e-06,
+      "loss": 1.4227,
+      "step": 11387
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.490932683912651e-06,
+      "loss": 0.9159,
+      "step": 11388
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.483683555760742e-06,
+      "loss": 0.9715,
+      "step": 11389
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476441828627175e-06,
+      "loss": 1.7419,
+      "step": 11390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.469207503067262e-06,
+      "loss": 1.1132,
+      "step": 11391
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.461980579635737e-06,
+      "loss": 1.0269,
+      "step": 11392
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4547610588867685e-06,
+      "loss": 0.9085,
+      "step": 11393
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.447548941373979e-06,
+      "loss": 1.0445,
+      "step": 11394
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4403442276503938e-06,
+      "loss": 1.5647,
+      "step": 11395
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4331469182684704e-06,
+      "loss": 0.8282,
+      "step": 11396
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4259570137801344e-06,
+      "loss": 1.1272,
+      "step": 11397
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4187745147367e-06,
+      "loss": 0.975,
+      "step": 11398
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.411599421688938e-06,
+      "loss": 0.8278,
+      "step": 11399
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4044317351870414e-06,
+      "loss": 1.4811,
+      "step": 11400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3972714557806484e-06,
+      "loss": 1.2944,
+      "step": 11401
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.390118584018809e-06,
+      "loss": 1.1823,
+      "step": 11402
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.382973120450028e-06,
+      "loss": 0.7277,
+      "step": 11403
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3758350656222126e-06,
+      "loss": 2.0824,
+      "step": 11404
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3687044200827356e-06,
+      "loss": 1.701,
+      "step": 11405
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.361581184378382e-06,
+      "loss": 0.7112,
+      "step": 11406
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3544653590553697e-06,
+      "loss": 1.4809,
+      "step": 11407
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3473569446593635e-06,
+      "loss": 1.9454,
+      "step": 11408
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3402559417354264e-06,
+      "loss": 1.569,
+      "step": 11409
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3331623508280896e-06,
+      "loss": 1.2563,
+      "step": 11410
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.326076172481296e-06,
+      "loss": 1.2658,
+      "step": 11411
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3189974072384223e-06,
+      "loss": 1.2193,
+      "step": 11412
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.311926055642278e-06,
+      "loss": 0.7517,
+      "step": 11413
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.304862118235108e-06,
+      "loss": 1.3897,
+      "step": 11414
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2978055955585895e-06,
+      "loss": 1.6621,
+      "step": 11415
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2907564881538343e-06,
+      "loss": 1.4635,
+      "step": 11416
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2837147965613545e-06,
+      "loss": 0.8473,
+      "step": 11417
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2766805213211295e-06,
+      "loss": 1.8929,
+      "step": 11418
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2696536629725603e-06,
+      "loss": 1.2935,
+      "step": 11419
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2626342220544614e-06,
+      "loss": 1.1634,
+      "step": 11420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.255622199105135e-06,
+      "loss": 1.379,
+      "step": 11421
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2486175946622512e-06,
+      "loss": 1.3837,
+      "step": 11422
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2416204092629133e-06,
+      "loss": 1.2287,
+      "step": 11423
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2346306434437036e-06,
+      "loss": 1.0817,
+      "step": 11424
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.227648297740593e-06,
+      "loss": 1.3843,
+      "step": 11425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.220673372689009e-06,
+      "loss": 1.3909,
+      "step": 11426
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.213705868823802e-06,
+      "loss": 1.7114,
+      "step": 11427
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2067457866792328e-06,
+      "loss": 1.166,
+      "step": 11428
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.199793126789019e-06,
+      "loss": 0.8155,
+      "step": 11429
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1928478896863014e-06,
+      "loss": 1.3812,
+      "step": 11430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.185910075903653e-06,
+      "loss": 1.1311,
+      "step": 11431
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1789796859730823e-06,
+      "loss": 1.5842,
+      "step": 11432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1720567204260087e-06,
+      "loss": 1.2499,
+      "step": 11433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1651411797933073e-06,
+      "loss": 0.4981,
+      "step": 11434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.158233064605276e-06,
+      "loss": 1.2905,
+      "step": 11435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1513323753916244e-06,
+      "loss": 1.2183,
+      "step": 11436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.144439112681519e-06,
+      "loss": 1.5731,
+      "step": 11437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1375532770035466e-06,
+      "loss": 1.3932,
+      "step": 11438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.130674868885708e-06,
+      "loss": 0.9806,
+      "step": 11439
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.12380388885547e-06,
+      "loss": 1.1765,
+      "step": 11440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1169403374396887e-06,
+      "loss": 1.0639,
+      "step": 11441
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.110084215164688e-06,
+      "loss": 1.3497,
+      "step": 11442
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1032355225562136e-06,
+      "loss": 1.0708,
+      "step": 11443
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0963942601394013e-06,
+      "loss": 0.9811,
+      "step": 11444
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.089560428438887e-06,
+      "loss": 0.8972,
+      "step": 11445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0827340279786733e-06,
+      "loss": 1.3326,
+      "step": 11446
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0759150592822307e-06,
+      "loss": 1.1709,
+      "step": 11447
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.069103522872463e-06,
+      "loss": 0.8045,
+      "step": 11448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0622994192716524e-06,
+      "loss": 1.2842,
+      "step": 11449
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0555027490015707e-06,
+      "loss": 1.177,
+      "step": 11450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0487135125833897e-06,
+      "loss": 1.6051,
+      "step": 11451
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0419317105377267e-06,
+      "loss": 0.9917,
+      "step": 11452
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0351573433846204e-06,
+      "loss": 1.1295,
+      "step": 11453
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0283904116435334e-06,
+      "loss": 0.8088,
+      "step": 11454
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0216309158333623e-06,
+      "loss": 1.3785,
+      "step": 11455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.014878856472436e-06,
+      "loss": 1.1817,
+      "step": 11456
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.008134234078519e-06,
+      "loss": 0.9039,
+      "step": 11457
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.001397049168786e-06,
+      "loss": 0.8532,
+      "step": 11458
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.994667302259879e-06,
+      "loss": 1.3921,
+      "step": 11459
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9879449938678304e-06,
+      "loss": 1.5639,
+      "step": 11460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9812301245081164e-06,
+      "loss": 1.0534,
+      "step": 11461
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9745226946956474e-06,
+      "loss": 0.4225,
+      "step": 11462
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.967822704944745e-06,
+      "loss": 1.1889,
+      "step": 11463
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9611301557692094e-06,
+      "loss": 1.2345,
+      "step": 11464
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.954445047682197e-06,
+      "loss": 0.8258,
+      "step": 11465
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9477673811963534e-06,
+      "loss": 0.9942,
+      "step": 11466
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9410971568237132e-06,
+      "loss": 0.8686,
+      "step": 11467
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9344343750757896e-06,
+      "loss": 0.9254,
+      "step": 11468
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.927779036463463e-06,
+      "loss": 1.5178,
+      "step": 11469
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.921131141497102e-06,
+      "loss": 1.2709,
+      "step": 11470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9144906906864555e-06,
+      "loss": 1.5285,
+      "step": 11471
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9078576845407377e-06,
+      "loss": 1.1642,
+      "step": 11472
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.901232123568576e-06,
+      "loss": 1.7292,
+      "step": 11473
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.894614008278018e-06,
+      "loss": 1.2059,
+      "step": 11474
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8880033391765815e-06,
+      "loss": 1.2478,
+      "step": 11475
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.881400116771138e-06,
+      "loss": 1.2398,
+      "step": 11476
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8748043415680605e-06,
+      "loss": 1.3438,
+      "step": 11477
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8682160140731106e-06,
+      "loss": 1.0232,
+      "step": 11478
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.861635134791507e-06,
+      "loss": 0.1285,
+      "step": 11479
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8550617042278684e-06,
+      "loss": 1.0125,
+      "step": 11480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.848495722886246e-06,
+      "loss": 0.9947,
+      "step": 11481
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.841937191270161e-06,
+      "loss": 0.7268,
+      "step": 11482
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8353861098824985e-06,
+      "loss": 1.1733,
+      "step": 11483
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.828842479225635e-06,
+      "loss": 1.2939,
+      "step": 11484
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.822306299801325e-06,
+      "loss": 0.9879,
+      "step": 11485
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.815777572110778e-06,
+      "loss": 1.3692,
+      "step": 11486
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8092562966546275e-06,
+      "loss": 1.3849,
+      "step": 11487
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.80274247393294e-06,
+      "loss": 1.324,
+      "step": 11488
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7962361044451934e-06,
+      "loss": 1.4406,
+      "step": 11489
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.789737188690311e-06,
+      "loss": 0.7021,
+      "step": 11490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.783245727166661e-06,
+      "loss": 2.2728,
+      "step": 11491
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7767617203719786e-06,
+      "loss": 1.0173,
+      "step": 11492
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.770285168803488e-06,
+      "loss": 1.3095,
+      "step": 11493
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7638160729578257e-06,
+      "loss": 1.264,
+      "step": 11494
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7573544333310385e-06,
+      "loss": 1.3032,
+      "step": 11495
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7509002504186423e-06,
+      "loss": 1.5676,
+      "step": 11496
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.744453524715529e-06,
+      "loss": 1.2641,
+      "step": 11497
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.738014256716037e-06,
+      "loss": 2.6504,
+      "step": 11498
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7315824469139607e-06,
+      "loss": 0.972,
+      "step": 11499
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7251580958024937e-06,
+      "loss": 1.766,
+      "step": 11500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.718741203874264e-06,
+      "loss": 0.8462,
+      "step": 11501
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7123317716213237e-06,
+      "loss": 0.8575,
+      "step": 11502
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7059297995351563e-06,
+      "loss": 1.244,
+      "step": 11503
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6995352881066803e-06,
+      "loss": 1.3274,
+      "step": 11504
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6931482378262485e-06,
+      "loss": 2.1502,
+      "step": 11505
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.686768649183602e-06,
+      "loss": 1.2508,
+      "step": 11506
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6803965226679605e-06,
+      "loss": 1.3244,
+      "step": 11507
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6740318587679445e-06,
+      "loss": 1.5243,
+      "step": 11508
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6676746579715974e-06,
+      "loss": 0.9462,
+      "step": 11509
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.661324920766406e-06,
+      "loss": 0.7888,
+      "step": 11510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6549826476392814e-06,
+      "loss": 1.2633,
+      "step": 11511
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.648647839076557e-06,
+      "loss": 1.1344,
+      "step": 11512
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.642320495563977e-06,
+      "loss": 1.3778,
+      "step": 11513
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6360006175867535e-06,
+      "loss": 1.0987,
+      "step": 11514
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.629688205629499e-06,
+      "loss": 0.9014,
+      "step": 11515
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.62338326017626e-06,
+      "loss": 1.3655,
+      "step": 11516
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6170857817105042e-06,
+      "loss": 0.8283,
+      "step": 11517
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6107957707151353e-06,
+      "loss": 1.2431,
+      "step": 11518
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.604513227672467e-06,
+      "loss": 1.614,
+      "step": 11519
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5982381530642807e-06,
+      "loss": 0.8452,
+      "step": 11520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.591970547371747e-06,
+      "loss": 1.7083,
+      "step": 11521
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5857104110754705e-06,
+      "loss": 1.0162,
+      "step": 11522
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5794577446555002e-06,
+      "loss": 1.0261,
+      "step": 11523
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.573212548591275e-06,
+      "loss": 1.1833,
+      "step": 11524
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.566974823361712e-06,
+      "loss": 1.2395,
+      "step": 11525
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.560744569445117e-06,
+      "loss": 0.9024,
+      "step": 11526
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.554521787319242e-06,
+      "loss": 1.2867,
+      "step": 11527
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5483064774612597e-06,
+      "loss": 0.8768,
+      "step": 11528
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5420986403477565e-06,
+      "loss": 1.3391,
+      "step": 11529
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5358982764547735e-06,
+      "loss": 1.1847,
+      "step": 11530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5297053862577525e-06,
+      "loss": 1.3996,
+      "step": 11531
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5235199702315804e-06,
+      "loss": 1.2881,
+      "step": 11532
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5173420288505668e-06,
+      "loss": 1.2139,
+      "step": 11533
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5111715625884323e-06,
+      "loss": 0.9587,
+      "step": 11534
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.505008571918366e-06,
+      "loss": 1.5079,
+      "step": 11535
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.498853057312922e-06,
+      "loss": 1.1368,
+      "step": 11536
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4927050192441348e-06,
+      "loss": 1.0157,
+      "step": 11537
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.486564458183449e-06,
+      "loss": 1.3778,
+      "step": 11538
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4804313746017103e-06,
+      "loss": 1.2313,
+      "step": 11539
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4743057689692318e-06,
+      "loss": 1.5439,
+      "step": 11540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.468187641755726e-06,
+      "loss": 1.273,
+      "step": 11541
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.462076993430351e-06,
+      "loss": 1.0841,
+      "step": 11542
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4559738244616658e-06,
+      "loss": 0.7929,
+      "step": 11543
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.449878135317674e-06,
+      "loss": 1.407,
+      "step": 11544
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4437899264658117e-06,
+      "loss": 1.1387,
+      "step": 11545
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4377091983729174e-06,
+      "loss": 1.0683,
+      "step": 11546
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4316359515052846e-06,
+      "loss": 1.1553,
+      "step": 11547
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4255701863286185e-06,
+      "loss": 1.4634,
+      "step": 11548
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.419511903308047e-06,
+      "loss": 1.0964,
+      "step": 11549
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4134611029081324e-06,
+      "loss": 1.1637,
+      "step": 11550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.407417785592836e-06,
+      "loss": 1.3409,
+      "step": 11551
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4013819518255986e-06,
+      "loss": 1.023,
+      "step": 11552
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.39535360206925e-06,
+      "loss": 1.273,
+      "step": 11553
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.389332736786054e-06,
+      "loss": 1.5662,
+      "step": 11554
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3833193564376852e-06,
+      "loss": 1.0779,
+      "step": 11555
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3773134614852534e-06,
+      "loss": 0.8617,
+      "step": 11556
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3713150523893336e-06,
+      "loss": 1.1721,
+      "step": 11557
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3653241296098694e-06,
+      "loss": 1.4516,
+      "step": 11558
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.359340693606271e-06,
+      "loss": 0.9367,
+      "step": 11559
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3533647448373274e-06,
+      "loss": 1.464,
+      "step": 11560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.347396283761316e-06,
+      "loss": 1.182,
+      "step": 11561
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3414353108358823e-06,
+      "loss": 1.0321,
+      "step": 11562
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.335481826518138e-06,
+      "loss": 1.3677,
+      "step": 11563
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3295358312645953e-06,
+      "loss": 1.0316,
+      "step": 11564
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.323597325531224e-06,
+      "loss": 1.3738,
+      "step": 11565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.317666309773359e-06,
+      "loss": 0.8382,
+      "step": 11566
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3117427844458362e-06,
+      "loss": 0.2961,
+      "step": 11567
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3058267500028594e-06,
+      "loss": 0.9114,
+      "step": 11568
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2999182068980882e-06,
+      "loss": 1.5516,
+      "step": 11569
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2940171555845934e-06,
+      "loss": 1.2447,
+      "step": 11570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2881235965148794e-06,
+      "loss": 1.0464,
+      "step": 11571
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2822375301408848e-06,
+      "loss": 1.1048,
+      "step": 11572
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2763589569139377e-06,
+      "loss": 0.9272,
+      "step": 11573
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2704878772848327e-06,
+      "loss": 1.3606,
+      "step": 11574
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2646242917037764e-06,
+      "loss": 1.2818,
+      "step": 11575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.258768200620387e-06,
+      "loss": 1.3065,
+      "step": 11576
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2529196044837274e-06,
+      "loss": 1.1152,
+      "step": 11577
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2470785037422614e-06,
+      "loss": 0.9713,
+      "step": 11578
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2412448988439085e-06,
+      "loss": 1.2878,
+      "step": 11579
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.235418790235999e-06,
+      "loss": 1.178,
+      "step": 11580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2296001783652764e-06,
+      "loss": 0.9798,
+      "step": 11581
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.223789063677928e-06,
+      "loss": 0.7428,
+      "step": 11582
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2179854466195526e-06,
+      "loss": 1.4308,
+      "step": 11583
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2121893276351948e-06,
+      "loss": 0.9867,
+      "step": 11584
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.206400707169298e-06,
+      "loss": 1.3484,
+      "step": 11585
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.200619585665753e-06,
+      "loss": 1.0027,
+      "step": 11586
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1948459635678375e-06,
+      "loss": 1.3637,
+      "step": 11587
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.18907984131832e-06,
+      "loss": 1.6404,
+      "step": 11588
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1833212193593244e-06,
+      "loss": 1.384,
+      "step": 11589
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1775700981324534e-06,
+      "loss": 1.8304,
+      "step": 11590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.171826478078698e-06,
+      "loss": 1.2311,
+      "step": 11591
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1660903596384953e-06,
+      "loss": 0.9378,
+      "step": 11592
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.160361743251682e-06,
+      "loss": 1.4929,
+      "step": 11593
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1546406293575516e-06,
+      "loss": 1.1243,
+      "step": 11594
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1489270183948197e-06,
+      "loss": 0.9465,
+      "step": 11595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1432209108016022e-06,
+      "loss": 0.8915,
+      "step": 11596
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.137522307015438e-06,
+      "loss": 1.4458,
+      "step": 11597
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1318312074733337e-06,
+      "loss": 1.3715,
+      "step": 11598
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1261476126116617e-06,
+      "loss": 0.2616,
+      "step": 11599
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1204715228662743e-06,
+      "loss": 1.091,
+      "step": 11600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.114802938672411e-06,
+      "loss": 1.6959,
+      "step": 11601
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.109141860464747e-06,
+      "loss": 0.6295,
+      "step": 11602
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.103488288677391e-06,
+      "loss": 1.5621,
+      "step": 11603
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0978422237438512e-06,
+      "loss": 1.4241,
+      "step": 11604
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.092203666097081e-06,
+      "loss": 1.1709,
+      "step": 11605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0865726161694577e-06,
+      "loss": 0.9897,
+      "step": 11606
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0809490743927906e-06,
+      "loss": 0.9256,
+      "step": 11607
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0753330411982796e-06,
+      "loss": 1.2134,
+      "step": 11608
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0697245170165804e-06,
+      "loss": 1.1497,
+      "step": 11609
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0641235022777707e-06,
+      "loss": 0.7967,
+      "step": 11610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0585299974113405e-06,
+      "loss": 0.4367,
+      "step": 11611
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.052944002846213e-06,
+      "loss": 1.1913,
+      "step": 11612
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0473655190107023e-06,
+      "loss": 1.3184,
+      "step": 11613
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0417945463326095e-06,
+      "loss": 1.0791,
+      "step": 11614
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0362310852391154e-06,
+      "loss": 0.797,
+      "step": 11615
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0306751361568234e-06,
+      "loss": 0.7977,
+      "step": 11616
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0251266995117814e-06,
+      "loss": 0.8846,
+      "step": 11617
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.019585775729449e-06,
+      "loss": 1.3024,
+      "step": 11618
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.01405236523472e-06,
+      "loss": 1.2122,
+      "step": 11619
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.008526468451888e-06,
+      "loss": 1.5669,
+      "step": 11620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.003008085804714e-06,
+      "loss": 1.1114,
+      "step": 11621
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9974972177163263e-06,
+      "loss": 0.7499,
+      "step": 11622
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9919938646093427e-06,
+      "loss": 1.6223,
+      "step": 11623
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9864980269057253e-06,
+      "loss": 1.4294,
+      "step": 11624
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.981009705026937e-06,
+      "loss": 1.5196,
+      "step": 11625
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.975528899393808e-06,
+      "loss": 1.5307,
+      "step": 11626
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9700556104266244e-06,
+      "loss": 1.474,
+      "step": 11627
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9645898385451055e-06,
+      "loss": 1.2577,
+      "step": 11628
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9591315841683387e-06,
+      "loss": 1.1277,
+      "step": 11629
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9536808477149006e-06,
+      "loss": 0.8952,
+      "step": 11630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9482376296027337e-06,
+      "loss": 0.9691,
+      "step": 11631
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9428019302492605e-06,
+      "loss": 1.1763,
+      "step": 11632
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.937373750071303e-06,
+      "loss": 0.9928,
+      "step": 11633
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9319530894850613e-06,
+      "loss": 1.7193,
+      "step": 11634
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9265399489062365e-06,
+      "loss": 1.2329,
+      "step": 11635
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.921134328749907e-06,
+      "loss": 1.0864,
+      "step": 11636
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.915736229430587e-06,
+      "loss": 1.2974,
+      "step": 11637
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9103456513621996e-06,
+      "loss": 1.0912,
+      "step": 11638
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9049625949581152e-06,
+      "loss": 1.1447,
+      "step": 11639
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8995870606311027e-06,
+      "loss": 1.2313,
+      "step": 11640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8942190487933775e-06,
+      "loss": 1.3092,
+      "step": 11641
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8888585598565544e-06,
+      "loss": 1.7913,
+      "step": 11642
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8835055942316938e-06,
+      "loss": 1.1677,
+      "step": 11643
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8781601523292669e-06,
+      "loss": 1.367,
+      "step": 11644
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8728222345591684e-06,
+      "loss": 1.2181,
+      "step": 11645
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8674918413307262e-06,
+      "loss": 0.8995,
+      "step": 11646
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8621689730526693e-06,
+      "loss": 0.9886,
+      "step": 11647
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8568536301331818e-06,
+      "loss": 1.1237,
+      "step": 11648
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.851545812979838e-06,
+      "loss": 1.2187,
+      "step": 11649
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8462455219996567e-06,
+      "loss": 1.6465,
+      "step": 11650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8409527575990571e-06,
+      "loss": 1.4134,
+      "step": 11651
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8356675201839145e-06,
+      "loss": 1.0192,
+      "step": 11652
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8303898101594942e-06,
+      "loss": 1.204,
+      "step": 11653
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8251196279305161e-06,
+      "loss": 0.4891,
+      "step": 11654
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8198569739010906e-06,
+      "loss": 1.0708,
+      "step": 11655
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8146018484747729e-06,
+      "loss": 1.1134,
+      "step": 11656
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8093542520545293e-06,
+      "loss": 1.2887,
+      "step": 11657
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8041141850427713e-06,
+      "loss": 1.2189,
+      "step": 11658
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7988816478412884e-06,
+      "loss": 1.3697,
+      "step": 11659
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.793656640851349e-06,
+      "loss": 0.7644,
+      "step": 11660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7884391644735764e-06,
+      "loss": 1.3554,
+      "step": 11661
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7832292191080957e-06,
+      "loss": 1.2762,
+      "step": 11662
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7780268051543759e-06,
+      "loss": 1.1135,
+      "step": 11663
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7728319230113755e-06,
+      "loss": 0.8322,
+      "step": 11664
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7676445730774428e-06,
+      "loss": 0.6853,
+      "step": 11665
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.762464755750326e-06,
+      "loss": 1.1964,
+      "step": 11666
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.757292471427252e-06,
+      "loss": 1.2881,
+      "step": 11667
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7521277205048148e-06,
+      "loss": 0.9457,
+      "step": 11668
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.746970503379064e-06,
+      "loss": 0.8422,
+      "step": 11669
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7418208204454833e-06,
+      "loss": 0.6079,
+      "step": 11670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7366786720989347e-06,
+      "loss": 1.271,
+      "step": 11671
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7315440587337362e-06,
+      "loss": 1.3865,
+      "step": 11672
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7264169807436059e-06,
+      "loss": 1.3611,
+      "step": 11673
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7212974385217072e-06,
+      "loss": 1.0335,
+      "step": 11674
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.716185432460604e-06,
+      "loss": 1.3329,
+      "step": 11675
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7110809629523162e-06,
+      "loss": 0.8288,
+      "step": 11676
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7059840303882303e-06,
+      "loss": 1.2531,
+      "step": 11677
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7008946351592115e-06,
+      "loss": 1.3188,
+      "step": 11678
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6958127776555033e-06,
+      "loss": 0.8411,
+      "step": 11679
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6907384582668051e-06,
+      "loss": 0.733,
+      "step": 11680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6856716773822167e-06,
+      "loss": 1.1189,
+      "step": 11681
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6806124353902719e-06,
+      "loss": 1.12,
+      "step": 11682
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6755607326789157e-06,
+      "loss": 2.0088,
+      "step": 11683
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6705165696355273e-06,
+      "loss": 1.356,
+      "step": 11684
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6654799466468861e-06,
+      "loss": 1.6201,
+      "step": 11685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6604508640992389e-06,
+      "loss": 1.002,
+      "step": 11686
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.655429322378188e-06,
+      "loss": 1.1627,
+      "step": 11687
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6504153218688035e-06,
+      "loss": 1.1398,
+      "step": 11688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6454088629555664e-06,
+      "loss": 1.6724,
+      "step": 11689
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.640409946022392e-06,
+      "loss": 1.3652,
+      "step": 11690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.635418571452596e-06,
+      "loss": 1.2133,
+      "step": 11691
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6304347396289277e-06,
+      "loss": 0.8944,
+      "step": 11692
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.625458450933548e-06,
+      "loss": 1.1608,
+      "step": 11693
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6204897057480517e-06,
+      "loss": 1.3472,
+      "step": 11694
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6155285044534452e-06,
+      "loss": 0.8524,
+      "step": 11695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6105748474301684e-06,
+      "loss": 1.1299,
+      "step": 11696
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6056287350580735e-06,
+      "loss": 1.2191,
+      "step": 11697
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6006901677164342e-06,
+      "loss": 1.3339,
+      "step": 11698
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5957591457839372e-06,
+      "loss": 0.9946,
+      "step": 11699
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5908356696387128e-06,
+      "loss": 1.2025,
+      "step": 11700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5859197396583037e-06,
+      "loss": 1.2224,
+      "step": 11701
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.581011356219675e-06,
+      "loss": 0.6547,
+      "step": 11702
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.576110519699181e-06,
+      "loss": 1.2152,
+      "step": 11703
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5712172304726547e-06,
+      "loss": 1.6516,
+      "step": 11704
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.566331488915307e-06,
+      "loss": 1.5675,
+      "step": 11705
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5614532954017823e-06,
+      "loss": 0.9904,
+      "step": 11706
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5565826503061486e-06,
+      "loss": 0.9874,
+      "step": 11707
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.551719554001918e-06,
+      "loss": 1.1013,
+      "step": 11708
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.54686400686197e-06,
+      "loss": 0.7628,
+      "step": 11709
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.542016009258651e-06,
+      "loss": 1.2856,
+      "step": 11710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5371755615636973e-06,
+      "loss": 1.4717,
+      "step": 11711
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5323426641483007e-06,
+      "loss": 1.6924,
+      "step": 11712
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5275173173830537e-06,
+      "loss": 1.1716,
+      "step": 11713
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5226995216379491e-06,
+      "loss": 1.1269,
+      "step": 11714
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.517889277282447e-06,
+      "loss": 0.8206,
+      "step": 11715
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5130865846853858e-06,
+      "loss": 0.8469,
+      "step": 11716
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5082914442150597e-06,
+      "loss": 1.3957,
+      "step": 11717
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5035038562391635e-06,
+      "loss": 1.1884,
+      "step": 11718
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4987238211248144e-06,
+      "loss": 0.6678,
+      "step": 11719
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4939513392385418e-06,
+      "loss": 1.2712,
+      "step": 11720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4891864109463305e-06,
+      "loss": 1.4556,
+      "step": 11721
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4844290366135439e-06,
+      "loss": 0.9312,
+      "step": 11722
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4796792166049899e-06,
+      "loss": 1.1802,
+      "step": 11723
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4749369512849e-06,
+      "loss": 1.5565,
+      "step": 11724
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4702022410168935e-06,
+      "loss": 1.6048,
+      "step": 11725
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4654750861640699e-06,
+      "loss": 1.2927,
+      "step": 11726
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4607554870888828e-06,
+      "loss": 1.3057,
+      "step": 11727
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4560434441532545e-06,
+      "loss": 1.582,
+      "step": 11728
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4513389577185177e-06,
+      "loss": 1.2583,
+      "step": 11729
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4466420281454064e-06,
+      "loss": 1.2551,
+      "step": 11730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4419526557940877e-06,
+      "loss": 1.5868,
+      "step": 11731
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.437270841024163e-06,
+      "loss": 1.4467,
+      "step": 11732
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4325965841946344e-06,
+      "loss": 1.3119,
+      "step": 11733
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4279298856639366e-06,
+      "loss": 1.7214,
+      "step": 11734
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4232707457898953e-06,
+      "loss": 0.9176,
+      "step": 11735
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.418619164929813e-06,
+      "loss": 1.0208,
+      "step": 11736
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4139751434403602e-06,
+      "loss": 1.3333,
+      "step": 11737
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.409338681677641e-06,
+      "loss": 1.076,
+      "step": 11738
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4047097799972154e-06,
+      "loss": 1.0673,
+      "step": 11739
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4000884387540102e-06,
+      "loss": 1.0574,
+      "step": 11740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3954746583023982e-06,
+      "loss": 0.8606,
+      "step": 11741
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3908684389961846e-06,
+      "loss": 1.3455,
+      "step": 11742
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.386269781188565e-06,
+      "loss": 1.2203,
+      "step": 11743
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3816786852321795e-06,
+      "loss": 1.1751,
+      "step": 11744
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.377095151479091e-06,
+      "loss": 1.1873,
+      "step": 11745
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3725191802807513e-06,
+      "loss": 1.4698,
+      "step": 11746
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3679507719880691e-06,
+      "loss": 1.2328,
+      "step": 11747
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3633899269513417e-06,
+      "loss": 0.9014,
+      "step": 11748
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3588366455203227e-06,
+      "loss": 1.6224,
+      "step": 11749
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.354290928044144e-06,
+      "loss": 1.019,
+      "step": 11750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3497527748713934e-06,
+      "loss": 1.1503,
+      "step": 11751
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.345222186350048e-06,
+      "loss": 1.2297,
+      "step": 11752
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.340699162827541e-06,
+      "loss": 0.7201,
+      "step": 11753
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3361837046506841e-06,
+      "loss": 0.9988,
+      "step": 11754
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3316758121657335e-06,
+      "loss": 1.3765,
+      "step": 11755
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3271754857183793e-06,
+      "loss": 1.1786,
+      "step": 11756
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.32268272565369e-06,
+      "loss": 0.8854,
+      "step": 11757
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3181975323161899e-06,
+      "loss": 0.7817,
+      "step": 11758
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.313719906049804e-06,
+      "loss": 1.704,
+      "step": 11759
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3092498471978908e-06,
+      "loss": 0.9468,
+      "step": 11760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.304787356103232e-06,
+      "loss": 1.1411,
+      "step": 11761
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3003324331079868e-06,
+      "loss": 1.2781,
+      "step": 11762
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2958850785537823e-06,
+      "loss": 1.3662,
+      "step": 11763
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2914452927816566e-06,
+      "loss": 1.3147,
+      "step": 11764
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2870130761320488e-06,
+      "loss": 1.0437,
+      "step": 11765
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.282588428944842e-06,
+      "loss": 1.7766,
+      "step": 11766
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2781713515592874e-06,
+      "loss": 1.1469,
+      "step": 11767
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2737618443141364e-06,
+      "loss": 1.0396,
+      "step": 11768
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.269359907547496e-06,
+      "loss": 1.216,
+      "step": 11769
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2649655415969185e-06,
+      "loss": 1.2965,
+      "step": 11770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2605787467993791e-06,
+      "loss": 0.9837,
+      "step": 11771
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2561995234912416e-06,
+      "loss": 1.1386,
+      "step": 11772
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2518278720083265e-06,
+      "loss": 0.9451,
+      "step": 11773
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2474637926858545e-06,
+      "loss": 0.9851,
+      "step": 11774
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2431072858584802e-06,
+      "loss": 1.5358,
+      "step": 11775
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2387583518602586e-06,
+      "loss": 0.7921,
+      "step": 11776
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2344169910246672e-06,
+      "loss": 0.8207,
+      "step": 11777
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2300832036846066e-06,
+      "loss": 1.1205,
+      "step": 11778
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2257569901724108e-06,
+      "loss": 1.3783,
+      "step": 11779
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2214383508198147e-06,
+      "loss": 1.9729,
+      "step": 11780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2171272859579751e-06,
+      "loss": 0.9838,
+      "step": 11781
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2128237959174837e-06,
+      "loss": 1.6153,
+      "step": 11782
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2085278810283317e-06,
+      "loss": 1.1872,
+      "step": 11783
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2042395416199227e-06,
+      "loss": 1.5247,
+      "step": 11784
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1999587780211042e-06,
+      "loss": 1.2826,
+      "step": 11785
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.195685590560136e-06,
+      "loss": 1.423,
+      "step": 11786
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1914199795647007e-06,
+      "loss": 1.228,
+      "step": 11787
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1871619453618588e-06,
+      "loss": 1.4657,
+      "step": 11788
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.182911488278149e-06,
+      "loss": 1.1958,
+      "step": 11789
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1786686086395104e-06,
+      "loss": 0.8537,
+      "step": 11790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.174433306771261e-06,
+      "loss": 1.3354,
+      "step": 11791
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1702055829982073e-06,
+      "loss": 0.669,
+      "step": 11792
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1659854376445124e-06,
+      "loss": 1.4137,
+      "step": 11793
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.161772871033784e-06,
+      "loss": 1.3294,
+      "step": 11794
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1575678834890635e-06,
+      "loss": 1.0263,
+      "step": 11795
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1533704753327823e-06,
+      "loss": 0.9505,
+      "step": 11796
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.149180646886816e-06,
+      "loss": 1.5357,
+      "step": 11797
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.144998398472441e-06,
+      "loss": 1.1915,
+      "step": 11798
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1408237304103564e-06,
+      "loss": 0.2469,
+      "step": 11799
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1366566430206837e-06,
+      "loss": 1.0393,
+      "step": 11800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1324971366229565e-06,
+      "loss": 0.9465,
+      "step": 11801
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1283452115361414e-06,
+      "loss": 1.7139,
+      "step": 11802
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1242008680786175e-06,
+      "loss": 1.0875,
+      "step": 11803
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1200641065681528e-06,
+      "loss": 1.0507,
+      "step": 11804
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1159349273219932e-06,
+      "loss": 1.2506,
+      "step": 11805
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1118133306567525e-06,
+      "loss": 1.8625,
+      "step": 11806
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1076993168884775e-06,
+      "loss": 1.2298,
+      "step": 11807
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1035928863326605e-06,
+      "loss": 1.351,
+      "step": 11808
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.099494039304172e-06,
+      "loss": 1.0776,
+      "step": 11809
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0954027761173158e-06,
+      "loss": 1.0727,
+      "step": 11810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0913190970858078e-06,
+      "loss": 1.7739,
+      "step": 11811
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0872430025228087e-06,
+      "loss": 0.9143,
+      "step": 11812
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0831744927408905e-06,
+      "loss": 1.1556,
+      "step": 11813
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0791135680519926e-06,
+      "loss": 1.1443,
+      "step": 11814
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0750602287675438e-06,
+      "loss": 0.995,
+      "step": 11815
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0710144751983509e-06,
+      "loss": 1.1959,
+      "step": 11816
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0669763076546547e-06,
+      "loss": 1.3254,
+      "step": 11817
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0629457264460963e-06,
+      "loss": 1.198,
+      "step": 11818
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0589227318817618e-06,
+      "loss": 1.1981,
+      "step": 11819
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0549073242701268e-06,
+      "loss": 1.5409,
+      "step": 11820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0508995039191116e-06,
+      "loss": 1.0616,
+      "step": 11821
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.046899271136026e-06,
+      "loss": 1.5168,
+      "step": 11822
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0429066262276243e-06,
+      "loss": 1.1701,
+      "step": 11823
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0389215695000731e-06,
+      "loss": 1.2495,
+      "step": 11824
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.03494410125895e-06,
+      "loss": 1.2969,
+      "step": 11825
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0309742218092444e-06,
+      "loss": 1.1812,
+      "step": 11826
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0270119314553684e-06,
+      "loss": 1.1194,
+      "step": 11827
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0230572305011787e-06,
+      "loss": 1.034,
+      "step": 11828
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0191101192499108e-06,
+      "loss": 1.8199,
+      "step": 11829
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0151705980042448e-06,
+      "loss": 1.2819,
+      "step": 11830
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.011238667066261e-06,
+      "loss": 0.8424,
+      "step": 11831
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.007314326737463e-06,
+      "loss": 1.2597,
+      "step": 11832
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0033975773187876e-06,
+      "loss": 1.1006,
+      "step": 11833
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.994884191105725e-07,
+      "loss": 1.3888,
+      "step": 11834
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.95586852412589e-07,
+      "loss": 1.3531,
+      "step": 11835
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.916928775239865e-07,
+      "loss": 0.9478,
+      "step": 11836
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.878064947433707e-07,
+      "loss": 1.2999,
+      "step": 11837
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.839277043687701e-07,
+      "loss": 1.5151,
+      "step": 11838
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.80056506697613e-07,
+      "loss": 0.8143,
+      "step": 11839
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.761929020267401e-07,
+      "loss": 1.351,
+      "step": 11840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.723368906524254e-07,
+      "loss": 0.9625,
+      "step": 11841
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.684884728703325e-07,
+      "loss": 1.0946,
+      "step": 11842
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.64647648975603e-07,
+      "loss": 1.3049,
+      "step": 11843
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.608144192627232e-07,
+      "loss": 1.0426,
+      "step": 11844
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.569887840256586e-07,
+      "loss": 1.3075,
+      "step": 11845
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.531707435577297e-07,
+      "loss": 1.595,
+      "step": 11846
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.493602981517468e-07,
+      "loss": 1.5094,
+      "step": 11847
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.455574480998763e-07,
+      "loss": 1.2978,
+      "step": 11848
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.417621936937293e-07,
+      "loss": 0.8571,
+      "step": 11849
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.379745352243397e-07,
+      "loss": 0.9948,
+      "step": 11850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.341944729821528e-07,
+      "loss": 1.1855,
+      "step": 11851
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.304220072570258e-07,
+      "loss": 0.9892,
+      "step": 11852
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.266571383382273e-07,
+      "loss": 0.8438,
+      "step": 11853
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.228998665144595e-07,
+      "loss": 1.5523,
+      "step": 11854
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.191501920738477e-07,
+      "loss": 1.0138,
+      "step": 11855
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.154081153039063e-07,
+      "loss": 1.208,
+      "step": 11856
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.116736364915945e-07,
+      "loss": 1.4273,
+      "step": 11857
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.079467559232613e-07,
+      "loss": 1.9473,
+      "step": 11858
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.042274738847112e-07,
+      "loss": 1.2587,
+      "step": 11859
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.005157906611383e-07,
+      "loss": 2.0988,
+      "step": 11860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.968117065371484e-07,
+      "loss": 1.4485,
+      "step": 11861
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.93115221796792e-07,
+      "loss": 1.5062,
+      "step": 11862
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.894263367235089e-07,
+      "loss": 1.4186,
+      "step": 11863
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.85745051600173e-07,
+      "loss": 1.2177,
+      "step": 11864
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.820713667090697e-07,
+      "loss": 0.9762,
+      "step": 11865
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.784052823319067e-07,
+      "loss": 0.8251,
+      "step": 11866
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.747467987498037e-07,
+      "loss": 1.2114,
+      "step": 11867
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.710959162432919e-07,
+      "loss": 1.558,
+      "step": 11868
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.674526350923362e-07,
+      "loss": 1.8013,
+      "step": 11869
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.63816955576302e-07,
+      "loss": 1.095,
+      "step": 11870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.601888779739775e-07,
+      "loss": 1.7123,
+      "step": 11871
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.565684025635845e-07,
+      "loss": 1.3214,
+      "step": 11872
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.529555296227342e-07,
+      "loss": 1.0054,
+      "step": 11873
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.493502594284608e-07,
+      "loss": 0.9236,
+      "step": 11874
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.457525922572318e-07,
+      "loss": 1.1836,
+      "step": 11875
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.421625283849266e-07,
+      "loss": 1.3434,
+      "step": 11876
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.38580068086825e-07,
+      "loss": 0.8678,
+      "step": 11877
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.350052116376406e-07,
+      "loss": 0.8597,
+      "step": 11878
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.314379593114874e-07,
+      "loss": 1.2277,
+      "step": 11879
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.278783113819244e-07,
+      "loss": 1.5356,
+      "step": 11880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.24326268121911e-07,
+      "loss": 0.6268,
+      "step": 11881
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.207818298038184e-07,
+      "loss": 1.6049,
+      "step": 11882
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.172449966994399e-07,
+      "loss": 1.0787,
+      "step": 11883
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.137157690799813e-07,
+      "loss": 1.321,
+      "step": 11884
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.101941472160701e-07,
+      "loss": 1.0817,
+      "step": 11885
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.066801313777462e-07,
+      "loss": 1.2663,
+      "step": 11886
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.031737218344826e-07,
+      "loss": 0.8714,
+      "step": 11887
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.996749188551422e-07,
+      "loss": 1.3278,
+      "step": 11888
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.961837227080326e-07,
+      "loss": 1.0074,
+      "step": 11889
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.927001336608509e-07,
+      "loss": 1.1688,
+      "step": 11890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.892241519807275e-07,
+      "loss": 1.5351,
+      "step": 11891
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.857557779342162e-07,
+      "loss": 0.9895,
+      "step": 11892
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.822950117872597e-07,
+      "loss": 1.5299,
+      "step": 11893
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.788418538052456e-07,
+      "loss": 1.0986,
+      "step": 11894
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.753963042529732e-07,
+      "loss": 1.1812,
+      "step": 11895
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.719583633946314e-07,
+      "loss": 2.0547,
+      "step": 11896
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.685280314938647e-07,
+      "loss": 1.4188,
+      "step": 11897
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.651053088137294e-07,
+      "loss": 0.6861,
+      "step": 11898
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.616901956166489e-07,
+      "loss": 1.1846,
+      "step": 11899
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.582826921645247e-07,
+      "loss": 0.651,
+      "step": 11900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.548827987186369e-07,
+      "loss": 0.3576,
+      "step": 11901
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.514905155396989e-07,
+      "loss": 0.9656,
+      "step": 11902
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.481058428878473e-07,
+      "loss": 1.7325,
+      "step": 11903
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.447287810226078e-07,
+      "loss": 1.24,
+      "step": 11904
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.413593302029397e-07,
+      "loss": 1.1154,
+      "step": 11905
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.379974906872255e-07,
+      "loss": 1.0802,
+      "step": 11906
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.346432627332589e-07,
+      "loss": 1.2012,
+      "step": 11907
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.312966465982452e-07,
+      "loss": 1.0102,
+      "step": 11908
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.279576425388013e-07,
+      "loss": 1.3352,
+      "step": 11909
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.246262508109669e-07,
+      "loss": 1.6508,
+      "step": 11910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.213024716702044e-07,
+      "loss": 0.6698,
+      "step": 11911
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.179863053713764e-07,
+      "loss": 1.5093,
+      "step": 11912
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.146777521687909e-07,
+      "loss": 1.2869,
+      "step": 11913
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.113768123161335e-07,
+      "loss": 1.4795,
+      "step": 11914
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.080834860665352e-07,
+      "loss": 1.6522,
+      "step": 11915
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.047977736725275e-07,
+      "loss": 0.9552,
+      "step": 11916
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.015196753860643e-07,
+      "loss": 1.4778,
+      "step": 11917
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.982491914585221e-07,
+      "loss": 0.7162,
+      "step": 11918
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.949863221406894e-07,
+      "loss": 1.245,
+      "step": 11919
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.917310676827548e-07,
+      "loss": 1.0487,
+      "step": 11920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.884834283343411e-07,
+      "loss": 1.0133,
+      "step": 11921
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.852434043444822e-07,
+      "loss": 0.716,
+      "step": 11922
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.820109959616239e-07,
+      "loss": 1.162,
+      "step": 11923
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.787862034336456e-07,
+      "loss": 0.5782,
+      "step": 11924
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.755690270078275e-07,
+      "loss": 1.0423,
+      "step": 11925
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.723594669308498e-07,
+      "loss": 1.3157,
+      "step": 11926
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.69157523448849e-07,
+      "loss": 1.1979,
+      "step": 11927
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.659631968073288e-07,
+      "loss": 0.7632,
+      "step": 11928
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.627764872512598e-07,
+      "loss": 1.0694,
+      "step": 11929
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.595973950249912e-07,
+      "loss": 1.0943,
+      "step": 11930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.564259203723056e-07,
+      "loss": 1.4511,
+      "step": 11931
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.532620635363973e-07,
+      "loss": 1.2009,
+      "step": 11932
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.501058247598724e-07,
+      "loss": 1.2367,
+      "step": 11933
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.469572042847482e-07,
+      "loss": 1.4458,
+      "step": 11934
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.438162023524875e-07,
+      "loss": 1.8695,
+      "step": 11935
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.406828192039305e-07,
+      "loss": 1.0482,
+      "step": 11936
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.37557055079352e-07,
+      "loss": 1.3604,
+      "step": 11937
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.344389102184379e-07,
+      "loss": 0.9478,
+      "step": 11938
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.313283848602969e-07,
+      "loss": 1.1258,
+      "step": 11939
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.282254792434494e-07,
+      "loss": 1.4816,
+      "step": 11940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.251301936058274e-07,
+      "loss": 1.016,
+      "step": 11941
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.220425281847741e-07,
+      "loss": 1.2664,
+      "step": 11942
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.189624832170781e-07,
+      "loss": 1.0694,
+      "step": 11943
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.158900589389171e-07,
+      "loss": 1.4528,
+      "step": 11944
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.128252555858693e-07,
+      "loss": 1.4964,
+      "step": 11945
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.09768073392969e-07,
+      "loss": 1.3773,
+      "step": 11946
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.067185125946395e-07,
+      "loss": 0.6606,
+      "step": 11947
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.036765734247163e-07,
+      "loss": 1.2918,
+      "step": 11948
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.006422561164793e-07,
+      "loss": 1.2678,
+      "step": 11949
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.976155609025869e-07,
+      "loss": 0.949,
+      "step": 11950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.945964880151423e-07,
+      "loss": 1.4321,
+      "step": 11951
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.91585037685638e-07,
+      "loss": 1.147,
+      "step": 11952
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.885812101450117e-07,
+      "loss": 1.0194,
+      "step": 11953
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.855850056235901e-07,
+      "loss": 1.7586,
+      "step": 11954
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.825964243511339e-07,
+      "loss": 1.2764,
+      "step": 11955
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.796154665568154e-07,
+      "loss": 1.4008,
+      "step": 11956
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.766421324692184e-07,
+      "loss": 1.2775,
+      "step": 11957
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.736764223163271e-07,
+      "loss": 0.8503,
+      "step": 11958
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.707183363255708e-07,
+      "loss": 1.3404,
+      "step": 11959
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.677678747237902e-07,
+      "loss": 1.4586,
+      "step": 11960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.648250377372044e-07,
+      "loss": 1.6379,
+      "step": 11961
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.618898255914995e-07,
+      "loss": 1.3775,
+      "step": 11962
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.589622385117288e-07,
+      "loss": 1.7578,
+      "step": 11963
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.560422767224016e-07,
+      "loss": 0.9744,
+      "step": 11964
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.531299404474277e-07,
+      "loss": 1.0268,
+      "step": 11965
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.502252299101063e-07,
+      "loss": 0.9412,
+      "step": 11966
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.473281453332036e-07,
+      "loss": 1.7514,
+      "step": 11967
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.444386869388529e-07,
+      "loss": 0.8398,
+      "step": 11968
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.415568549486328e-07,
+      "loss": 1.6191,
+      "step": 11969
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.386826495835107e-07,
+      "loss": 1.7756,
+      "step": 11970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.358160710639104e-07,
+      "loss": 0.8683,
+      "step": 11971
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.329571196096339e-07,
+      "loss": 1.5099,
+      "step": 11972
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.301057954398947e-07,
+      "loss": 1.5853,
+      "step": 11973
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.272620987733623e-07,
+      "loss": 1.1231,
+      "step": 11974
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.244260298280735e-07,
+      "loss": 1.5535,
+      "step": 11975
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.215975888215208e-07,
+      "loss": 1.074,
+      "step": 11976
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.187767759705753e-07,
+      "loss": 0.5838,
+      "step": 11977
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.159635914915639e-07,
+      "loss": 1.2828,
+      "step": 11978
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.131580356001808e-07,
+      "loss": 1.0596,
+      "step": 11979
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.103601085115761e-07,
+      "loss": 1.2679,
+      "step": 11980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.075698104403004e-07,
+      "loss": 1.0932,
+      "step": 11981
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.047871416003047e-07,
+      "loss": 1.4943,
+      "step": 11982
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.02012102204985e-07,
+      "loss": 0.9065,
+      "step": 11983
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.992446924671157e-07,
+      "loss": 1.2467,
+      "step": 11984
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.964849125989269e-07,
+      "loss": 1.4512,
+      "step": 11985
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.93732762812027e-07,
+      "loss": 1.229,
+      "step": 11986
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.909882433174695e-07,
+      "loss": 1.3792,
+      "step": 11987
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.882513543256973e-07,
+      "loss": 1.1814,
+      "step": 11988
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.855220960465756e-07,
+      "loss": 1.1204,
+      "step": 11989
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.828004686893928e-07,
+      "loss": 1.3017,
+      "step": 11990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.800864724628596e-07,
+      "loss": 1.0054,
+      "step": 11991
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.77380107575065e-07,
+      "loss": 1.2535,
+      "step": 11992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.746813742335654e-07,
+      "loss": 1.6252,
+      "step": 11993
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7199027264528406e-07,
+      "loss": 1.2622,
+      "step": 11994
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6930680301657814e-07,
+      "loss": 2.0249,
+      "step": 11995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.666309655532386e-07,
+      "loss": 1.2979,
+      "step": 11996
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6396276046042354e-07,
+      "loss": 1.1443,
+      "step": 11997
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6130218794276924e-07,
+      "loss": 1.4969,
+      "step": 11998
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5864924820426816e-07,
+      "loss": 1.1514,
+      "step": 11999
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5600394144835747e-07,
+      "loss": 1.7518,
+      "step": 12000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5336626787788607e-07,
+      "loss": 1.9576,
+      "step": 12001
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5073622769511436e-07,
+      "loss": 1.046,
+      "step": 12002
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.481138211017033e-07,
+      "loss": 1.0854,
+      "step": 12003
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4549904829876975e-07,
+      "loss": 1.62,
+      "step": 12004
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4289190948679784e-07,
+      "loss": 1.3354,
+      "step": 12005
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.402924048657275e-07,
+      "loss": 0.7761,
+      "step": 12006
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3770053463486616e-07,
+      "loss": 1.0642,
+      "step": 12007
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.351162989929769e-07,
+      "loss": 1.6353,
+      "step": 12008
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.325396981382346e-07,
+      "loss": 1.7411,
+      "step": 12009
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2997073226819227e-07,
+      "loss": 0.7963,
+      "step": 12010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.27409401579848e-07,
+      "loss": 0.9305,
+      "step": 12011
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2485570626962233e-07,
+      "loss": 1.7514,
+      "step": 12012
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2230964653332536e-07,
+      "loss": 0.6688,
+      "step": 12013
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.197712225662009e-07,
+      "loss": 2.0107,
+      "step": 12014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.172404345628933e-07,
+      "loss": 1.1578,
+      "step": 12015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1471728271746945e-07,
+      "loss": 1.4352,
+      "step": 12016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.122017672233969e-07,
+      "loss": 1.1303,
+      "step": 12017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.096938882735879e-07,
+      "loss": 1.8898,
+      "step": 12018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.071936460603443e-07,
+      "loss": 0.8179,
+      "step": 12019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.047010407753904e-07,
+      "loss": 1.2389,
+      "step": 12020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.022160726098512e-07,
+      "loss": 0.9324,
+      "step": 12021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9973874175428527e-07,
+      "loss": 1.4947,
+      "step": 12022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9726904839865187e-07,
+      "loss": 1.1819,
+      "step": 12023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.948069927323439e-07,
+      "loss": 1.2245,
+      "step": 12024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9235257494415476e-07,
+      "loss": 1.0506,
+      "step": 12025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8990579522227846e-07,
+      "loss": 0.8846,
+      "step": 12026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8746665375435367e-07,
+      "loss": 1.6257,
+      "step": 12027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.850351507274086e-07,
+      "loss": 1.3354,
+      "step": 12028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8261128632788303e-07,
+      "loss": 1.1289,
+      "step": 12029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8019506074167264e-07,
+      "loss": 1.1865,
+      "step": 12030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.777864741540404e-07,
+      "loss": 1.367,
+      "step": 12031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7538552674967196e-07,
+      "loss": 1.2167,
+      "step": 12032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7299221871268665e-07,
+      "loss": 1.454,
+      "step": 12033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.706065502266043e-07,
+      "loss": 1.3926,
+      "step": 12034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.682285214743564e-07,
+      "loss": 1.3014,
+      "step": 12035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6585813263830817e-07,
+      "loss": 1.0347,
+      "step": 12036
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.634953839002142e-07,
+      "loss": 1.0241,
+      "step": 12037
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.611402754412518e-07,
+      "loss": 1.0028,
+      "step": 12038
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5879280744202104e-07,
+      "loss": 1.3777,
+      "step": 12039
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5645298008252224e-07,
+      "loss": 1.0804,
+      "step": 12040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5412079354217866e-07,
+      "loss": 0.9357,
+      "step": 12041
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.51796247999836e-07,
+      "loss": 1.2924,
+      "step": 12042
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.494793436337296e-07,
+      "loss": 1.3476,
+      "step": 12043
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4717008062153947e-07,
+      "loss": 1.4824,
+      "step": 12044
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4486845914032394e-07,
+      "loss": 1.3474,
+      "step": 12045
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4257447936658636e-07,
+      "loss": 1.2103,
+      "step": 12046
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.402881414762304e-07,
+      "loss": 1.2419,
+      "step": 12047
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3800944564457147e-07,
+      "loss": 1.205,
+      "step": 12048
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.357383920463586e-07,
+      "loss": 1.2866,
+      "step": 12049
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3347498085571916e-07,
+      "loss": 1.4016,
+      "step": 12050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.312192122462254e-07,
+      "loss": 1.3002,
+      "step": 12051
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2897108639086127e-07,
+      "loss": 0.9666,
+      "step": 12052
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.267306034619999e-07,
+      "loss": 1.194,
+      "step": 12053
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.244977636314595e-07,
+      "loss": 1.3528,
+      "step": 12054
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.222725670704363e-07,
+      "loss": 1.0766,
+      "step": 12055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.200550139495939e-07,
+      "loss": 2.1022,
+      "step": 12056
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.178451044389519e-07,
+      "loss": 0.7668,
+      "step": 12057
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.156428387079857e-07,
+      "loss": 1.377,
+      "step": 12058
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.134482169255493e-07,
+      "loss": 1.1957,
+      "step": 12059
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1126123925995234e-07,
+      "loss": 1.6864,
+      "step": 12060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0908190587887184e-07,
+      "loss": 0.9322,
+      "step": 12061
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.069102169494409e-07,
+      "loss": 1.2028,
+      "step": 12062
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0474617263818174e-07,
+      "loss": 1.355,
+      "step": 12063
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0258977311103944e-07,
+      "loss": 0.6777,
+      "step": 12064
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.004410185333706e-07,
+      "loss": 1.3805,
+      "step": 12065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.982999090699323e-07,
+      "loss": 1.2075,
+      "step": 12066
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.961664448849155e-07,
+      "loss": 0.8701,
+      "step": 12067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9404062614192265e-07,
+      "loss": 1.339,
+      "step": 12068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.919224530039677e-07,
+      "loss": 1.0195,
+      "step": 12069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8981192563346526e-07,
+      "loss": 1.0444,
+      "step": 12070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8770904419225255e-07,
+      "loss": 1.3023,
+      "step": 12071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.856138088415894e-07,
+      "loss": 0.9929,
+      "step": 12072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8352621974214734e-07,
+      "loss": 1.8318,
+      "step": 12073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.814462770539872e-07,
+      "loss": 1.1797,
+      "step": 12074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.793739809366258e-07,
+      "loss": 1.1203,
+      "step": 12075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.773093315489472e-07,
+      "loss": 1.0613,
+      "step": 12076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7525232904928033e-07,
+      "loss": 0.9163,
+      "step": 12077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.732029735953545e-07,
+      "loss": 0.8767,
+      "step": 12078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7116126534433296e-07,
+      "loss": 1.5622,
+      "step": 12079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6912720445276817e-07,
+      "loss": 0.8644,
+      "step": 12080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.671007910766354e-07,
+      "loss": 1.0177,
+      "step": 12081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.650820253713104e-07,
+      "loss": 1.2128,
+      "step": 12082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.630709074916249e-07,
+      "loss": 0.9248,
+      "step": 12083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6106743759176656e-07,
+      "loss": 1.0623,
+      "step": 12084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.590716158253681e-07,
+      "loss": 1.1556,
+      "step": 12085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5708344234548485e-07,
+      "loss": 1.2253,
+      "step": 12086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.551029173045616e-07,
+      "loss": 1.3056,
+      "step": 12087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5313004085447676e-07,
+      "loss": 0.6911,
+      "step": 12088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.511648131465205e-07,
+      "loss": 0.9544,
+      "step": 12089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4920723433136116e-07,
+      "loss": 1.0208,
+      "step": 12090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.472573045591342e-07,
+      "loss": 1.3056,
+      "step": 12091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4531502397935334e-07,
+      "loss": 1.3848,
+      "step": 12092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4338039274095503e-07,
+      "loss": 0.7657,
+      "step": 12093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.414534109922983e-07,
+      "loss": 1.5215,
+      "step": 12094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3953407888114286e-07,
+      "loss": 1.5213,
+      "step": 12095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.376223965546598e-07,
+      "loss": 1.1866,
+      "step": 12096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3571836415944293e-07,
+      "loss": 1.375,
+      "step": 12097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3382198184149772e-07,
+      "loss": 1.2095,
+      "step": 12098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.319332497462523e-07,
+      "loss": 1.5753,
+      "step": 12099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3005216801851304e-07,
+      "loss": 0.8195,
+      "step": 12100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2817873680254232e-07,
+      "loss": 1.2783,
+      "step": 12101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2631295624199188e-07,
+      "loss": 1.1094,
+      "step": 12102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.244548264799362e-07,
+      "loss": 1.4652,
+      "step": 12103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2260434765886128e-07,
+      "loss": 1.6006,
+      "step": 12104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2076151992065364e-07,
+      "loss": 1.2245,
+      "step": 12105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1892634340663353e-07,
+      "loss": 0.7913,
+      "step": 12106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1709881825753286e-07,
+      "loss": 0.8517,
+      "step": 12107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1527894461346177e-07,
+      "loss": 1.0568,
+      "step": 12108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1346672261399747e-07,
+      "loss": 1.2625,
+      "step": 12109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.116621523980955e-07,
+      "loss": 0.8952,
+      "step": 12110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0986523410413405e-07,
+      "loss": 1.3131,
+      "step": 12111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0807596786990292e-07,
+      "loss": 0.9848,
+      "step": 12112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0629435383259232e-07,
+      "loss": 1.0399,
+      "step": 12113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0452039212883744e-07,
+      "loss": 0.8161,
+      "step": 12114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0275408289467391e-07,
+      "loss": 1.5891,
+      "step": 12115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.009954262655156e-07,
+      "loss": 1.6243,
+      "step": 12116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9924442237624351e-07,
+      "loss": 1.5853,
+      "step": 12117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9750107136112806e-07,
+      "loss": 1.2416,
+      "step": 12118
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9576537335384006e-07,
+      "loss": 1.5888,
+      "step": 12119
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9403732848747302e-07,
+      "loss": 1.6577,
+      "step": 12120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9231693689455432e-07,
+      "loss": 1.4998,
+      "step": 12121
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9060419870698953e-07,
+      "loss": 1.1071,
+      "step": 12122
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8889911405611804e-07,
+      "loss": 1.4975,
+      "step": 12123
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.872016830726908e-07,
+      "loss": 1.254,
+      "step": 12124
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.855119058868704e-07,
+      "loss": 1.7561,
+      "step": 12125
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8382978262821983e-07,
+      "loss": 0.8619,
+      "step": 12126
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8215531342574698e-07,
+      "loss": 0.9827,
+      "step": 12127
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.804884984078381e-07,
+      "loss": 1.1885,
+      "step": 12128
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7882933770231313e-07,
+      "loss": 1.453,
+      "step": 12129
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7717783143639256e-07,
+      "loss": 0.7687,
+      "step": 12130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.755339797367306e-07,
+      "loss": 1.3898,
+      "step": 12131
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7389778272935976e-07,
+      "loss": 1.1292,
+      "step": 12132
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7226924053976856e-07,
+      "loss": 0.8594,
+      "step": 12133
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.706483532928127e-07,
+      "loss": 1.4897,
+      "step": 12134
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6903512111280384e-07,
+      "loss": 1.5007,
+      "step": 12135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6742954412344303e-07,
+      "loss": 0.8282,
+      "step": 12136
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.658316224478318e-07,
+      "loss": 0.9324,
+      "step": 12137
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6424135620852764e-07,
+      "loss": 0.9884,
+      "step": 12138
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.626587455274553e-07,
+      "loss": 1.453,
+      "step": 12139
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6108379052597323e-07,
+      "loss": 1.265,
+      "step": 12140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5951649132486256e-07,
+      "loss": 1.3952,
+      "step": 12141
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.579568480442939e-07,
+      "loss": 0.9236,
+      "step": 12142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.564048608038715e-07,
+      "loss": 1.3237,
+      "step": 12143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5486052972260025e-07,
+      "loss": 1.1714,
+      "step": 12144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5332385491890756e-07,
+      "loss": 1.3972,
+      "step": 12145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5179483651061034e-07,
+      "loss": 1.034,
+      "step": 12146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5027347461498143e-07,
+      "loss": 1.2848,
+      "step": 12147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4875976934867198e-07,
+      "loss": 1.5105,
+      "step": 12148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.472537208277447e-07,
+      "loss": 1.3308,
+      "step": 12149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4575532916768497e-07,
+      "loss": 0.9506,
+      "step": 12150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.44264594483412e-07,
+      "loss": 1.0619,
+      "step": 12151
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4278151688922327e-07,
+      "loss": 1.3496,
+      "step": 12152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4130609649883886e-07,
+      "loss": 1.0879,
+      "step": 12153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3983833342541274e-07,
+      "loss": 1.4016,
+      "step": 12154
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3837822778147712e-07,
+      "loss": 0.9406,
+      "step": 12155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.36925779678998e-07,
+      "loss": 1.1891,
+      "step": 12156
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3548098922936404e-07,
+      "loss": 1.1558,
+      "step": 12157
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3404385654336437e-07,
+      "loss": 1.2268,
+      "step": 12158
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3261438173118868e-07,
+      "loss": 1.2286,
+      "step": 12159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3119256490244924e-07,
+      "loss": 1.5089,
+      "step": 12160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.297784061661811e-07,
+      "loss": 1.8276,
+      "step": 12161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.283719056308197e-07,
+      "loss": 1.2936,
+      "step": 12162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.269730634042232e-07,
+      "loss": 1.2963,
+      "step": 12163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2558187959366141e-07,
+      "loss": 1.0935,
+      "step": 12164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.241983543057934e-07,
+      "loss": 0.6589,
+      "step": 12165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2282248764672321e-07,
+      "loss": 0.9485,
+      "step": 12166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2145427972195533e-07,
+      "loss": 1.1972,
+      "step": 12167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2009373063640583e-07,
+      "loss": 0.9524,
+      "step": 12168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1874084049440238e-07,
+      "loss": 1.0717,
+      "step": 12169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1739560939967309e-07,
+      "loss": 2.3438,
+      "step": 12170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1605803745540211e-07,
+      "loss": 1.1463,
+      "step": 12171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1472812476412964e-07,
+      "loss": 1.3249,
+      "step": 12172
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1340587142784076e-07,
+      "loss": 1.1673,
+      "step": 12173
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1209127754794324e-07,
+      "loss": 1.1392,
+      "step": 12174
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1078434322522313e-07,
+      "loss": 1.3352,
+      "step": 12175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0948506855990027e-07,
+      "loss": 1.0595,
+      "step": 12176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0819345365162825e-07,
+      "loss": 0.9179,
+      "step": 12177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0690949859941679e-07,
+      "loss": 1.0802,
+      "step": 12178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0563320350174267e-07,
+      "loss": 1.0853,
+      "step": 12179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0436456845647203e-07,
+      "loss": 1.5431,
+      "step": 12180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0310359356087151e-07,
+      "loss": 1.9498,
+      "step": 12181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0185027891166376e-07,
+      "loss": 0.9274,
+      "step": 12182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0060462460491638e-07,
+      "loss": 1.4933,
+      "step": 12183
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.936663073618625e-08,
+      "loss": 1.4985,
+      "step": 12184
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.813629740037522e-08,
+      "loss": 1.1772,
+      "step": 12185
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.691362469184117e-08,
+      "loss": 1.0086,
+      "step": 12186
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.569861270434244e-08,
+      "loss": 2.0031,
+      "step": 12187
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.449126153103782e-08,
+      "loss": 1.6952,
+      "step": 12188
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.329157126451992e-08,
+      "loss": 1.1475,
+      "step": 12189
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.209954199677073e-08,
+      "loss": 1.1458,
+      "step": 12190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.0915173819206e-08,
+      "loss": 1.6255,
+      "step": 12191
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.973846682264197e-08,
+      "loss": 0.8544,
+      "step": 12192
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.856942109731758e-08,
+      "loss": 0.8664,
+      "step": 12193
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.740803673286113e-08,
+      "loss": 0.9992,
+      "step": 12194
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.625431381834582e-08,
+      "loss": 1.0736,
+      "step": 12195
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.510825244222309e-08,
+      "loss": 0.7134,
+      "step": 12196
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.396985269238933e-08,
+      "loss": 0.9641,
+      "step": 12197
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.283911465614136e-08,
+      "loss": 0.9286,
+      "step": 12198
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.17160384201765e-08,
+      "loss": 1.0181,
+      "step": 12199
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.060062407061475e-08,
+      "loss": 1.004,
+      "step": 12200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.949287169298769e-08,
+      "loss": 1.2596,
+      "step": 12201
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.839278137224959e-08,
+      "loss": 1.1603,
+      "step": 12202
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.730035319273298e-08,
+      "loss": 1.1944,
+      "step": 12203
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.621558723823752e-08,
+      "loss": 1.1918,
+      "step": 12204
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.513848359191888e-08,
+      "loss": 0.8643,
+      "step": 12205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.406904233638878e-08,
+      "loss": 1.4891,
+      "step": 12206
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.300726355364829e-08,
+      "loss": 1.0387,
+      "step": 12207
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.195314732511005e-08,
+      "loss": 0.9012,
+      "step": 12208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.090669373162051e-08,
+      "loss": 1.3281,
+      "step": 12209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.986790285340438e-08,
+      "loss": 1.3199,
+      "step": 12210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.883677477012018e-08,
+      "loss": 1.2852,
+      "step": 12211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.781330956084908e-08,
+      "loss": 1.3163,
+      "step": 12212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.679750730407274e-08,
+      "loss": 1.5188,
+      "step": 12213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.578936807766223e-08,
+      "loss": 1.2016,
+      "step": 12214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.478889195894456e-08,
+      "loss": 0.8755,
+      "step": 12215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.379607902463613e-08,
+      "loss": 0.8266,
+      "step": 12216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.281092935085386e-08,
+      "loss": 1.3925,
+      "step": 12217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.183344301315952e-08,
+      "loss": 1.4314,
+      "step": 12218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.086362008649316e-08,
+      "loss": 1.3902,
+      "step": 12219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.990146064521752e-08,
+      "loss": 1.1091,
+      "step": 12220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.894696476314021e-08,
+      "loss": 1.2361,
+      "step": 12221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.800013251342496e-08,
+      "loss": 1.9846,
+      "step": 12222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.706096396868033e-08,
+      "loss": 1.3607,
+      "step": 12223
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.612945920093759e-08,
+      "loss": 1.1576,
+      "step": 12224
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.5205618281617414e-08,
+      "loss": 1.0688,
+      "step": 12225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.4289441281552e-08,
+      "loss": 1.6328,
+      "step": 12226
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3380928271018485e-08,
+      "loss": 1.2586,
+      "step": 12227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.248007931966115e-08,
+      "loss": 1.3049,
+      "step": 12228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.158689449656917e-08,
+      "loss": 1.0003,
+      "step": 12229
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0701373870232215e-08,
+      "loss": 1.2159,
+      "step": 12230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.982351750854042e-08,
+      "loss": 1.1178,
+      "step": 12231
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8953325478839904e-08,
+      "loss": 1.153,
+      "step": 12232
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8090797847821775e-08,
+      "loss": 0.83,
+      "step": 12233
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.723593468165532e-08,
+      "loss": 1.3438,
+      "step": 12234
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.638873604587701e-08,
+      "loss": 1.9252,
+      "step": 12235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5549202005457094e-08,
+      "loss": 1.0994,
+      "step": 12236
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4717332624766297e-08,
+      "loss": 0.7929,
+      "step": 12237
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3893127967598034e-08,
+      "loss": 1.1163,
+      "step": 12238
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.307658809715731e-08,
+      "loss": 0.8804,
+      "step": 12239
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2267713076038495e-08,
+      "loss": 0.8037,
+      "step": 12240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.146650296630306e-08,
+      "loss": 1.2642,
+      "step": 12241
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.067295782935743e-08,
+      "loss": 1.2577,
+      "step": 12242
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9887077726064036e-08,
+      "loss": 1.4823,
+      "step": 12243
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.910886271668579e-08,
+      "loss": 1.3895,
+      "step": 12244
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8338312860886074e-08,
+      "loss": 0.8578,
+      "step": 12245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.757542821777316e-08,
+      "loss": 1.0596,
+      "step": 12246
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.682020884583359e-08,
+      "loss": 0.1259,
+      "step": 12247
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.607265480297661e-08,
+      "loss": 1.2098,
+      "step": 12248
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5332766146523036e-08,
+      "loss": 0.6215,
+      "step": 12249
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4600542933227456e-08,
+      "loss": 1.3964,
+      "step": 12250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.387598521921165e-08,
+      "loss": 1.2294,
+      "step": 12251
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.315909306006448e-08,
+      "loss": 0.6066,
+      "step": 12252
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2449866510730896e-08,
+      "loss": 1.1774,
+      "step": 12253
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.174830562561182e-08,
+      "loss": 0.7614,
+      "step": 12254
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.105441045849755e-08,
+      "loss": 0.6054,
+      "step": 12255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0368181062589983e-08,
+      "loss": 1.6249,
+      "step": 12256
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.968961749053589e-08,
+      "loss": 1.2764,
+      "step": 12257
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9018719794338124e-08,
+      "loss": 0.9377,
+      "step": 12258
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8355488025466613e-08,
+      "loss": 1.007,
+      "step": 12259
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7699922234769582e-08,
+      "loss": 1.5509,
+      "step": 12260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.705202247250682e-08,
+      "loss": 0.6998,
+      "step": 12261
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6411788788371916e-08,
+      "loss": 0.9641,
+      "step": 12262
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.577922123145893e-08,
+      "loss": 1.1618,
+      "step": 12263
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5154319850262397e-08,
+      "loss": 0.8804,
+      "step": 12264
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4537084692721756e-08,
+      "loss": 0.8952,
+      "step": 12265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3927515806143608e-08,
+      "loss": 1.554,
+      "step": 12266
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3325613237290545e-08,
+      "loss": 1.4356,
+      "step": 12267
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2731377032314536e-08,
+      "loss": 1.126,
+      "step": 12268
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2144807236768038e-08,
+      "loss": 0.7813,
+      "step": 12269
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1565903895637284e-08,
+      "loss": 1.1522,
+      "step": 12270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0994667053320095e-08,
+      "loss": 1.6668,
+      "step": 12271
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0431096753614766e-08,
+      "loss": 1.6428,
+      "step": 12272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.987519303974228e-08,
+      "loss": 1.6649,
+      "step": 12273
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9326955954312996e-08,
+      "loss": 0.7769,
+      "step": 12274
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.878638553939327e-08,
+      "loss": 0.8113,
+      "step": 12275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.825348183640552e-08,
+      "loss": 1.3565,
+      "step": 12276
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7728244886239254e-08,
+      "loss": 0.8613,
+      "step": 12277
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7210674729151167e-08,
+      "loss": 0.8168,
+      "step": 12278
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.670077140484283e-08,
+      "loss": 0.9057,
+      "step": 12279
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6198534952405198e-08,
+      "loss": 1.2612,
+      "step": 12280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.570396541036301e-08,
+      "loss": 1.1087,
+      "step": 12281
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.521706281663038e-08,
+      "loss": 1.0413,
+      "step": 12282
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4737827208533005e-08,
+      "loss": 1.2127,
+      "step": 12283
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4266258622841477e-08,
+      "loss": 1.0666,
+      "step": 12284
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3802357095715757e-08,
+      "loss": 1.0177,
+      "step": 12285
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3346122662705184e-08,
+      "loss": 0.7789,
+      "step": 12286
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2897555358815094e-08,
+      "loss": 1.2957,
+      "step": 12287
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2456655218429092e-08,
+      "loss": 0.8594,
+      "step": 12288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2023422275375674e-08,
+      "loss": 1.6355,
+      "step": 12289
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1597856562850506e-08,
+      "loss": 1.3646,
+      "step": 12290
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1179958113505252e-08,
+      "loss": 1.0527,
+      "step": 12291
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0769726959380944e-08,
+      "loss": 0.9848,
+      "step": 12292
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0367163131930202e-08,
+      "loss": 1.1434,
+      "step": 12293
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.972266662028329e-09,
+      "loss": 1.0212,
+      "step": 12294
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.585037579951106e-09,
+      "loss": 1.177,
+      "step": 12295
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.205475915397e-09,
+      "loss": 0.9333,
+      "step": 12296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.833581697464955e-09,
+      "loss": 1.3125,
+      "step": 12297
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.469354954665498e-09,
+      "loss": 1.0514,
+      "step": 12298
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.112795714954046e-09,
+      "loss": 1.8175,
+      "step": 12299
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.763904005653189e-09,
+      "loss": 1.4841,
+      "step": 12300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.422679853519299e-09,
+      "loss": 1.0442,
+      "step": 12301
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.089123284709232e-09,
+      "loss": 1.3661,
+      "step": 12302
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.763234324813628e-09,
+      "loss": 1.318,
+      "step": 12303
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.445012998823608e-09,
+      "loss": 0.8471,
+      "step": 12304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.134459331130771e-09,
+      "loss": 0.9033,
+      "step": 12305
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.831573345560504e-09,
+      "loss": 1.1194,
+      "step": 12306
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.53635506532757e-09,
+      "loss": 1.1604,
+      "step": 12307
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.2488045130694166e-09,
+      "loss": 1.2888,
+      "step": 12308
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.968921710846175e-09,
+      "loss": 1.0376,
+      "step": 12309
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6967066801184565e-09,
+      "loss": 1.2491,
+      "step": 12310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.432159441747352e-09,
+      "loss": 1.4866,
+      "step": 12311
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.175280016027739e-09,
+      "loss": 1.1779,
+      "step": 12312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.926068422654972e-09,
+      "loss": 1.4242,
+      "step": 12313
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6845246807470923e-09,
+      "loss": 1.0216,
+      "step": 12314
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4506488088115186e-09,
+      "loss": 1.2477,
+      "step": 12315
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2244408248005563e-09,
+      "loss": 0.7536,
+      "step": 12316
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0059007460447875e-09,
+      "loss": 1.3773,
+      "step": 12317
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.795028589297477e-09,
+      "loss": 1.0767,
+      "step": 12318
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.591824370745677e-09,
+      "loss": 0.8808,
+      "step": 12319
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.396288105965816e-09,
+      "loss": 1.1378,
+      "step": 12320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2084198099570076e-09,
+      "loss": 1.6013,
+      "step": 12321
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0282194971077416e-09,
+      "loss": 0.7246,
+      "step": 12322
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8556871812513976e-09,
+      "loss": 0.8441,
+      "step": 12323
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6908228756107315e-09,
+      "loss": 1.7302,
+      "step": 12324
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5336265928422856e-09,
+      "loss": 1.1964,
+      "step": 12325
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3840983449697752e-09,
+      "loss": 1.2011,
+      "step": 12326
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2422381434951113e-09,
+      "loss": 0.9576,
+      "step": 12327
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1080459992651725e-09,
+      "loss": 0.9559,
+      "step": 12328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.815219225939309e-10,
+      "loss": 1.6083,
+      "step": 12329
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.626659231625311e-10,
+      "loss": 1.2586,
+      "step": 12330
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.514780101081087e-10,
+      "loss": 1.3441,
+      "step": 12331
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.47958191934972e-10,
+      "loss": 1.2001,
+      "step": 12332
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.521064765923178e-10,
+      "loss": 0.9128,
+      "step": 12333
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6392287144092495e-10,
+      "loss": 1.2022,
+      "step": 12334
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8340738321984705e-10,
+      "loss": 1.4532,
+      "step": 12335
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.105600181130264e-10,
+      "loss": 1.1688,
+      "step": 12336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.453807817048848e-10,
+      "loss": 1.4868,
+      "step": 12337
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8786967900252807e-10,
+      "loss": 0.59,
+      "step": 12338
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.380267144024394e-10,
+      "loss": 2.1156,
+      "step": 12339
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.585189174599052e-11,
+      "loss": 1.0746,
+      "step": 12340
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.13452142417259e-11,
+      "loss": 1.0834,
+      "step": 12341
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.450668455418082e-11,
+      "loss": 0.8806,
+      "step": 12342
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.533630473726788e-11,
+      "loss": 1.0409,
+      "step": 12343
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.834076256481467e-12,
+      "loss": 1.1888,
+      "step": 12344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.0238,
+      "step": 12345
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 12345,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 1.5693072851533824e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}