diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,10756 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 1784,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 4.8979,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 5.0308,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 5.1394,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-08,
+      "loss": 4.684,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-08,
+      "loss": 4.7728,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 4.8141,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-08,
+      "loss": 4.6878,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 4.8455,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 5.0892,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4e-07,
+      "loss": 4.7674,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6e-07,
+      "loss": 4.8139,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-07,
+      "loss": 4.7889,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 5.0036,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2e-07,
+      "loss": 4.621,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 4.7464,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6e-07,
+      "loss": 4.8966,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8e-07,
+      "loss": 4.9026,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 4.8133,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2e-07,
+      "loss": 4.7758,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 5.0714,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-07,
+      "loss": 4.7965,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8e-07,
+      "loss": 5.1218,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 4.9747,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 4.7182,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4e-07,
+      "loss": 4.9117,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 5.0119,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 4.849,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 4.5669,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2e-07,
+      "loss": 4.9379,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4e-07,
+      "loss": 4.7604,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6e-07,
+      "loss": 4.6528,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 5.042,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 4.7666,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 4.6258,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.4e-07,
+      "loss": 4.8797,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.6e-07,
+      "loss": 5.1419,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 4.7931,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 4.8808,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.2e-07,
+      "loss": 5.0122,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.4e-07,
+      "loss": 4.9645,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.6e-07,
+      "loss": 4.8041,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.8e-07,
+      "loss": 4.6033,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 4.9098,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 4.7344,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 5.0339,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.6e-07,
+      "loss": 5.0145,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.8e-07,
+      "loss": 5.2033,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 5.6315,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 5.6405,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 5.7597,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 4.5845,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 4.7264,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 4.7007,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.02e-06,
+      "loss": 4.5986,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.04e-06,
+      "loss": 4.6163,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.06e-06,
+      "loss": 4.7264,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.08e-06,
+      "loss": 4.6375,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1e-06,
+      "loss": 4.5965,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.12e-06,
+      "loss": 4.5969,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.14e-06,
+      "loss": 4.5947,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 4.5887,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 4.5323,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 4.9012,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 4.3647,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 4.6395,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.26e-06,
+      "loss": 4.676,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.28e-06,
+      "loss": 4.7523,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3e-06,
+      "loss": 4.7299,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.32e-06,
+      "loss": 4.5901,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.34e-06,
+      "loss": 4.5266,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 4.6522,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 4.4009,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 4.6197,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.42e-06,
+      "loss": 4.3639,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.44e-06,
+      "loss": 4.9628,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.46e-06,
+      "loss": 4.7202,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.48e-06,
+      "loss": 4.5287,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5e-06,
+      "loss": 4.4449,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.52e-06,
+      "loss": 4.4654,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.54e-06,
+      "loss": 4.78,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.56e-06,
+      "loss": 4.4214,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 4.516,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 4.48,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 4.4536,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 4.3779,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 4.4318,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 4.7914,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 4.4019,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.72e-06,
+      "loss": 4.7843,
+      "step": 89
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.74e-06,
+      "loss": 4.278,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.76e-06,
+      "loss": 4.7026,
+      "step": 91
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 4.4068,
+      "step": 92
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 4.2567,
+      "step": 93
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 4.7279,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 4.9394,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8600000000000002e-06,
+      "loss": 4.5352,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 4.6746,
+      "step": 97
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 4.7662,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 5.1079,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 5.029,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.94e-06,
+      "loss": 4.496,
+      "step": 101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 4.2873,
+      "step": 102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98e-06,
+      "loss": 4.4015,
+      "step": 103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 4.3324,
+      "step": 104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.02e-06,
+      "loss": 4.4186,
+      "step": 105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.04e-06,
+      "loss": 4.3571,
+      "step": 106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.06e-06,
+      "loss": 4.3992,
+      "step": 107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.08e-06,
+      "loss": 4.4966,
+      "step": 108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 4.3935,
+      "step": 109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.12e-06,
+      "loss": 4.4388,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1400000000000003e-06,
+      "loss": 4.3539,
+      "step": 111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.16e-06,
+      "loss": 4.2627,
+      "step": 112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 4.4298,
+      "step": 113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2e-06,
+      "loss": 4.1223,
+      "step": 114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 4.168,
+      "step": 115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.24e-06,
+      "loss": 4.3208,
+      "step": 116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2600000000000004e-06,
+      "loss": 4.4926,
+      "step": 117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.28e-06,
+      "loss": 4.3425,
+      "step": 118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 4.4036,
+      "step": 119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 4.4742,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 4.5283,
+      "step": 121
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 4.3519,
+      "step": 122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.38e-06,
+      "loss": 4.5656,
+      "step": 123
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 4.4429,
+      "step": 124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.42e-06,
+      "loss": 4.1031,
+      "step": 125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 4.7698,
+      "step": 126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.46e-06,
+      "loss": 4.402,
+      "step": 127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 4.3798,
+      "step": 128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5e-06,
+      "loss": 4.2812,
+      "step": 129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.52e-06,
+      "loss": 4.3277,
+      "step": 130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5400000000000002e-06,
+      "loss": 4.6332,
+      "step": 131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.56e-06,
+      "loss": 4.3139,
+      "step": 132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 4.4162,
+      "step": 133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6e-06,
+      "loss": 4.3782,
+      "step": 134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 4.3454,
+      "step": 135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.64e-06,
+      "loss": 4.466,
+      "step": 136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6600000000000004e-06,
+      "loss": 4.2361,
+      "step": 137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.68e-06,
+      "loss": 4.4187,
+      "step": 138
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 4.5573,
+      "step": 139
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 4.2393,
+      "step": 140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7400000000000004e-06,
+      "loss": 4.4812,
+      "step": 141
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 4.7115,
+      "step": 142
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 4.3691,
+      "step": 143
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 4.5167,
+      "step": 144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.82e-06,
+      "loss": 4.7135,
+      "step": 145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.84e-06,
+      "loss": 4.5908,
+      "step": 146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.86e-06,
+      "loss": 4.5714,
+      "step": 147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.88e-06,
+      "loss": 4.7001,
+      "step": 148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9e-06,
+      "loss": 4.8567,
+      "step": 149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.92e-06,
+      "loss": 4.4613,
+      "step": 150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 4.087,
+      "step": 151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.96e-06,
+      "loss": 4.4216,
+      "step": 152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 4.0582,
+      "step": 153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3e-06,
+      "loss": 4.5296,
+      "step": 154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 4.371,
+      "step": 155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.04e-06,
+      "loss": 4.5969,
+      "step": 156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0600000000000003e-06,
+      "loss": 4.3693,
+      "step": 157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.08e-06,
+      "loss": 4.5456,
+      "step": 158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 4.3506,
+      "step": 159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.12e-06,
+      "loss": 4.3094,
+      "step": 160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1400000000000004e-06,
+      "loss": 4.3081,
+      "step": 161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 4.4913,
+      "step": 162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 4.3098,
+      "step": 163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 4.5428,
+      "step": 164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2200000000000005e-06,
+      "loss": 4.3314,
+      "step": 165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 4.5463,
+      "step": 166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2600000000000006e-06,
+      "loss": 4.3964,
+      "step": 167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 4.088,
+      "step": 168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 4.178,
+      "step": 169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 4.344,
+      "step": 170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 4.4334,
+      "step": 171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 4.3306,
+      "step": 172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 4.3677,
+      "step": 173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 4.2183,
+      "step": 174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 4.4863,
+      "step": 175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.44e-06,
+      "loss": 4.2,
+      "step": 176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.46e-06,
+      "loss": 4.2847,
+      "step": 177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.48e-06,
+      "loss": 4.2671,
+      "step": 178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.5e-06,
+      "loss": 4.0324,
+      "step": 179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.52e-06,
+      "loss": 4.4798,
+      "step": 180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.54e-06,
+      "loss": 4.3183,
+      "step": 181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 4.2923,
+      "step": 182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.58e-06,
+      "loss": 4.3861,
+      "step": 183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 4.4197,
+      "step": 184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.62e-06,
+      "loss": 4.4057,
+      "step": 185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 4.2489,
+      "step": 186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.66e-06,
+      "loss": 4.2751,
+      "step": 187
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 4.5355,
+      "step": 188
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7e-06,
+      "loss": 4.2379,
+      "step": 189
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 4.2707,
+      "step": 190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.74e-06,
+      "loss": 4.4653,
+      "step": 191
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 4.246,
+      "step": 192
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 4.1605,
+      "step": 193
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 4.3331,
+      "step": 194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 4.5936,
+      "step": 195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 4.4333,
+      "step": 196
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.86e-06,
+      "loss": 4.3837,
+      "step": 197
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.88e-06,
+      "loss": 4.4062,
+      "step": 198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 4.5159,
+      "step": 199
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 4.4136,
+      "step": 200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.94e-06,
+      "loss": 4.3524,
+      "step": 201
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.96e-06,
+      "loss": 4.2316,
+      "step": 202
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 4.4342,
+      "step": 203
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 4.1104,
+      "step": 204
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 4.2423,
+      "step": 205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.04e-06,
+      "loss": 4.3397,
+      "step": 206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.060000000000001e-06,
+      "loss": 4.4085,
+      "step": 207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.08e-06,
+      "loss": 4.0921,
+      "step": 208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.1e-06,
+      "loss": 4.1782,
+      "step": 209
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.12e-06,
+      "loss": 4.3254,
+      "step": 210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.14e-06,
+      "loss": 4.3316,
+      "step": 211
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.16e-06,
+      "loss": 4.2394,
+      "step": 212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.18e-06,
+      "loss": 4.2667,
+      "step": 213
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 4.0512,
+      "step": 214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.22e-06,
+      "loss": 4.4445,
+      "step": 215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.24e-06,
+      "loss": 3.9252,
+      "step": 216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.26e-06,
+      "loss": 4.1777,
+      "step": 217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 4.2564,
+      "step": 218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3e-06,
+      "loss": 4.231,
+      "step": 219
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.32e-06,
+      "loss": 4.4956,
+      "step": 220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.34e-06,
+      "loss": 4.3815,
+      "step": 221
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 4.3849,
+      "step": 222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.38e-06,
+      "loss": 4.167,
+      "step": 223
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4e-06,
+      "loss": 4.351,
+      "step": 224
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.42e-06,
+      "loss": 4.3551,
+      "step": 225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 4.4939,
+      "step": 226
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 4.4145,
+      "step": 227
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.48e-06,
+      "loss": 4.0854,
+      "step": 228
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.5e-06,
+      "loss": 4.2662,
+      "step": 229
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 4.2702,
+      "step": 230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 4.28,
+      "step": 231
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.56e-06,
+      "loss": 4.2866,
+      "step": 232
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.58e-06,
+      "loss": 4.447,
+      "step": 233
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 4.3667,
+      "step": 234
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 4.1456,
+      "step": 235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 4.6387,
+      "step": 236
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.66e-06,
+      "loss": 4.3739,
+      "step": 237
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 4.3531,
+      "step": 238
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7e-06,
+      "loss": 4.3174,
+      "step": 239
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 4.2655,
+      "step": 240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.74e-06,
+      "loss": 4.5726,
+      "step": 241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.76e-06,
+      "loss": 4.2863,
+      "step": 242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.78e-06,
+      "loss": 4.3672,
+      "step": 243
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 4.5211,
+      "step": 244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 4.6593,
+      "step": 245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.84e-06,
+      "loss": 4.5861,
+      "step": 246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.86e-06,
+      "loss": 4.4221,
+      "step": 247
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 4.4456,
+      "step": 248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 4.179,
+      "step": 249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.92e-06,
+      "loss": 4.2583,
+      "step": 250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.94e-06,
+      "loss": 4.2367,
+      "step": 251
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 4.149,
+      "step": 252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 4.2105,
+      "step": 253
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5e-06,
+      "loss": 4.3214,
+      "step": 254
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.02e-06,
+      "loss": 4.2818,
+      "step": 255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.04e-06,
+      "loss": 4.2106,
+      "step": 256
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.060000000000001e-06,
+      "loss": 4.2392,
+      "step": 257
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 4.3545,
+      "step": 258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.1e-06,
+      "loss": 4.4534,
+      "step": 259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.12e-06,
+      "loss": 3.9962,
+      "step": 260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.140000000000001e-06,
+      "loss": 4.1125,
+      "step": 261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.1600000000000006e-06,
+      "loss": 4.3527,
+      "step": 262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.18e-06,
+      "loss": 4.2578,
+      "step": 263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.2e-06,
+      "loss": 4.1116,
+      "step": 264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.220000000000001e-06,
+      "loss": 4.19,
+      "step": 265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 4.0246,
+      "step": 266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.2600000000000005e-06,
+      "loss": 4.3317,
+      "step": 267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.28e-06,
+      "loss": 4.2264,
+      "step": 268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 4.4712,
+      "step": 269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.320000000000001e-06,
+      "loss": 4.0864,
+      "step": 270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.3400000000000005e-06,
+      "loss": 4.3615,
+      "step": 271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.36e-06,
+      "loss": 4.1232,
+      "step": 272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 4.0899,
+      "step": 273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 4.3393,
+      "step": 274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.420000000000001e-06,
+      "loss": 4.3412,
+      "step": 275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 4.2632,
+      "step": 276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.460000000000001e-06,
+      "loss": 4.3143,
+      "step": 277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 4.3781,
+      "step": 278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 4.2352,
+      "step": 279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 4.2026,
+      "step": 280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.540000000000001e-06,
+      "loss": 4.312,
+      "step": 281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.560000000000001e-06,
+      "loss": 4.3772,
+      "step": 282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 4.3619,
+      "step": 283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 4.1302,
+      "step": 284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.620000000000001e-06,
+      "loss": 4.2813,
+      "step": 285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.64e-06,
+      "loss": 4.0457,
+      "step": 286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.66e-06,
+      "loss": 4.271,
+      "step": 287
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.68e-06,
+      "loss": 4.5288,
+      "step": 288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.7e-06,
+      "loss": 4.2303,
+      "step": 289
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.72e-06,
+      "loss": 4.3454,
+      "step": 290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.74e-06,
+      "loss": 4.2488,
+      "step": 291
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.76e-06,
+      "loss": 4.6298,
+      "step": 292
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.78e-06,
+      "loss": 4.432,
+      "step": 293
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.8e-06,
+      "loss": 4.5399,
+      "step": 294
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.82e-06,
+      "loss": 4.4457,
+      "step": 295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.84e-06,
+      "loss": 4.3277,
+      "step": 296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.86e-06,
+      "loss": 4.3389,
+      "step": 297
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 4.5077,
+      "step": 298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9e-06,
+      "loss": 4.285,
+      "step": 299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.92e-06,
+      "loss": 4.5822,
+      "step": 300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.94e-06,
+      "loss": 4.2964,
+      "step": 301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9600000000000005e-06,
+      "loss": 4.1312,
+      "step": 302
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.98e-06,
+      "loss": 4.3298,
+      "step": 303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6e-06,
+      "loss": 4.1611,
+      "step": 304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.02e-06,
+      "loss": 4.3209,
+      "step": 305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.040000000000001e-06,
+      "loss": 4.2239,
+      "step": 306
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 4.4311,
+      "step": 307
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.08e-06,
+      "loss": 4.4931,
+      "step": 308
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.1e-06,
+      "loss": 4.4734,
+      "step": 309
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.120000000000001e-06,
+      "loss": 4.2552,
+      "step": 310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.1400000000000005e-06,
+      "loss": 4.2722,
+      "step": 311
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.16e-06,
+      "loss": 4.2364,
+      "step": 312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.18e-06,
+      "loss": 4.2464,
+      "step": 313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 4.3668,
+      "step": 314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.220000000000001e-06,
+      "loss": 4.4123,
+      "step": 315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.24e-06,
+      "loss": 4.1796,
+      "step": 316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.26e-06,
+      "loss": 4.2291,
+      "step": 317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 3.9427,
+      "step": 318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 4.134,
+      "step": 319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 4.2965,
+      "step": 320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.34e-06,
+      "loss": 4.5014,
+      "step": 321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.360000000000001e-06,
+      "loss": 4.2937,
+      "step": 322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 4.3505,
+      "step": 323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 4.2249,
+      "step": 324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.42e-06,
+      "loss": 4.2019,
+      "step": 325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.440000000000001e-06,
+      "loss": 4.1783,
+      "step": 326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.460000000000001e-06,
+      "loss": 4.0979,
+      "step": 327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 4.2332,
+      "step": 328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 4.1688,
+      "step": 329
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.520000000000001e-06,
+      "loss": 4.4648,
+      "step": 330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.540000000000001e-06,
+      "loss": 4.2999,
+      "step": 331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 4.3966,
+      "step": 332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 4.3984,
+      "step": 333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 4.3907,
+      "step": 334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.620000000000001e-06,
+      "loss": 4.3986,
+      "step": 335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 4.1703,
+      "step": 336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 4.2845,
+      "step": 337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 4.1571,
+      "step": 338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 4.2387,
+      "step": 339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 4.2163,
+      "step": 340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 4.2032,
+      "step": 341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.760000000000001e-06,
+      "loss": 4.4022,
+      "step": 342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 4.4713,
+      "step": 343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 4.0314,
+      "step": 344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.820000000000001e-06,
+      "loss": 4.2946,
+      "step": 345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.8400000000000014e-06,
+      "loss": 4.2412,
+      "step": 346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.860000000000001e-06,
+      "loss": 4.7193,
+      "step": 347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.88e-06,
+      "loss": 4.5665,
+      "step": 348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.9e-06,
+      "loss": 4.2407,
+      "step": 349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.92e-06,
+      "loss": 4.2379,
+      "step": 350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 4.3604,
+      "step": 351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.96e-06,
+      "loss": 4.412,
+      "step": 352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.98e-06,
+      "loss": 4.0509,
+      "step": 353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7e-06,
+      "loss": 4.222,
+      "step": 354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 3.8822,
+      "step": 355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.04e-06,
+      "loss": 4.1164,
+      "step": 356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.06e-06,
+      "loss": 4.3387,
+      "step": 357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.08e-06,
+      "loss": 4.3317,
+      "step": 358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 4.4643,
+      "step": 359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 4.2113,
+      "step": 360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.14e-06,
+      "loss": 4.2384,
+      "step": 361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.16e-06,
+      "loss": 4.0246,
+      "step": 362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 4.2922,
+      "step": 363
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 4.319,
+      "step": 364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.22e-06,
+      "loss": 4.2642,
+      "step": 365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.24e-06,
+      "loss": 4.3832,
+      "step": 366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.260000000000001e-06,
+      "loss": 4.2273,
+      "step": 367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 4.1541,
+      "step": 368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.3e-06,
+      "loss": 4.0488,
+      "step": 369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.32e-06,
+      "loss": 4.1406,
+      "step": 370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.340000000000001e-06,
+      "loss": 4.2007,
+      "step": 371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 4.5768,
+      "step": 372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 4.2027,
+      "step": 373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.4e-06,
+      "loss": 4.0602,
+      "step": 374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.420000000000001e-06,
+      "loss": 4.0183,
+      "step": 375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.440000000000001e-06,
+      "loss": 4.1212,
+      "step": 376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.4600000000000006e-06,
+      "loss": 4.1234,
+      "step": 377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.48e-06,
+      "loss": 4.0016,
+      "step": 378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 4.183,
+      "step": 379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 4.3496,
+      "step": 380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.540000000000001e-06,
+      "loss": 3.9576,
+      "step": 381
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 4.1224,
+      "step": 382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.58e-06,
+      "loss": 4.2602,
+      "step": 383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 4.0886,
+      "step": 384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.620000000000001e-06,
+      "loss": 4.3628,
+      "step": 385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.640000000000001e-06,
+      "loss": 4.3552,
+      "step": 386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.660000000000001e-06,
+      "loss": 4.3068,
+      "step": 387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 4.0677,
+      "step": 388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.7e-06,
+      "loss": 4.385,
+      "step": 389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.72e-06,
+      "loss": 4.591,
+      "step": 390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.74e-06,
+      "loss": 4.1402,
+      "step": 391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.76e-06,
+      "loss": 4.5987,
+      "step": 392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.78e-06,
+      "loss": 4.6117,
+      "step": 393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 4.2869,
+      "step": 394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.820000000000001e-06,
+      "loss": 4.4133,
+      "step": 395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 4.3002,
+      "step": 396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 4.3726,
+      "step": 397
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.88e-06,
+      "loss": 4.7235,
+      "step": 398
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.9e-06,
+      "loss": 5.0038,
+      "step": 399
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.92e-06,
+      "loss": 4.9442,
+      "step": 400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.94e-06,
+      "loss": 4.3973,
+      "step": 401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.960000000000002e-06,
+      "loss": 4.0756,
+      "step": 402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 4.1005,
+      "step": 403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 4.2536,
+      "step": 404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.020000000000001e-06,
+      "loss": 4.0213,
+      "step": 405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 4.2101,
+      "step": 406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.06e-06,
+      "loss": 4.1213,
+      "step": 407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.08e-06,
+      "loss": 4.2718,
+      "step": 408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.1e-06,
+      "loss": 4.3338,
+      "step": 409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.120000000000002e-06,
+      "loss": 4.2651,
+      "step": 410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.14e-06,
+      "loss": 4.2211,
+      "step": 411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.16e-06,
+      "loss": 4.2575,
+      "step": 412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.18e-06,
+      "loss": 4.2992,
+      "step": 413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.2e-06,
+      "loss": 4.258,
+      "step": 414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 4.2931,
+      "step": 415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.24e-06,
+      "loss": 4.1584,
+      "step": 416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.26e-06,
+      "loss": 4.2493,
+      "step": 417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.28e-06,
+      "loss": 4.2601,
+      "step": 418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.3e-06,
+      "loss": 4.2966,
+      "step": 419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.32e-06,
+      "loss": 4.2481,
+      "step": 420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.34e-06,
+      "loss": 4.1389,
+      "step": 421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.36e-06,
+      "loss": 4.4645,
+      "step": 422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 4.0371,
+      "step": 423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 4.5059,
+      "step": 424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.42e-06,
+      "loss": 4.4168,
+      "step": 425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.44e-06,
+      "loss": 4.275,
+      "step": 426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.46e-06,
+      "loss": 4.06,
+      "step": 427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.48e-06,
+      "loss": 4.0722,
+      "step": 428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.5e-06,
+      "loss": 4.2012,
+      "step": 429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.52e-06,
+      "loss": 4.2943,
+      "step": 430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.540000000000001e-06,
+      "loss": 4.1564,
+      "step": 431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.560000000000001e-06,
+      "loss": 4.3186,
+      "step": 432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 4.2468,
+      "step": 433
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.6e-06,
+      "loss": 4.4037,
+      "step": 434
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.62e-06,
+      "loss": 4.3876,
+      "step": 435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.64e-06,
+      "loss": 4.0163,
+      "step": 436
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.66e-06,
+      "loss": 4.2362,
+      "step": 437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.68e-06,
+      "loss": 4.3478,
+      "step": 438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 4.1701,
+      "step": 439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 4.4183,
+      "step": 440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.740000000000001e-06,
+      "loss": 4.3777,
+      "step": 441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.76e-06,
+      "loss": 4.3285,
+      "step": 442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.78e-06,
+      "loss": 4.3068,
+      "step": 443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.8e-06,
+      "loss": 4.5496,
+      "step": 444
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.82e-06,
+      "loss": 4.3222,
+      "step": 445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.84e-06,
+      "loss": 4.9054,
+      "step": 446
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.860000000000002e-06,
+      "loss": 4.5172,
+      "step": 447
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 5.0985,
+      "step": 448
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 4.4318,
+      "step": 449
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.920000000000001e-06,
+      "loss": 4.4205,
+      "step": 450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.94e-06,
+      "loss": 3.9559,
+      "step": 451
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.96e-06,
+      "loss": 4.0948,
+      "step": 452
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.98e-06,
+      "loss": 4.2409,
+      "step": 453
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9e-06,
+      "loss": 3.9455,
+      "step": 454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.020000000000002e-06,
+      "loss": 3.8765,
+      "step": 455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.040000000000002e-06,
+      "loss": 4.164,
+      "step": 456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.060000000000001e-06,
+      "loss": 4.0202,
+      "step": 457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 3.9943,
+      "step": 458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 3.8719,
+      "step": 459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.12e-06,
+      "loss": 4.161,
+      "step": 460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.14e-06,
+      "loss": 4.0348,
+      "step": 461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.16e-06,
+      "loss": 4.0563,
+      "step": 462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 4.1192,
+      "step": 463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 4.1267,
+      "step": 464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.220000000000002e-06,
+      "loss": 4.0574,
+      "step": 465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.240000000000001e-06,
+      "loss": 4.2841,
+      "step": 466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.260000000000001e-06,
+      "loss": 4.1138,
+      "step": 467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 4.2195,
+      "step": 468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.3e-06,
+      "loss": 4.0567,
+      "step": 469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.32e-06,
+      "loss": 4.1701,
+      "step": 470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.340000000000002e-06,
+      "loss": 4.3347,
+      "step": 471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.360000000000002e-06,
+      "loss": 4.1338,
+      "step": 472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.38e-06,
+      "loss": 4.1579,
+      "step": 473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.4e-06,
+      "loss": 4.2513,
+      "step": 474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.42e-06,
+      "loss": 4.2336,
+      "step": 475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 4.3623,
+      "step": 476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.460000000000001e-06,
+      "loss": 4.3537,
+      "step": 477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.48e-06,
+      "loss": 4.2287,
+      "step": 478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.5e-06,
+      "loss": 4.2965,
+      "step": 479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.52e-06,
+      "loss": 4.1659,
+      "step": 480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.54e-06,
+      "loss": 4.017,
+      "step": 481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.56e-06,
+      "loss": 4.1384,
+      "step": 482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.58e-06,
+      "loss": 4.4777,
+      "step": 483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 4.0477,
+      "step": 484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.620000000000001e-06,
+      "loss": 4.3787,
+      "step": 485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.640000000000001e-06,
+      "loss": 4.2479,
+      "step": 486
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.66e-06,
+      "loss": 4.3696,
+      "step": 487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.68e-06,
+      "loss": 4.5017,
+      "step": 488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.7e-06,
+      "loss": 4.4585,
+      "step": 489
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.72e-06,
+      "loss": 4.2926,
+      "step": 490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.74e-06,
+      "loss": 4.1693,
+      "step": 491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 4.4713,
+      "step": 492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 4.3967,
+      "step": 493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 4.4192,
+      "step": 494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.820000000000001e-06,
+      "loss": 4.4366,
+      "step": 495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.84e-06,
+      "loss": 4.3373,
+      "step": 496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.86e-06,
+      "loss": 4.2001,
+      "step": 497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.88e-06,
+      "loss": 4.6219,
+      "step": 498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.9e-06,
+      "loss": 4.7205,
+      "step": 499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 4.1576,
+      "step": 500
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 4.256221294403076,
+      "eval_runtime": 979.265,
+      "eval_samples_per_second": 2.698,
+      "eval_steps_per_second": 0.338,
+      "eval_wer": 1.9558111860372869,
+      "step": 500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.940000000000001e-06,
+      "loss": 4.2677,
+      "step": 501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 4.0057,
+      "step": 502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 4.0918,
+      "step": 503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1e-05,
+      "loss": 4.2983,
+      "step": 504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.992211838006231e-06,
+      "loss": 3.9936,
+      "step": 505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.984423676012462e-06,
+      "loss": 4.314,
+      "step": 506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.976635514018693e-06,
+      "loss": 4.1491,
+      "step": 507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.968847352024923e-06,
+      "loss": 4.1031,
+      "step": 508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.961059190031154e-06,
+      "loss": 4.2648,
+      "step": 509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.953271028037384e-06,
+      "loss": 4.3805,
+      "step": 510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.945482866043615e-06,
+      "loss": 4.2409,
+      "step": 511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.937694704049845e-06,
+      "loss": 4.0743,
+      "step": 512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.929906542056076e-06,
+      "loss": 4.079,
+      "step": 513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.922118380062306e-06,
+      "loss": 4.0243,
+      "step": 514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.914330218068537e-06,
+      "loss": 4.1766,
+      "step": 515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.906542056074768e-06,
+      "loss": 4.3067,
+      "step": 516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.898753894080998e-06,
+      "loss": 4.2048,
+      "step": 517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.890965732087229e-06,
+      "loss": 4.5795,
+      "step": 518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.883177570093458e-06,
+      "loss": 4.318,
+      "step": 519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.87538940809969e-06,
+      "loss": 3.9178,
+      "step": 520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.86760124610592e-06,
+      "loss": 3.9701,
+      "step": 521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.859813084112151e-06,
+      "loss": 4.054,
+      "step": 522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.85202492211838e-06,
+      "loss": 4.024,
+      "step": 523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.844236760124612e-06,
+      "loss": 4.2859,
+      "step": 524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.836448598130843e-06,
+      "loss": 4.3807,
+      "step": 525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.828660436137073e-06,
+      "loss": 4.2869,
+      "step": 526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.820872274143302e-06,
+      "loss": 4.3896,
+      "step": 527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.813084112149533e-06,
+      "loss": 3.9939,
+      "step": 528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.805295950155765e-06,
+      "loss": 4.3391,
+      "step": 529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.797507788161996e-06,
+      "loss": 4.2874,
+      "step": 530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.789719626168224e-06,
+      "loss": 4.1203,
+      "step": 531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.781931464174455e-06,
+      "loss": 3.9538,
+      "step": 532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.774143302180686e-06,
+      "loss": 4.0216,
+      "step": 533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.766355140186918e-06,
+      "loss": 3.9364,
+      "step": 534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.758566978193147e-06,
+      "loss": 4.4056,
+      "step": 535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.750778816199377e-06,
+      "loss": 4.2598,
+      "step": 536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.742990654205608e-06,
+      "loss": 4.1149,
+      "step": 537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.73520249221184e-06,
+      "loss": 4.3797,
+      "step": 538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.727414330218069e-06,
+      "loss": 4.2991,
+      "step": 539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.7196261682243e-06,
+      "loss": 4.3411,
+      "step": 540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.71183800623053e-06,
+      "loss": 4.368,
+      "step": 541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.70404984423676e-06,
+      "loss": 4.3842,
+      "step": 542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.696261682242991e-06,
+      "loss": 4.1804,
+      "step": 543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.688473520249222e-06,
+      "loss": 4.3239,
+      "step": 544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.680685358255452e-06,
+      "loss": 4.1238,
+      "step": 545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.672897196261683e-06,
+      "loss": 4.2495,
+      "step": 546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.665109034267914e-06,
+      "loss": 4.6969,
+      "step": 547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.657320872274144e-06,
+      "loss": 4.4121,
+      "step": 548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.649532710280375e-06,
+      "loss": 4.8561,
+      "step": 549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.641744548286605e-06,
+      "loss": 4.7053,
+      "step": 550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.633956386292836e-06,
+      "loss": 4.4564,
+      "step": 551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.626168224299066e-06,
+      "loss": 4.3437,
+      "step": 552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.618380062305297e-06,
+      "loss": 4.1783,
+      "step": 553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.610591900311527e-06,
+      "loss": 4.1788,
+      "step": 554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.602803738317758e-06,
+      "loss": 4.0739,
+      "step": 555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.595015576323989e-06,
+      "loss": 4.1816,
+      "step": 556
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.58722741433022e-06,
+      "loss": 4.2924,
+      "step": 557
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.57943925233645e-06,
+      "loss": 4.127,
+      "step": 558
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.57165109034268e-06,
+      "loss": 4.012,
+      "step": 559
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.563862928348911e-06,
+      "loss": 4.0712,
+      "step": 560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.556074766355141e-06,
+      "loss": 4.339,
+      "step": 561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.548286604361372e-06,
+      "loss": 4.0574,
+      "step": 562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.540498442367601e-06,
+      "loss": 3.9701,
+      "step": 563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.532710280373833e-06,
+      "loss": 4.4059,
+      "step": 564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.524922118380064e-06,
+      "loss": 4.3194,
+      "step": 565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.517133956386294e-06,
+      "loss": 4.1761,
+      "step": 566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.509345794392523e-06,
+      "loss": 4.0658,
+      "step": 567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.501557632398755e-06,
+      "loss": 4.2957,
+      "step": 568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.493769470404986e-06,
+      "loss": 4.2191,
+      "step": 569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.485981308411217e-06,
+      "loss": 4.4104,
+      "step": 570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.478193146417445e-06,
+      "loss": 4.1213,
+      "step": 571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.470404984423676e-06,
+      "loss": 4.3653,
+      "step": 572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.462616822429908e-06,
+      "loss": 4.0531,
+      "step": 573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.454828660436139e-06,
+      "loss": 4.0227,
+      "step": 574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.447040498442368e-06,
+      "loss": 4.0987,
+      "step": 575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.439252336448598e-06,
+      "loss": 4.1056,
+      "step": 576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.431464174454829e-06,
+      "loss": 4.187,
+      "step": 577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.423676012461061e-06,
+      "loss": 4.0711,
+      "step": 578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.41588785046729e-06,
+      "loss": 4.1468,
+      "step": 579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.40809968847352e-06,
+      "loss": 4.4059,
+      "step": 580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.400311526479751e-06,
+      "loss": 4.2542,
+      "step": 581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.392523364485983e-06,
+      "loss": 4.3021,
+      "step": 582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.384735202492212e-06,
+      "loss": 3.9786,
+      "step": 583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.376947040498443e-06,
+      "loss": 4.4636,
+      "step": 584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.369158878504673e-06,
+      "loss": 4.2705,
+      "step": 585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.361370716510904e-06,
+      "loss": 4.1328,
+      "step": 586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.353582554517135e-06,
+      "loss": 4.3207,
+      "step": 587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.345794392523365e-06,
+      "loss": 4.3203,
+      "step": 588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.338006230529596e-06,
+      "loss": 4.3819,
+      "step": 589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.330218068535826e-06,
+      "loss": 4.3436,
+      "step": 590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.322429906542057e-06,
+      "loss": 4.1969,
+      "step": 591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.314641744548287e-06,
+      "loss": 4.4116,
+      "step": 592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.306853582554518e-06,
+      "loss": 4.1355,
+      "step": 593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.299065420560748e-06,
+      "loss": 4.4157,
+      "step": 594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.291277258566979e-06,
+      "loss": 4.3975,
+      "step": 595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.28348909657321e-06,
+      "loss": 4.4729,
+      "step": 596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.27570093457944e-06,
+      "loss": 4.488,
+      "step": 597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.26791277258567e-06,
+      "loss": 4.2787,
+      "step": 598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.260124610591901e-06,
+      "loss": 4.7134,
+      "step": 599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.252336448598132e-06,
+      "loss": 4.8995,
+      "step": 600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.244548286604362e-06,
+      "loss": 3.9255,
+      "step": 601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.236760124610593e-06,
+      "loss": 4.2446,
+      "step": 602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.228971962616824e-06,
+      "loss": 4.1358,
+      "step": 603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.221183800623054e-06,
+      "loss": 4.2149,
+      "step": 604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.213395638629285e-06,
+      "loss": 4.0799,
+      "step": 605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.205607476635515e-06,
+      "loss": 4.0453,
+      "step": 606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.197819314641744e-06,
+      "loss": 4.1612,
+      "step": 607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.190031152647976e-06,
+      "loss": 4.1618,
+      "step": 608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.182242990654207e-06,
+      "loss": 4.306,
+      "step": 609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.174454828660438e-06,
+      "loss": 4.0926,
+      "step": 610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 4.3992,
+      "step": 611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.158878504672899e-06,
+      "loss": 4.1634,
+      "step": 612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.15109034267913e-06,
+      "loss": 4.2863,
+      "step": 613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.14330218068536e-06,
+      "loss": 4.1789,
+      "step": 614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.135514018691589e-06,
+      "loss": 4.2619,
+      "step": 615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.12772585669782e-06,
+      "loss": 4.071,
+      "step": 616
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.119937694704052e-06,
+      "loss": 4.0913,
+      "step": 617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.112149532710282e-06,
+      "loss": 4.1295,
+      "step": 618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.104361370716511e-06,
+      "loss": 4.0624,
+      "step": 619
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.096573208722742e-06,
+      "loss": 4.1754,
+      "step": 620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.088785046728972e-06,
+      "loss": 4.2665,
+      "step": 621
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.080996884735204e-06,
+      "loss": 4.1339,
+      "step": 622
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.073208722741433e-06,
+      "loss": 4.2665,
+      "step": 623
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.065420560747664e-06,
+      "loss": 4.4326,
+      "step": 624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.057632398753894e-06,
+      "loss": 4.3469,
+      "step": 625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.049844236760127e-06,
+      "loss": 4.1828,
+      "step": 626
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.042056074766356e-06,
+      "loss": 4.139,
+      "step": 627
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.034267912772586e-06,
+      "loss": 4.4259,
+      "step": 628
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.026479750778817e-06,
+      "loss": 4.474,
+      "step": 629
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.018691588785047e-06,
+      "loss": 4.1752,
+      "step": 630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.010903426791278e-06,
+      "loss": 4.2072,
+      "step": 631
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.003115264797508e-06,
+      "loss": 4.2673,
+      "step": 632
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.995327102803739e-06,
+      "loss": 4.126,
+      "step": 633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.98753894080997e-06,
+      "loss": 4.1688,
+      "step": 634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.9797507788162e-06,
+      "loss": 4.2611,
+      "step": 635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.97196261682243e-06,
+      "loss": 4.0809,
+      "step": 636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.964174454828661e-06,
+      "loss": 4.3,
+      "step": 637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.956386292834892e-06,
+      "loss": 3.949,
+      "step": 638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.948598130841122e-06,
+      "loss": 4.3013,
+      "step": 639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.940809968847353e-06,
+      "loss": 4.5105,
+      "step": 640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.933021806853583e-06,
+      "loss": 4.0846,
+      "step": 641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.925233644859814e-06,
+      "loss": 4.3201,
+      "step": 642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.917445482866045e-06,
+      "loss": 4.2269,
+      "step": 643
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.909657320872275e-06,
+      "loss": 4.3943,
+      "step": 644
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.901869158878506e-06,
+      "loss": 4.2926,
+      "step": 645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.894080996884736e-06,
+      "loss": 4.4902,
+      "step": 646
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.886292834890967e-06,
+      "loss": 4.7154,
+      "step": 647
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.878504672897197e-06,
+      "loss": 4.8186,
+      "step": 648
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.870716510903428e-06,
+      "loss": 4.7584,
+      "step": 649
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.862928348909659e-06,
+      "loss": 4.6648,
+      "step": 650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.855140186915887e-06,
+      "loss": 4.1074,
+      "step": 651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.84735202492212e-06,
+      "loss": 4.0953,
+      "step": 652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.83956386292835e-06,
+      "loss": 4.131,
+      "step": 653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.83177570093458e-06,
+      "loss": 4.1487,
+      "step": 654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.82398753894081e-06,
+      "loss": 4.1152,
+      "step": 655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.81619937694704e-06,
+      "loss": 4.1064,
+      "step": 656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.808411214953273e-06,
+      "loss": 4.131,
+      "step": 657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.800623052959503e-06,
+      "loss": 4.3666,
+      "step": 658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.792834890965732e-06,
+      "loss": 4.3815,
+      "step": 659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.785046728971963e-06,
+      "loss": 4.1963,
+      "step": 660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.777258566978195e-06,
+      "loss": 4.2838,
+      "step": 661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.769470404984425e-06,
+      "loss": 4.2513,
+      "step": 662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.761682242990654e-06,
+      "loss": 4.2834,
+      "step": 663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.753894080996885e-06,
+      "loss": 4.2635,
+      "step": 664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.746105919003115e-06,
+      "loss": 4.0828,
+      "step": 665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.738317757009348e-06,
+      "loss": 4.1673,
+      "step": 666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.730529595015576e-06,
+      "loss": 4.1687,
+      "step": 667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.722741433021807e-06,
+      "loss": 4.0363,
+      "step": 668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.714953271028038e-06,
+      "loss": 4.1119,
+      "step": 669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.707165109034268e-06,
+      "loss": 4.3305,
+      "step": 670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.699376947040499e-06,
+      "loss": 4.3001,
+      "step": 671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.69158878504673e-06,
+      "loss": 4.3189,
+      "step": 672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.68380062305296e-06,
+      "loss": 4.1437,
+      "step": 673
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.67601246105919e-06,
+      "loss": 4.1328,
+      "step": 674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.668224299065421e-06,
+      "loss": 3.9901,
+      "step": 675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.660436137071652e-06,
+      "loss": 4.2261,
+      "step": 676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.652647975077882e-06,
+      "loss": 4.1611,
+      "step": 677
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.644859813084113e-06,
+      "loss": 4.2216,
+      "step": 678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.637071651090343e-06,
+      "loss": 4.0273,
+      "step": 679
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.629283489096574e-06,
+      "loss": 4.1831,
+      "step": 680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.621495327102804e-06,
+      "loss": 4.1493,
+      "step": 681
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.613707165109035e-06,
+      "loss": 4.3,
+      "step": 682
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.605919003115266e-06,
+      "loss": 4.3376,
+      "step": 683
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.598130841121496e-06,
+      "loss": 4.3089,
+      "step": 684
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.590342679127727e-06,
+      "loss": 4.4079,
+      "step": 685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.582554517133957e-06,
+      "loss": 4.2835,
+      "step": 686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.574766355140188e-06,
+      "loss": 3.9759,
+      "step": 687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.566978193146418e-06,
+      "loss": 4.0905,
+      "step": 688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.559190031152649e-06,
+      "loss": 4.2157,
+      "step": 689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.55140186915888e-06,
+      "loss": 4.2714,
+      "step": 690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.54361370716511e-06,
+      "loss": 4.1373,
+      "step": 691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.53582554517134e-06,
+      "loss": 4.3353,
+      "step": 692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.528037383177571e-06,
+      "loss": 4.1958,
+      "step": 693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.520249221183802e-06,
+      "loss": 4.3141,
+      "step": 694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.51246105919003e-06,
+      "loss": 4.4398,
+      "step": 695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.504672897196263e-06,
+      "loss": 4.5299,
+      "step": 696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.496884735202494e-06,
+      "loss": 3.9704,
+      "step": 697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.489096573208724e-06,
+      "loss": 4.3336,
+      "step": 698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.481308411214953e-06,
+      "loss": 4.2485,
+      "step": 699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.473520249221184e-06,
+      "loss": 4.3074,
+      "step": 700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.465732087227416e-06,
+      "loss": 4.0492,
+      "step": 701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.457943925233646e-06,
+      "loss": 4.1611,
+      "step": 702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.450155763239875e-06,
+      "loss": 4.1531,
+      "step": 703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.442367601246106e-06,
+      "loss": 4.3696,
+      "step": 704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.434579439252338e-06,
+      "loss": 4.0249,
+      "step": 705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.426791277258569e-06,
+      "loss": 4.0972,
+      "step": 706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.419003115264797e-06,
+      "loss": 3.9853,
+      "step": 707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.411214953271028e-06,
+      "loss": 4.1267,
+      "step": 708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.403426791277259e-06,
+      "loss": 3.9886,
+      "step": 709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.395638629283491e-06,
+      "loss": 4.2274,
+      "step": 710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.38785046728972e-06,
+      "loss": 4.2635,
+      "step": 711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.38006230529595e-06,
+      "loss": 4.096,
+      "step": 712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.372274143302181e-06,
+      "loss": 4.0992,
+      "step": 713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.364485981308411e-06,
+      "loss": 4.5115,
+      "step": 714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.356697819314642e-06,
+      "loss": 4.1638,
+      "step": 715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.348909657320873e-06,
+      "loss": 3.9764,
+      "step": 716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.341121495327103e-06,
+      "loss": 4.1803,
+      "step": 717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 4.3886,
+      "step": 718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.325545171339564e-06,
+      "loss": 4.0808,
+      "step": 719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.317757009345795e-06,
+      "loss": 4.2577,
+      "step": 720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.309968847352025e-06,
+      "loss": 4.3291,
+      "step": 721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.302180685358256e-06,
+      "loss": 4.2901,
+      "step": 722
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.294392523364487e-06,
+      "loss": 4.1644,
+      "step": 723
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.286604361370717e-06,
+      "loss": 4.4149,
+      "step": 724
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.278816199376948e-06,
+      "loss": 4.2612,
+      "step": 725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.271028037383178e-06,
+      "loss": 4.1666,
+      "step": 726
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.263239875389409e-06,
+      "loss": 3.8837,
+      "step": 727
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.25545171339564e-06,
+      "loss": 4.2903,
+      "step": 728
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.24766355140187e-06,
+      "loss": 4.195,
+      "step": 729
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.2398753894081e-06,
+      "loss": 4.1302,
+      "step": 730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.232087227414331e-06,
+      "loss": 4.2015,
+      "step": 731
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.224299065420562e-06,
+      "loss": 4.2175,
+      "step": 732
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.216510903426792e-06,
+      "loss": 4.2349,
+      "step": 733
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.208722741433023e-06,
+      "loss": 4.2237,
+      "step": 734
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.200934579439253e-06,
+      "loss": 4.2773,
+      "step": 735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.193146417445484e-06,
+      "loss": 4.1539,
+      "step": 736
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.185358255451715e-06,
+      "loss": 4.141,
+      "step": 737
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.177570093457945e-06,
+      "loss": 3.914,
+      "step": 738
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.169781931464174e-06,
+      "loss": 4.1081,
+      "step": 739
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.161993769470406e-06,
+      "loss": 4.1523,
+      "step": 740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.154205607476637e-06,
+      "loss": 4.3682,
+      "step": 741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.146417445482867e-06,
+      "loss": 4.035,
+      "step": 742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.138629283489096e-06,
+      "loss": 4.1855,
+      "step": 743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.130841121495327e-06,
+      "loss": 4.146,
+      "step": 744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.123052959501559e-06,
+      "loss": 4.6021,
+      "step": 745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.11526479750779e-06,
+      "loss": 4.1236,
+      "step": 746
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.107476635514018e-06,
+      "loss": 4.4978,
+      "step": 747
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.099688473520249e-06,
+      "loss": 4.6986,
+      "step": 748
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.091900311526481e-06,
+      "loss": 4.2746,
+      "step": 749
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.084112149532712e-06,
+      "loss": 4.6037,
+      "step": 750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.07632398753894e-06,
+      "loss": 4.2044,
+      "step": 751
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.068535825545171e-06,
+      "loss": 4.2127,
+      "step": 752
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.060747663551402e-06,
+      "loss": 4.2551,
+      "step": 753
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.052959501557634e-06,
+      "loss": 3.9347,
+      "step": 754
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.045171339563863e-06,
+      "loss": 4.0872,
+      "step": 755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.037383177570094e-06,
+      "loss": 4.2794,
+      "step": 756
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.029595015576324e-06,
+      "loss": 3.9968,
+      "step": 757
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.021806853582555e-06,
+      "loss": 4.1948,
+      "step": 758
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.014018691588785e-06,
+      "loss": 4.248,
+      "step": 759
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.006230529595016e-06,
+      "loss": 4.1119,
+      "step": 760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.998442367601246e-06,
+      "loss": 4.4344,
+      "step": 761
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.990654205607477e-06,
+      "loss": 3.9615,
+      "step": 762
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.982866043613708e-06,
+      "loss": 4.2202,
+      "step": 763
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.975077881619938e-06,
+      "loss": 3.9416,
+      "step": 764
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.967289719626169e-06,
+      "loss": 3.93,
+      "step": 765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.9595015576324e-06,
+      "loss": 4.0406,
+      "step": 766
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.95171339563863e-06,
+      "loss": 4.28,
+      "step": 767
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.94392523364486e-06,
+      "loss": 3.9428,
+      "step": 768
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.936137071651091e-06,
+      "loss": 4.4859,
+      "step": 769
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.928348909657322e-06,
+      "loss": 4.2239,
+      "step": 770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.920560747663552e-06,
+      "loss": 4.0879,
+      "step": 771
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.912772585669783e-06,
+      "loss": 4.1591,
+      "step": 772
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.904984423676013e-06,
+      "loss": 4.0768,
+      "step": 773
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.897196261682244e-06,
+      "loss": 4.2007,
+      "step": 774
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.889408099688474e-06,
+      "loss": 4.0522,
+      "step": 775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.881619937694705e-06,
+      "loss": 4.1272,
+      "step": 776
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.873831775700936e-06,
+      "loss": 3.976,
+      "step": 777
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.866043613707166e-06,
+      "loss": 4.3929,
+      "step": 778
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.858255451713395e-06,
+      "loss": 4.2813,
+      "step": 779
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.850467289719627e-06,
+      "loss": 4.3425,
+      "step": 780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.842679127725858e-06,
+      "loss": 4.291,
+      "step": 781
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.834890965732088e-06,
+      "loss": 4.3224,
+      "step": 782
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.827102803738317e-06,
+      "loss": 4.236,
+      "step": 783
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.81931464174455e-06,
+      "loss": 4.1546,
+      "step": 784
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.81152647975078e-06,
+      "loss": 3.9727,
+      "step": 785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.80373831775701e-06,
+      "loss": 4.2866,
+      "step": 786
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.79595015576324e-06,
+      "loss": 3.8694,
+      "step": 787
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.78816199376947e-06,
+      "loss": 4.2966,
+      "step": 788
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.780373831775702e-06,
+      "loss": 4.2564,
+      "step": 789
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.772585669781933e-06,
+      "loss": 4.2773,
+      "step": 790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.764797507788162e-06,
+      "loss": 4.4553,
+      "step": 791
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.757009345794392e-06,
+      "loss": 4.0103,
+      "step": 792
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.749221183800623e-06,
+      "loss": 4.2524,
+      "step": 793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.741433021806855e-06,
+      "loss": 4.1566,
+      "step": 794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.733644859813084e-06,
+      "loss": 4.5188,
+      "step": 795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.725856697819315e-06,
+      "loss": 4.2645,
+      "step": 796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.718068535825545e-06,
+      "loss": 4.0065,
+      "step": 797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.710280373831777e-06,
+      "loss": 4.6659,
+      "step": 798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.702492211838006e-06,
+      "loss": 4.405,
+      "step": 799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.694704049844237e-06,
+      "loss": 4.4123,
+      "step": 800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.686915887850467e-06,
+      "loss": 4.2232,
+      "step": 801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.679127725856698e-06,
+      "loss": 4.1487,
+      "step": 802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.671339563862929e-06,
+      "loss": 4.1774,
+      "step": 803
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.663551401869159e-06,
+      "loss": 4.1412,
+      "step": 804
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.65576323987539e-06,
+      "loss": 4.2344,
+      "step": 805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.64797507788162e-06,
+      "loss": 4.2893,
+      "step": 806
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.64018691588785e-06,
+      "loss": 4.1144,
+      "step": 807
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.632398753894081e-06,
+      "loss": 3.8777,
+      "step": 808
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.624610591900312e-06,
+      "loss": 4.2978,
+      "step": 809
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.616822429906543e-06,
+      "loss": 4.2677,
+      "step": 810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.609034267912772e-06,
+      "loss": 4.2522,
+      "step": 811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.601246105919004e-06,
+      "loss": 4.1791,
+      "step": 812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.593457943925234e-06,
+      "loss": 4.1521,
+      "step": 813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.585669781931465e-06,
+      "loss": 4.016,
+      "step": 814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.5778816199376945e-06,
+      "loss": 4.1516,
+      "step": 815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.570093457943926e-06,
+      "loss": 4.1297,
+      "step": 816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.5623052959501565e-06,
+      "loss": 4.0589,
+      "step": 817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.554517133956387e-06,
+      "loss": 4.2112,
+      "step": 818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.546728971962617e-06,
+      "loss": 4.1802,
+      "step": 819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.538940809968847e-06,
+      "loss": 4.1495,
+      "step": 820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.531152647975079e-06,
+      "loss": 4.3631,
+      "step": 821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.523364485981309e-06,
+      "loss": 4.0962,
+      "step": 822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.515576323987539e-06,
+      "loss": 4.3062,
+      "step": 823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.50778816199377e-06,
+      "loss": 4.2867,
+      "step": 824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 4.1595,
+      "step": 825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.492211838006232e-06,
+      "loss": 4.2508,
+      "step": 826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.484423676012462e-06,
+      "loss": 4.3047,
+      "step": 827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.476635514018692e-06,
+      "loss": 4.0445,
+      "step": 828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.4688473520249225e-06,
+      "loss": 4.0754,
+      "step": 829
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.461059190031154e-06,
+      "loss": 4.0939,
+      "step": 830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.4532710280373844e-06,
+      "loss": 4.5133,
+      "step": 831
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.445482866043614e-06,
+      "loss": 4.4704,
+      "step": 832
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.437694704049845e-06,
+      "loss": 4.2372,
+      "step": 833
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.429906542056075e-06,
+      "loss": 3.9155,
+      "step": 834
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.422118380062307e-06,
+      "loss": 3.9785,
+      "step": 835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.4143302180685364e-06,
+      "loss": 4.1249,
+      "step": 836
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.406542056074767e-06,
+      "loss": 4.3165,
+      "step": 837
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.3987538940809976e-06,
+      "loss": 4.1402,
+      "step": 838
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.390965732087229e-06,
+      "loss": 4.0376,
+      "step": 839
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.383177570093458e-06,
+      "loss": 4.154,
+      "step": 840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.375389408099689e-06,
+      "loss": 4.1849,
+      "step": 841
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.36760124610592e-06,
+      "loss": 4.1869,
+      "step": 842
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.35981308411215e-06,
+      "loss": 4.1591,
+      "step": 843
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.35202492211838e-06,
+      "loss": 4.2733,
+      "step": 844
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.3442367601246115e-06,
+      "loss": 4.032,
+      "step": 845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.336448598130842e-06,
+      "loss": 4.1691,
+      "step": 846
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.328660436137073e-06,
+      "loss": 4.0699,
+      "step": 847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.320872274143302e-06,
+      "loss": 4.4413,
+      "step": 848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.313084112149533e-06,
+      "loss": 4.6893,
+      "step": 849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.305295950155764e-06,
+      "loss": 4.7372,
+      "step": 850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.297507788161995e-06,
+      "loss": 4.2294,
+      "step": 851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.289719626168225e-06,
+      "loss": 4.096,
+      "step": 852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.281931464174455e-06,
+      "loss": 4.165,
+      "step": 853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.274143302180686e-06,
+      "loss": 4.2159,
+      "step": 854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.266355140186917e-06,
+      "loss": 4.2525,
+      "step": 855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.258566978193147e-06,
+      "loss": 4.0841,
+      "step": 856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.2507788161993775e-06,
+      "loss": 4.1443,
+      "step": 857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.242990654205608e-06,
+      "loss": 4.46,
+      "step": 858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.2352024922118395e-06,
+      "loss": 4.15,
+      "step": 859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.227414330218069e-06,
+      "loss": 4.2259,
+      "step": 860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.2196261682243e-06,
+      "loss": 3.8453,
+      "step": 861
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.21183800623053e-06,
+      "loss": 4.0966,
+      "step": 862
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.204049844236761e-06,
+      "loss": 4.1801,
+      "step": 863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.196261682242991e-06,
+      "loss": 4.0659,
+      "step": 864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.188473520249222e-06,
+      "loss": 4.0819,
+      "step": 865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.180685358255453e-06,
+      "loss": 4.3268,
+      "step": 866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.172897196261683e-06,
+      "loss": 4.1791,
+      "step": 867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.165109034267913e-06,
+      "loss": 4.2028,
+      "step": 868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.1573208722741435e-06,
+      "loss": 4.1953,
+      "step": 869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.149532710280375e-06,
+      "loss": 4.1251,
+      "step": 870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.1417445482866054e-06,
+      "loss": 4.1843,
+      "step": 871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.133956386292835e-06,
+      "loss": 4.2022,
+      "step": 872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.126168224299066e-06,
+      "loss": 4.1832,
+      "step": 873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.118380062305297e-06,
+      "loss": 4.1114,
+      "step": 874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.110591900311528e-06,
+      "loss": 4.3403,
+      "step": 875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.1028037383177574e-06,
+      "loss": 4.1978,
+      "step": 876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.095015576323988e-06,
+      "loss": 4.5599,
+      "step": 877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.0872274143302186e-06,
+      "loss": 4.2294,
+      "step": 878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.07943925233645e-06,
+      "loss": 4.1294,
+      "step": 879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.07165109034268e-06,
+      "loss": 4.2876,
+      "step": 880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.06386292834891e-06,
+      "loss": 4.3559,
+      "step": 881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.056074766355141e-06,
+      "loss": 3.9312,
+      "step": 882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.048286604361371e-06,
+      "loss": 4.2502,
+      "step": 883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.040498442367601e-06,
+      "loss": 4.3173,
+      "step": 884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.0327102803738325e-06,
+      "loss": 4.3656,
+      "step": 885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.024922118380063e-06,
+      "loss": 4.4091,
+      "step": 886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.017133956386294e-06,
+      "loss": 4.5098,
+      "step": 887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.009345794392523e-06,
+      "loss": 4.0576,
+      "step": 888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.001557632398755e-06,
+      "loss": 3.9532,
+      "step": 889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.993769470404985e-06,
+      "loss": 4.3666,
+      "step": 890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.985981308411216e-06,
+      "loss": 4.2842,
+      "step": 891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.978193146417446e-06,
+      "loss": 4.3728,
+      "step": 892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.970404984423676e-06,
+      "loss": 4.5453,
+      "step": 893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.962616822429908e-06,
+      "loss": 4.2992,
+      "step": 894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.954828660436138e-06,
+      "loss": 4.2889,
+      "step": 895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.947040498442368e-06,
+      "loss": 4.082,
+      "step": 896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.9392523364485985e-06,
+      "loss": 4.2857,
+      "step": 897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.931464174454829e-06,
+      "loss": 4.4102,
+      "step": 898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.9236760124610605e-06,
+      "loss": 4.1493,
+      "step": 899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.91588785046729e-06,
+      "loss": 4.7925,
+      "step": 900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.908099688473521e-06,
+      "loss": 4.2732,
+      "step": 901
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.900311526479751e-06,
+      "loss": 4.0334,
+      "step": 902
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.892523364485983e-06,
+      "loss": 4.4657,
+      "step": 903
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.8847352024922125e-06,
+      "loss": 4.083,
+      "step": 904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.876947040498443e-06,
+      "loss": 4.2409,
+      "step": 905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.869158878504674e-06,
+      "loss": 4.3499,
+      "step": 906
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.861370716510904e-06,
+      "loss": 4.2303,
+      "step": 907
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.853582554517134e-06,
+      "loss": 4.147,
+      "step": 908
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.845794392523365e-06,
+      "loss": 4.305,
+      "step": 909
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.838006230529596e-06,
+      "loss": 4.1374,
+      "step": 910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.8302180685358264e-06,
+      "loss": 4.2121,
+      "step": 911
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.822429906542056e-06,
+      "loss": 4.057,
+      "step": 912
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.814641744548287e-06,
+      "loss": 4.1286,
+      "step": 913
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.806853582554518e-06,
+      "loss": 4.1481,
+      "step": 914
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.799065420560749e-06,
+      "loss": 4.1702,
+      "step": 915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.791277258566978e-06,
+      "loss": 4.151,
+      "step": 916
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.783489096573209e-06,
+      "loss": 4.0026,
+      "step": 917
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.77570093457944e-06,
+      "loss": 4.1047,
+      "step": 918
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.767912772585671e-06,
+      "loss": 4.2237,
+      "step": 919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.760124610591901e-06,
+      "loss": 3.9509,
+      "step": 920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.752336448598131e-06,
+      "loss": 4.193,
+      "step": 921
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.744548286604362e-06,
+      "loss": 3.9485,
+      "step": 922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.736760124610593e-06,
+      "loss": 4.0655,
+      "step": 923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.728971962616823e-06,
+      "loss": 4.1233,
+      "step": 924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.7211838006230535e-06,
+      "loss": 4.3378,
+      "step": 925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.713395638629284e-06,
+      "loss": 4.2682,
+      "step": 926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.705607476635515e-06,
+      "loss": 4.1078,
+      "step": 927
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.697819314641744e-06,
+      "loss": 4.02,
+      "step": 928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.690031152647976e-06,
+      "loss": 4.052,
+      "step": 929
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.682242990654206e-06,
+      "loss": 4.4095,
+      "step": 930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.674454828660437e-06,
+      "loss": 4.2643,
+      "step": 931
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 4.147,
+      "step": 932
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.658878504672898e-06,
+      "loss": 4.1059,
+      "step": 933
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.651090342679129e-06,
+      "loss": 4.3059,
+      "step": 934
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.643302180685359e-06,
+      "loss": 4.2833,
+      "step": 935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.635514018691589e-06,
+      "loss": 4.3016,
+      "step": 936
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.6277258566978195e-06,
+      "loss": 4.0642,
+      "step": 937
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.619937694704051e-06,
+      "loss": 3.9033,
+      "step": 938
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.6121495327102815e-06,
+      "loss": 3.9594,
+      "step": 939
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.604361370716511e-06,
+      "loss": 4.1977,
+      "step": 940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.596573208722742e-06,
+      "loss": 4.1937,
+      "step": 941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.588785046728972e-06,
+      "loss": 4.2532,
+      "step": 942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.580996884735204e-06,
+      "loss": 4.3022,
+      "step": 943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.5732087227414335e-06,
+      "loss": 4.1748,
+      "step": 944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.565420560747664e-06,
+      "loss": 4.3186,
+      "step": 945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.557632398753895e-06,
+      "loss": 4.4111,
+      "step": 946
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.549844236760126e-06,
+      "loss": 3.8426,
+      "step": 947
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.542056074766355e-06,
+      "loss": 4.374,
+      "step": 948
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.534267912772586e-06,
+      "loss": 5.0921,
+      "step": 949
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.526479750778817e-06,
+      "loss": 3.9403,
+      "step": 950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.5186915887850474e-06,
+      "loss": 4.1462,
+      "step": 951
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.510903426791277e-06,
+      "loss": 4.2318,
+      "step": 952
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.5031152647975086e-06,
+      "loss": 4.2531,
+      "step": 953
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.495327102803739e-06,
+      "loss": 3.9739,
+      "step": 954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.48753894080997e-06,
+      "loss": 4.3401,
+      "step": 955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.479750778816199e-06,
+      "loss": 4.0881,
+      "step": 956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.47196261682243e-06,
+      "loss": 4.2385,
+      "step": 957
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.464174454828661e-06,
+      "loss": 3.9696,
+      "step": 958
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.456386292834892e-06,
+      "loss": 4.2788,
+      "step": 959
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.448598130841122e-06,
+      "loss": 4.1104,
+      "step": 960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.440809968847352e-06,
+      "loss": 3.9995,
+      "step": 961
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.433021806853583e-06,
+      "loss": 4.2005,
+      "step": 962
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.425233644859814e-06,
+      "loss": 4.0187,
+      "step": 963
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.417445482866044e-06,
+      "loss": 4.1059,
+      "step": 964
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.4096573208722745e-06,
+      "loss": 4.1502,
+      "step": 965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.401869158878505e-06,
+      "loss": 4.0396,
+      "step": 966
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.3940809968847365e-06,
+      "loss": 4.1785,
+      "step": 967
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.386292834890966e-06,
+      "loss": 4.105,
+      "step": 968
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.378504672897197e-06,
+      "loss": 4.125,
+      "step": 969
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.370716510903427e-06,
+      "loss": 4.0963,
+      "step": 970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.362928348909658e-06,
+      "loss": 4.1919,
+      "step": 971
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.355140186915888e-06,
+      "loss": 4.0685,
+      "step": 972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.347352024922119e-06,
+      "loss": 4.2269,
+      "step": 973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.33956386292835e-06,
+      "loss": 4.2733,
+      "step": 974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.33177570093458e-06,
+      "loss": 4.2635,
+      "step": 975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.32398753894081e-06,
+      "loss": 4.1389,
+      "step": 976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.3161993769470405e-06,
+      "loss": 4.4508,
+      "step": 977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.308411214953272e-06,
+      "loss": 4.1115,
+      "step": 978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.3006230529595025e-06,
+      "loss": 3.9011,
+      "step": 979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.292834890965732e-06,
+      "loss": 4.0471,
+      "step": 980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.285046728971963e-06,
+      "loss": 3.8682,
+      "step": 981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.277258566978194e-06,
+      "loss": 3.907,
+      "step": 982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.269470404984425e-06,
+      "loss": 4.2864,
+      "step": 983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.2616822429906544e-06,
+      "loss": 4.2423,
+      "step": 984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.253894080996885e-06,
+      "loss": 4.0897,
+      "step": 985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.246105919003116e-06,
+      "loss": 4.0951,
+      "step": 986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.238317757009347e-06,
+      "loss": 4.203,
+      "step": 987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.230529595015577e-06,
+      "loss": 4.1875,
+      "step": 988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.222741433021807e-06,
+      "loss": 4.3312,
+      "step": 989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.214953271028038e-06,
+      "loss": 4.0853,
+      "step": 990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.207165109034268e-06,
+      "loss": 4.1096,
+      "step": 991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.199376947040498e-06,
+      "loss": 4.6277,
+      "step": 992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.1915887850467296e-06,
+      "loss": 4.3306,
+      "step": 993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.18380062305296e-06,
+      "loss": 4.2287,
+      "step": 994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.176012461059191e-06,
+      "loss": 4.1912,
+      "step": 995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.16822429906542e-06,
+      "loss": 3.8345,
+      "step": 996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.160436137071652e-06,
+      "loss": 4.216,
+      "step": 997
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.152647975077882e-06,
+      "loss": 4.877,
+      "step": 998
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.144859813084113e-06,
+      "loss": 4.4606,
+      "step": 999
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.137071651090343e-06,
+      "loss": 4.5621,
+      "step": 1000
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 4.196638584136963,
+      "eval_runtime": 980.7532,
+      "eval_samples_per_second": 2.694,
+      "eval_steps_per_second": 0.337,
+      "eval_wer": 1.9576556921856407,
+      "step": 1000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.129283489096573e-06,
+      "loss": 4.3725,
+      "step": 1001
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.121495327102805e-06,
+      "loss": 4.1794,
+      "step": 1002
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.113707165109035e-06,
+      "loss": 4.2651,
+      "step": 1003
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.105919003115265e-06,
+      "loss": 4.1898,
+      "step": 1004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.0981308411214955e-06,
+      "loss": 4.184,
+      "step": 1005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.090342679127726e-06,
+      "loss": 4.2282,
+      "step": 1006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.0825545171339575e-06,
+      "loss": 4.0185,
+      "step": 1007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.074766355140187e-06,
+      "loss": 4.2411,
+      "step": 1008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.066978193146418e-06,
+      "loss": 4.229,
+      "step": 1009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.059190031152648e-06,
+      "loss": 4.0801,
+      "step": 1010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.05140186915888e-06,
+      "loss": 4.2039,
+      "step": 1011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.0436137071651095e-06,
+      "loss": 4.1816,
+      "step": 1012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.03582554517134e-06,
+      "loss": 4.2359,
+      "step": 1013
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.028037383177571e-06,
+      "loss": 4.2413,
+      "step": 1014
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.020249221183801e-06,
+      "loss": 4.4647,
+      "step": 1015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.012461059190031e-06,
+      "loss": 4.1725,
+      "step": 1016
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.004672897196262e-06,
+      "loss": 4.2985,
+      "step": 1017
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.996884735202493e-06,
+      "loss": 4.2147,
+      "step": 1018
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.9890965732087235e-06,
+      "loss": 4.093,
+      "step": 1019
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.981308411214953e-06,
+      "loss": 4.1873,
+      "step": 1020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.973520249221184e-06,
+      "loss": 4.1716,
+      "step": 1021
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.965732087227415e-06,
+      "loss": 4.2233,
+      "step": 1022
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.957943925233646e-06,
+      "loss": 4.1742,
+      "step": 1023
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.9501557632398754e-06,
+      "loss": 4.2005,
+      "step": 1024
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.942367601246106e-06,
+      "loss": 4.0917,
+      "step": 1025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.9345794392523374e-06,
+      "loss": 3.8496,
+      "step": 1026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.926791277258568e-06,
+      "loss": 4.2641,
+      "step": 1027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.919003115264798e-06,
+      "loss": 4.1411,
+      "step": 1028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.911214953271028e-06,
+      "loss": 4.1717,
+      "step": 1029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.903426791277259e-06,
+      "loss": 4.2969,
+      "step": 1030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.89563862928349e-06,
+      "loss": 4.1628,
+      "step": 1031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.88785046728972e-06,
+      "loss": 4.0558,
+      "step": 1032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.8800623052959505e-06,
+      "loss": 4.2549,
+      "step": 1033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.872274143302181e-06,
+      "loss": 3.9411,
+      "step": 1034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.864485981308412e-06,
+      "loss": 4.133,
+      "step": 1035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.856697819314641e-06,
+      "loss": 4.1527,
+      "step": 1036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.848909657320873e-06,
+      "loss": 4.1898,
+      "step": 1037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.841121495327103e-06,
+      "loss": 4.2407,
+      "step": 1038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 4.2877,
+      "step": 1039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.825545171339564e-06,
+      "loss": 4.3204,
+      "step": 1040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.817757009345795e-06,
+      "loss": 4.4031,
+      "step": 1041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.809968847352026e-06,
+      "loss": 4.388,
+      "step": 1042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.802180685358256e-06,
+      "loss": 4.4793,
+      "step": 1043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.794392523364486e-06,
+      "loss": 4.2258,
+      "step": 1044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.7866043613707165e-06,
+      "loss": 4.3164,
+      "step": 1045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.778816199376948e-06,
+      "loss": 4.2953,
+      "step": 1046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.7710280373831785e-06,
+      "loss": 4.2508,
+      "step": 1047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.763239875389408e-06,
+      "loss": 4.1276,
+      "step": 1048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.755451713395639e-06,
+      "loss": 4.3287,
+      "step": 1049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.747663551401869e-06,
+      "loss": 5.1358,
+      "step": 1050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.739875389408101e-06,
+      "loss": 4.4907,
+      "step": 1051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.7320872274143305e-06,
+      "loss": 4.135,
+      "step": 1052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.724299065420561e-06,
+      "loss": 4.062,
+      "step": 1053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.716510903426792e-06,
+      "loss": 3.9202,
+      "step": 1054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.708722741433023e-06,
+      "loss": 4.0609,
+      "step": 1055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.700934579439253e-06,
+      "loss": 4.1416,
+      "step": 1056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.693146417445483e-06,
+      "loss": 4.3627,
+      "step": 1057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.685358255451714e-06,
+      "loss": 4.1286,
+      "step": 1058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6775700934579444e-06,
+      "loss": 4.1857,
+      "step": 1059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.669781931464174e-06,
+      "loss": 4.0396,
+      "step": 1060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.661993769470406e-06,
+      "loss": 4.0697,
+      "step": 1061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.654205607476636e-06,
+      "loss": 4.2515,
+      "step": 1062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.646417445482867e-06,
+      "loss": 4.3237,
+      "step": 1063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6386292834890964e-06,
+      "loss": 3.8274,
+      "step": 1064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.630841121495327e-06,
+      "loss": 4.1534,
+      "step": 1065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.623052959501558e-06,
+      "loss": 3.8641,
+      "step": 1066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.615264797507789e-06,
+      "loss": 4.2398,
+      "step": 1067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.607476635514019e-06,
+      "loss": 4.0672,
+      "step": 1068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.599688473520249e-06,
+      "loss": 4.355,
+      "step": 1069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.591900311526481e-06,
+      "loss": 3.9519,
+      "step": 1070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.584112149532711e-06,
+      "loss": 4.1608,
+      "step": 1071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.576323987538941e-06,
+      "loss": 3.9147,
+      "step": 1072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.5685358255451715e-06,
+      "loss": 4.0154,
+      "step": 1073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.560747663551402e-06,
+      "loss": 4.0972,
+      "step": 1074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.5529595015576335e-06,
+      "loss": 4.1326,
+      "step": 1075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.545171339563863e-06,
+      "loss": 4.2631,
+      "step": 1076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.537383177570094e-06,
+      "loss": 4.1387,
+      "step": 1077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.529595015576324e-06,
+      "loss": 4.1601,
+      "step": 1078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.521806853582555e-06,
+      "loss": 4.1318,
+      "step": 1079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.514018691588785e-06,
+      "loss": 4.3249,
+      "step": 1080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.506230529595016e-06,
+      "loss": 4.1001,
+      "step": 1081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.498442367601247e-06,
+      "loss": 4.1754,
+      "step": 1082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.490654205607477e-06,
+      "loss": 3.9749,
+      "step": 1083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.482866043613707e-06,
+      "loss": 4.1963,
+      "step": 1084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.4750778816199375e-06,
+      "loss": 4.0747,
+      "step": 1085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.467289719626169e-06,
+      "loss": 4.2849,
+      "step": 1086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.4595015576323995e-06,
+      "loss": 4.3783,
+      "step": 1087
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.451713395638629e-06,
+      "loss": 4.2105,
+      "step": 1088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.44392523364486e-06,
+      "loss": 4.3097,
+      "step": 1089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.436137071651091e-06,
+      "loss": 4.1305,
+      "step": 1090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.428348909657322e-06,
+      "loss": 4.0727,
+      "step": 1091
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.4205607476635515e-06,
+      "loss": 4.2829,
+      "step": 1092
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.412772585669782e-06,
+      "loss": 4.3875,
+      "step": 1093
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.404984423676013e-06,
+      "loss": 4.2861,
+      "step": 1094
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.397196261682244e-06,
+      "loss": 4.3298,
+      "step": 1095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.389408099688474e-06,
+      "loss": 4.6253,
+      "step": 1096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.381619937694704e-06,
+      "loss": 4.3327,
+      "step": 1097
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.373831775700935e-06,
+      "loss": 4.9714,
+      "step": 1098
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.3660436137071654e-06,
+      "loss": 4.4205,
+      "step": 1099
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.358255451713395e-06,
+      "loss": 4.7827,
+      "step": 1100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.350467289719627e-06,
+      "loss": 3.7994,
+      "step": 1101
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.342679127725857e-06,
+      "loss": 4.3291,
+      "step": 1102
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.334890965732088e-06,
+      "loss": 4.2071,
+      "step": 1103
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.3271028037383174e-06,
+      "loss": 3.9857,
+      "step": 1104
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.319314641744549e-06,
+      "loss": 3.9589,
+      "step": 1105
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.311526479750779e-06,
+      "loss": 4.0422,
+      "step": 1106
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.30373831775701e-06,
+      "loss": 3.8273,
+      "step": 1107
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.29595015576324e-06,
+      "loss": 4.0342,
+      "step": 1108
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.28816199376947e-06,
+      "loss": 4.2659,
+      "step": 1109
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.280373831775702e-06,
+      "loss": 4.1993,
+      "step": 1110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.272585669781932e-06,
+      "loss": 3.957,
+      "step": 1111
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.264797507788162e-06,
+      "loss": 4.3666,
+      "step": 1112
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.2570093457943925e-06,
+      "loss": 3.8206,
+      "step": 1113
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.249221183800623e-06,
+      "loss": 4.1937,
+      "step": 1114
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.2414330218068545e-06,
+      "loss": 4.3663,
+      "step": 1115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.233644859813084e-06,
+      "loss": 4.2154,
+      "step": 1116
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.225856697819315e-06,
+      "loss": 4.1445,
+      "step": 1117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.218068535825545e-06,
+      "loss": 4.1574,
+      "step": 1118
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.210280373831777e-06,
+      "loss": 4.1081,
+      "step": 1119
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.2024922118380065e-06,
+      "loss": 4.0098,
+      "step": 1120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.194704049844237e-06,
+      "loss": 4.2196,
+      "step": 1121
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.186915887850468e-06,
+      "loss": 4.1207,
+      "step": 1122
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.179127725856698e-06,
+      "loss": 3.9498,
+      "step": 1123
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.171339563862928e-06,
+      "loss": 4.0744,
+      "step": 1124
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.163551401869159e-06,
+      "loss": 4.3314,
+      "step": 1125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.15576323987539e-06,
+      "loss": 3.9115,
+      "step": 1126
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.1479750778816205e-06,
+      "loss": 4.0456,
+      "step": 1127
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.14018691588785e-06,
+      "loss": 4.4835,
+      "step": 1128
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.132398753894081e-06,
+      "loss": 3.8169,
+      "step": 1129
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.124610591900312e-06,
+      "loss": 4.4524,
+      "step": 1130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.116822429906543e-06,
+      "loss": 4.3265,
+      "step": 1131
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.1090342679127725e-06,
+      "loss": 4.1266,
+      "step": 1132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.101246105919003e-06,
+      "loss": 4.0143,
+      "step": 1133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.0934579439252344e-06,
+      "loss": 4.203,
+      "step": 1134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.085669781931465e-06,
+      "loss": 4.4167,
+      "step": 1135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.077881619937695e-06,
+      "loss": 4.1966,
+      "step": 1136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.070093457943925e-06,
+      "loss": 4.1847,
+      "step": 1137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.062305295950156e-06,
+      "loss": 4.3202,
+      "step": 1138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.054517133956387e-06,
+      "loss": 4.0995,
+      "step": 1139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.046728971962617e-06,
+      "loss": 4.1432,
+      "step": 1140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.0389408099688476e-06,
+      "loss": 4.1481,
+      "step": 1141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.031152647975078e-06,
+      "loss": 4.5226,
+      "step": 1142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.023364485981309e-06,
+      "loss": 4.3355,
+      "step": 1143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.0155763239875384e-06,
+      "loss": 4.4553,
+      "step": 1144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.00778816199377e-06,
+      "loss": 4.2385,
+      "step": 1145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5e-06,
+      "loss": 4.3004,
+      "step": 1146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.992211838006231e-06,
+      "loss": 4.283,
+      "step": 1147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.9844236760124615e-06,
+      "loss": 4.0697,
+      "step": 1148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.976635514018692e-06,
+      "loss": 4.0599,
+      "step": 1149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.968847352024923e-06,
+      "loss": 4.2851,
+      "step": 1150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.961059190031153e-06,
+      "loss": 4.0193,
+      "step": 1151
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.953271028037384e-06,
+      "loss": 4.1886,
+      "step": 1152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.945482866043614e-06,
+      "loss": 4.1533,
+      "step": 1153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.937694704049845e-06,
+      "loss": 4.304,
+      "step": 1154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.9299065420560755e-06,
+      "loss": 4.2972,
+      "step": 1155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.922118380062306e-06,
+      "loss": 4.2194,
+      "step": 1156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.914330218068537e-06,
+      "loss": 4.1332,
+      "step": 1157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.906542056074766e-06,
+      "loss": 4.1679,
+      "step": 1158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.898753894080998e-06,
+      "loss": 4.0246,
+      "step": 1159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8909657320872275e-06,
+      "loss": 3.9246,
+      "step": 1160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.883177570093459e-06,
+      "loss": 4.309,
+      "step": 1161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.875389408099689e-06,
+      "loss": 4.1444,
+      "step": 1162
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.86760124610592e-06,
+      "loss": 4.1607,
+      "step": 1163
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.85981308411215e-06,
+      "loss": 4.0334,
+      "step": 1164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.85202492211838e-06,
+      "loss": 4.1034,
+      "step": 1165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.844236760124611e-06,
+      "loss": 4.2553,
+      "step": 1166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8364485981308415e-06,
+      "loss": 4.0834,
+      "step": 1167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.828660436137072e-06,
+      "loss": 3.8848,
+      "step": 1168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.820872274143303e-06,
+      "loss": 4.2794,
+      "step": 1169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.813084112149533e-06,
+      "loss": 4.2984,
+      "step": 1170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.805295950155764e-06,
+      "loss": 3.8147,
+      "step": 1171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.797507788161994e-06,
+      "loss": 3.9968,
+      "step": 1172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.789719626168225e-06,
+      "loss": 4.3356,
+      "step": 1173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.7819314641744554e-06,
+      "loss": 4.0179,
+      "step": 1174
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.774143302180686e-06,
+      "loss": 3.9366,
+      "step": 1175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.766355140186917e-06,
+      "loss": 4.2173,
+      "step": 1176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.758566978193147e-06,
+      "loss": 4.2946,
+      "step": 1177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.750778816199378e-06,
+      "loss": 4.1843,
+      "step": 1178
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.742990654205608e-06,
+      "loss": 3.8046,
+      "step": 1179
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.735202492211838e-06,
+      "loss": 4.1641,
+      "step": 1180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.727414330218069e-06,
+      "loss": 4.0206,
+      "step": 1181
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.719626168224299e-06,
+      "loss": 4.0824,
+      "step": 1182
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.7118380062305305e-06,
+      "loss": 4.0093,
+      "step": 1183
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.70404984423676e-06,
+      "loss": 4.0309,
+      "step": 1184
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.696261682242992e-06,
+      "loss": 4.5318,
+      "step": 1185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.688473520249221e-06,
+      "loss": 3.9508,
+      "step": 1186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.680685358255452e-06,
+      "loss": 4.1999,
+      "step": 1187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6728971962616825e-06,
+      "loss": 4.0963,
+      "step": 1188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.665109034267913e-06,
+      "loss": 4.0061,
+      "step": 1189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.657320872274144e-06,
+      "loss": 3.8852,
+      "step": 1190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.649532710280374e-06,
+      "loss": 4.2554,
+      "step": 1191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.641744548286605e-06,
+      "loss": 4.2202,
+      "step": 1192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.633956386292835e-06,
+      "loss": 4.6014,
+      "step": 1193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.626168224299066e-06,
+      "loss": 4.2183,
+      "step": 1194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6183800623052965e-06,
+      "loss": 4.2598,
+      "step": 1195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.610591900311527e-06,
+      "loss": 4.0301,
+      "step": 1196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.602803738317758e-06,
+      "loss": 4.1544,
+      "step": 1197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.595015576323988e-06,
+      "loss": 4.4703,
+      "step": 1198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.587227414330219e-06,
+      "loss": 4.3941,
+      "step": 1199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.579439252336449e-06,
+      "loss": 4.0286,
+      "step": 1200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.57165109034268e-06,
+      "loss": 4.0868,
+      "step": 1201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.56386292834891e-06,
+      "loss": 4.1264,
+      "step": 1202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.556074766355141e-06,
+      "loss": 4.0422,
+      "step": 1203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.548286604361371e-06,
+      "loss": 3.9118,
+      "step": 1204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.540498442367602e-06,
+      "loss": 3.9195,
+      "step": 1205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.532710280373832e-06,
+      "loss": 4.2063,
+      "step": 1206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.524922118380063e-06,
+      "loss": 4.3008,
+      "step": 1207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.517133956386293e-06,
+      "loss": 4.1016,
+      "step": 1208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.509345794392524e-06,
+      "loss": 4.0203,
+      "step": 1209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.501557632398754e-06,
+      "loss": 3.9111,
+      "step": 1210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.493769470404985e-06,
+      "loss": 4.2008,
+      "step": 1211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.485981308411215e-06,
+      "loss": 4.1822,
+      "step": 1212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.478193146417446e-06,
+      "loss": 4.2201,
+      "step": 1213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.4704049844236764e-06,
+      "loss": 4.0373,
+      "step": 1214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.462616822429907e-06,
+      "loss": 4.1144,
+      "step": 1215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.4548286604361376e-06,
+      "loss": 4.1333,
+      "step": 1216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.447040498442368e-06,
+      "loss": 4.0006,
+      "step": 1217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.439252336448599e-06,
+      "loss": 3.9949,
+      "step": 1218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.431464174454829e-06,
+      "loss": 3.9206,
+      "step": 1219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.42367601246106e-06,
+      "loss": 4.3099,
+      "step": 1220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.41588785046729e-06,
+      "loss": 4.1466,
+      "step": 1221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.40809968847352e-06,
+      "loss": 3.9109,
+      "step": 1222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4003115264797515e-06,
+      "loss": 4.2666,
+      "step": 1223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.392523364485981e-06,
+      "loss": 4.0382,
+      "step": 1224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.384735202492213e-06,
+      "loss": 4.2135,
+      "step": 1225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.376947040498442e-06,
+      "loss": 4.2339,
+      "step": 1226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.369158878504674e-06,
+      "loss": 4.1069,
+      "step": 1227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3613707165109035e-06,
+      "loss": 4.0805,
+      "step": 1228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.353582554517134e-06,
+      "loss": 4.1242,
+      "step": 1229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.345794392523365e-06,
+      "loss": 4.1509,
+      "step": 1230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.338006230529595e-06,
+      "loss": 4.1737,
+      "step": 1231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.330218068535826e-06,
+      "loss": 4.0042,
+      "step": 1232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.322429906542056e-06,
+      "loss": 4.1197,
+      "step": 1233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.314641744548287e-06,
+      "loss": 4.0168,
+      "step": 1234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3068535825545175e-06,
+      "loss": 4.4241,
+      "step": 1235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.299065420560748e-06,
+      "loss": 4.1408,
+      "step": 1236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.291277258566979e-06,
+      "loss": 4.0476,
+      "step": 1237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.283489096573209e-06,
+      "loss": 3.9629,
+      "step": 1238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.27570093457944e-06,
+      "loss": 4.2337,
+      "step": 1239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.26791277258567e-06,
+      "loss": 4.2204,
+      "step": 1240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.260124610591901e-06,
+      "loss": 4.2513,
+      "step": 1241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2523364485981315e-06,
+      "loss": 4.5944,
+      "step": 1242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.244548286604362e-06,
+      "loss": 4.1722,
+      "step": 1243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.236760124610592e-06,
+      "loss": 4.2814,
+      "step": 1244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.228971962616823e-06,
+      "loss": 3.9282,
+      "step": 1245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.221183800623053e-06,
+      "loss": 4.4213,
+      "step": 1246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.213395638629284e-06,
+      "loss": 4.3468,
+      "step": 1247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.205607476635514e-06,
+      "loss": 4.2278,
+      "step": 1248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1978193146417454e-06,
+      "loss": 4.5141,
+      "step": 1249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.190031152647975e-06,
+      "loss": 4.7197,
+      "step": 1250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.182242990654206e-06,
+      "loss": 4.2493,
+      "step": 1251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.174454828660436e-06,
+      "loss": 3.9824,
+      "step": 1252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 4.2274,
+      "step": 1253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1588785046728974e-06,
+      "loss": 4.2098,
+      "step": 1254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.151090342679128e-06,
+      "loss": 4.3067,
+      "step": 1255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1433021806853586e-06,
+      "loss": 4.0981,
+      "step": 1256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.135514018691589e-06,
+      "loss": 4.0265,
+      "step": 1257
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.12772585669782e-06,
+      "loss": 3.8669,
+      "step": 1258
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.11993769470405e-06,
+      "loss": 4.1094,
+      "step": 1259
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.112149532710281e-06,
+      "loss": 4.4217,
+      "step": 1260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.104361370716511e-06,
+      "loss": 3.9477,
+      "step": 1261
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.096573208722742e-06,
+      "loss": 4.2195,
+      "step": 1262
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0887850467289725e-06,
+      "loss": 4.2896,
+      "step": 1263
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.080996884735203e-06,
+      "loss": 4.1322,
+      "step": 1264
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.073208722741434e-06,
+      "loss": 4.2486,
+      "step": 1265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.065420560747663e-06,
+      "loss": 4.2184,
+      "step": 1266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.057632398753895e-06,
+      "loss": 3.9987,
+      "step": 1267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0498442367601245e-06,
+      "loss": 4.0496,
+      "step": 1268
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.042056074766356e-06,
+      "loss": 4.2509,
+      "step": 1269
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.034267912772586e-06,
+      "loss": 4.0227,
+      "step": 1270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.026479750778817e-06,
+      "loss": 4.0178,
+      "step": 1271
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.018691588785047e-06,
+      "loss": 4.0382,
+      "step": 1272
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.010903426791277e-06,
+      "loss": 4.3418,
+      "step": 1273
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.003115264797508e-06,
+      "loss": 4.0959,
+      "step": 1274
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9953271028037385e-06,
+      "loss": 4.1287,
+      "step": 1275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.987538940809969e-06,
+      "loss": 4.4,
+      "step": 1276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9797507788162e-06,
+      "loss": 4.2729,
+      "step": 1277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.97196261682243e-06,
+      "loss": 4.1811,
+      "step": 1278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.964174454828661e-06,
+      "loss": 4.3166,
+      "step": 1279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.956386292834891e-06,
+      "loss": 4.25,
+      "step": 1280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.948598130841122e-06,
+      "loss": 4.2391,
+      "step": 1281
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9408099688473525e-06,
+      "loss": 4.1273,
+      "step": 1282
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.933021806853583e-06,
+      "loss": 4.3166,
+      "step": 1283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.925233644859814e-06,
+      "loss": 4.128,
+      "step": 1284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.917445482866044e-06,
+      "loss": 4.124,
+      "step": 1285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.909657320872275e-06,
+      "loss": 4.2077,
+      "step": 1286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.901869158878505e-06,
+      "loss": 4.1628,
+      "step": 1287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.894080996884735e-06,
+      "loss": 4.138,
+      "step": 1288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8862928348909664e-06,
+      "loss": 4.1165,
+      "step": 1289
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.878504672897196e-06,
+      "loss": 3.8616,
+      "step": 1290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8707165109034276e-06,
+      "loss": 4.2893,
+      "step": 1291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.862928348909657e-06,
+      "loss": 4.2285,
+      "step": 1292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855140186915889e-06,
+      "loss": 4.5745,
+      "step": 1293
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8473520249221184e-06,
+      "loss": 4.0336,
+      "step": 1294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.839563862928349e-06,
+      "loss": 4.0037,
+      "step": 1295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8317757009345796e-06,
+      "loss": 3.9747,
+      "step": 1296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.82398753894081e-06,
+      "loss": 4.0977,
+      "step": 1297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.816199376947041e-06,
+      "loss": 3.8144,
+      "step": 1298
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8084112149532717e-06,
+      "loss": 4.6197,
+      "step": 1299
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.800623052959502e-06,
+      "loss": 3.9711,
+      "step": 1300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7928348909657324e-06,
+      "loss": 4.179,
+      "step": 1301
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.785046728971963e-06,
+      "loss": 4.3084,
+      "step": 1302
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7772585669781935e-06,
+      "loss": 4.1713,
+      "step": 1303
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7694704049844237e-06,
+      "loss": 4.177,
+      "step": 1304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7616822429906547e-06,
+      "loss": 4.0054,
+      "step": 1305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.753894080996885e-06,
+      "loss": 4.1164,
+      "step": 1306
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.746105919003116e-06,
+      "loss": 4.1165,
+      "step": 1307
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.738317757009346e-06,
+      "loss": 3.9715,
+      "step": 1308
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.730529595015577e-06,
+      "loss": 4.139,
+      "step": 1309
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.722741433021807e-06,
+      "loss": 3.9481,
+      "step": 1310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7149532710280376e-06,
+      "loss": 4.1057,
+      "step": 1311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7071651090342682e-06,
+      "loss": 4.2007,
+      "step": 1312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6993769470404988e-06,
+      "loss": 4.1304,
+      "step": 1313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.691588785046729e-06,
+      "loss": 3.9657,
+      "step": 1314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.68380062305296e-06,
+      "loss": 4.3177,
+      "step": 1315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.67601246105919e-06,
+      "loss": 4.0668,
+      "step": 1316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.668224299065421e-06,
+      "loss": 4.0704,
+      "step": 1317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.660436137071651e-06,
+      "loss": 4.4594,
+      "step": 1318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.652647975077882e-06,
+      "loss": 4.0802,
+      "step": 1319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6448598130841123e-06,
+      "loss": 3.8377,
+      "step": 1320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.637071651090343e-06,
+      "loss": 3.8676,
+      "step": 1321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6292834890965735e-06,
+      "loss": 4.0835,
+      "step": 1322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.621495327102804e-06,
+      "loss": 4.0337,
+      "step": 1323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6137071651090346e-06,
+      "loss": 4.1022,
+      "step": 1324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.605919003115265e-06,
+      "loss": 4.0927,
+      "step": 1325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5981308411214953e-06,
+      "loss": 3.9976,
+      "step": 1326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5903426791277263e-06,
+      "loss": 3.8563,
+      "step": 1327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5825545171339564e-06,
+      "loss": 3.978,
+      "step": 1328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5747663551401874e-06,
+      "loss": 4.2367,
+      "step": 1329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5669781931464176e-06,
+      "loss": 4.1864,
+      "step": 1330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5591900311526486e-06,
+      "loss": 3.9428,
+      "step": 1331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5514018691588787e-06,
+      "loss": 3.9893,
+      "step": 1332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5436137071651093e-06,
+      "loss": 4.1073,
+      "step": 1333
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.53582554517134e-06,
+      "loss": 4.1768,
+      "step": 1334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5280373831775704e-06,
+      "loss": 3.9439,
+      "step": 1335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5202492211838006e-06,
+      "loss": 4.0671,
+      "step": 1336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5124610591900315e-06,
+      "loss": 4.2565,
+      "step": 1337
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5046728971962617e-06,
+      "loss": 4.3157,
+      "step": 1338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4968847352024927e-06,
+      "loss": 4.4216,
+      "step": 1339
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.489096573208723e-06,
+      "loss": 4.2919,
+      "step": 1340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.481308411214954e-06,
+      "loss": 4.3521,
+      "step": 1341
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.473520249221184e-06,
+      "loss": 4.0826,
+      "step": 1342
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4657320872274145e-06,
+      "loss": 4.3514,
+      "step": 1343
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.457943925233645e-06,
+      "loss": 4.0152,
+      "step": 1344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4501557632398757e-06,
+      "loss": 4.233,
+      "step": 1345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4423676012461062e-06,
+      "loss": 4.3225,
+      "step": 1346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.434579439252337e-06,
+      "loss": 4.5704,
+      "step": 1347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.426791277258567e-06,
+      "loss": 4.1839,
+      "step": 1348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.419003115264798e-06,
+      "loss": 4.1595,
+      "step": 1349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.411214953271028e-06,
+      "loss": 4.6417,
+      "step": 1350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.403426791277259e-06,
+      "loss": 4.2985,
+      "step": 1351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.395638629283489e-06,
+      "loss": 4.2394,
+      "step": 1352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.38785046728972e-06,
+      "loss": 4.1823,
+      "step": 1353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3800623052959503e-06,
+      "loss": 4.2178,
+      "step": 1354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.372274143302181e-06,
+      "loss": 4.0713,
+      "step": 1355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3644859813084115e-06,
+      "loss": 4.3914,
+      "step": 1356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.356697819314642e-06,
+      "loss": 3.903,
+      "step": 1357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.348909657320872e-06,
+      "loss": 4.2348,
+      "step": 1358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.341121495327103e-06,
+      "loss": 4.2705,
+      "step": 1359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 4.5239,
+      "step": 1360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3255451713395643e-06,
+      "loss": 4.111,
+      "step": 1361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3177570093457945e-06,
+      "loss": 4.0084,
+      "step": 1362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3099688473520254e-06,
+      "loss": 4.0295,
+      "step": 1363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3021806853582556e-06,
+      "loss": 4.4013,
+      "step": 1364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.294392523364486e-06,
+      "loss": 4.1019,
+      "step": 1365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.2866043613707167e-06,
+      "loss": 3.8966,
+      "step": 1366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.2788161993769473e-06,
+      "loss": 4.1694,
+      "step": 1367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.2710280373831774e-06,
+      "loss": 3.9826,
+      "step": 1368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.2632398753894084e-06,
+      "loss": 4.3527,
+      "step": 1369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.2554517133956386e-06,
+      "loss": 4.3275,
+      "step": 1370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.2476635514018696e-06,
+      "loss": 4.2342,
+      "step": 1371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.2398753894080997e-06,
+      "loss": 3.9893,
+      "step": 1372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.2320872274143307e-06,
+      "loss": 4.1548,
+      "step": 1373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.224299065420561e-06,
+      "loss": 4.2497,
+      "step": 1374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.2165109034267914e-06,
+      "loss": 3.8813,
+      "step": 1375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.208722741433022e-06,
+      "loss": 3.9037,
+      "step": 1376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.2009345794392525e-06,
+      "loss": 4.0796,
+      "step": 1377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.193146417445483e-06,
+      "loss": 4.0803,
+      "step": 1378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1853582554517137e-06,
+      "loss": 4.1428,
+      "step": 1379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.177570093457944e-06,
+      "loss": 4.035,
+      "step": 1380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.169781931464175e-06,
+      "loss": 3.8469,
+      "step": 1381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.161993769470405e-06,
+      "loss": 4.3625,
+      "step": 1382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.154205607476636e-06,
+      "loss": 4.0223,
+      "step": 1383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.146417445482866e-06,
+      "loss": 4.2868,
+      "step": 1384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.138629283489097e-06,
+      "loss": 4.3609,
+      "step": 1385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.1308411214953272e-06,
+      "loss": 4.2169,
+      "step": 1386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.123052959501558e-06,
+      "loss": 4.234,
+      "step": 1387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.1152647975077884e-06,
+      "loss": 4.3051,
+      "step": 1388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.107476635514019e-06,
+      "loss": 4.1314,
+      "step": 1389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.099688473520249e-06,
+      "loss": 4.1721,
+      "step": 1390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.09190031152648e-06,
+      "loss": 3.9602,
+      "step": 1391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.08411214953271e-06,
+      "loss": 4.1206,
+      "step": 1392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.076323987538941e-06,
+      "loss": 4.2263,
+      "step": 1393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0685358255451713e-06,
+      "loss": 4.4255,
+      "step": 1394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0607476635514023e-06,
+      "loss": 4.2539,
+      "step": 1395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0529595015576325e-06,
+      "loss": 4.2001,
+      "step": 1396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.045171339563863e-06,
+      "loss": 4.2896,
+      "step": 1397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0373831775700936e-06,
+      "loss": 4.217,
+      "step": 1398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.029595015576324e-06,
+      "loss": 4.4061,
+      "step": 1399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0218068535825547e-06,
+      "loss": 4.4829,
+      "step": 1400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0140186915887853e-06,
+      "loss": 4.0196,
+      "step": 1401
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0062305295950155e-06,
+      "loss": 4.1599,
+      "step": 1402
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9984423676012464e-06,
+      "loss": 4.0471,
+      "step": 1403
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9906542056074766e-06,
+      "loss": 3.9683,
+      "step": 1404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9828660436137076e-06,
+      "loss": 3.9772,
+      "step": 1405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9750778816199377e-06,
+      "loss": 4.3239,
+      "step": 1406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9672897196261687e-06,
+      "loss": 4.1608,
+      "step": 1407
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.959501557632399e-06,
+      "loss": 3.9765,
+      "step": 1408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9517133956386294e-06,
+      "loss": 4.041,
+      "step": 1409
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.94392523364486e-06,
+      "loss": 3.9123,
+      "step": 1410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9361370716510906e-06,
+      "loss": 3.9731,
+      "step": 1411
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9283489096573207e-06,
+      "loss": 4.0073,
+      "step": 1412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.9205607476635517e-06,
+      "loss": 4.2421,
+      "step": 1413
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.912772585669782e-06,
+      "loss": 4.1833,
+      "step": 1414
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.904984423676013e-06,
+      "loss": 4.2527,
+      "step": 1415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.897196261682243e-06,
+      "loss": 4.3496,
+      "step": 1416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.889408099688474e-06,
+      "loss": 4.1174,
+      "step": 1417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.881619937694704e-06,
+      "loss": 4.1074,
+      "step": 1418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8738317757009347e-06,
+      "loss": 4.1737,
+      "step": 1419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8660436137071652e-06,
+      "loss": 3.9407,
+      "step": 1420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.858255451713396e-06,
+      "loss": 4.1445,
+      "step": 1421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8504672897196264e-06,
+      "loss": 4.2814,
+      "step": 1422
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.842679127725857e-06,
+      "loss": 4.1821,
+      "step": 1423
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.834890965732087e-06,
+      "loss": 4.0798,
+      "step": 1424
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.827102803738318e-06,
+      "loss": 4.2576,
+      "step": 1425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8193146417445482e-06,
+      "loss": 4.2851,
+      "step": 1426
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.811526479750779e-06,
+      "loss": 4.0919,
+      "step": 1427
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8037383177570094e-06,
+      "loss": 4.0561,
+      "step": 1428
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7959501557632403e-06,
+      "loss": 4.2283,
+      "step": 1429
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7881619937694705e-06,
+      "loss": 4.3297,
+      "step": 1430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.780373831775701e-06,
+      "loss": 4.1965,
+      "step": 1431
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7725856697819316e-06,
+      "loss": 4.2335,
+      "step": 1432
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.764797507788162e-06,
+      "loss": 4.3696,
+      "step": 1433
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7570093457943923e-06,
+      "loss": 3.9912,
+      "step": 1434
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7492211838006233e-06,
+      "loss": 4.0042,
+      "step": 1435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7414330218068535e-06,
+      "loss": 3.955,
+      "step": 1436
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7336448598130845e-06,
+      "loss": 4.3391,
+      "step": 1437
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7258566978193146e-06,
+      "loss": 4.0451,
+      "step": 1438
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7180685358255456e-06,
+      "loss": 4.2457,
+      "step": 1439
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7102803738317757e-06,
+      "loss": 4.1973,
+      "step": 1440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7024922118380063e-06,
+      "loss": 4.1781,
+      "step": 1441
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.694704049844237e-06,
+      "loss": 4.0822,
+      "step": 1442
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6869158878504674e-06,
+      "loss": 4.0379,
+      "step": 1443
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6791277258566976e-06,
+      "loss": 4.184,
+      "step": 1444
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6713395638629286e-06,
+      "loss": 4.2757,
+      "step": 1445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6635514018691587e-06,
+      "loss": 4.292,
+      "step": 1446
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6557632398753897e-06,
+      "loss": 4.2613,
+      "step": 1447
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.64797507788162e-06,
+      "loss": 4.4555,
+      "step": 1448
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.640186915887851e-06,
+      "loss": 4.4066,
+      "step": 1449
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.632398753894081e-06,
+      "loss": 4.2581,
+      "step": 1450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6246105919003116e-06,
+      "loss": 4.1151,
+      "step": 1451
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.616822429906542e-06,
+      "loss": 4.1259,
+      "step": 1452
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6090342679127727e-06,
+      "loss": 3.9906,
+      "step": 1453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6012461059190033e-06,
+      "loss": 3.9352,
+      "step": 1454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.593457943925234e-06,
+      "loss": 3.9873,
+      "step": 1455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.585669781931464e-06,
+      "loss": 4.0714,
+      "step": 1456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.577881619937695e-06,
+      "loss": 4.3342,
+      "step": 1457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.570093457943925e-06,
+      "loss": 4.1305,
+      "step": 1458
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.562305295950156e-06,
+      "loss": 4.0147,
+      "step": 1459
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.5545171339563862e-06,
+      "loss": 4.1794,
+      "step": 1460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.5467289719626172e-06,
+      "loss": 4.0886,
+      "step": 1461
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.5389408099688474e-06,
+      "loss": 4.2048,
+      "step": 1462
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.531152647975078e-06,
+      "loss": 4.1895,
+      "step": 1463
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.5233644859813085e-06,
+      "loss": 4.126,
+      "step": 1464
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.515576323987539e-06,
+      "loss": 4.3393,
+      "step": 1465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.5077881619937692e-06,
+      "loss": 3.8572,
+      "step": 1466
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.5e-06,
+      "loss": 4.1,
+      "step": 1467
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.4922118380062308e-06,
+      "loss": 4.1728,
+      "step": 1468
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.4844236760124613e-06,
+      "loss": 3.8288,
+      "step": 1469
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.476635514018692e-06,
+      "loss": 3.9488,
+      "step": 1470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.4688473520249225e-06,
+      "loss": 4.0971,
+      "step": 1471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.461059190031153e-06,
+      "loss": 3.8855,
+      "step": 1472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.453271028037383e-06,
+      "loss": 4.0871,
+      "step": 1473
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4454828660436138e-06,
+      "loss": 3.8353,
+      "step": 1474
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4376947040498443e-06,
+      "loss": 4.1605,
+      "step": 1475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.429906542056075e-06,
+      "loss": 4.1325,
+      "step": 1476
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4221183800623055e-06,
+      "loss": 4.2061,
+      "step": 1477
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.414330218068536e-06,
+      "loss": 3.871,
+      "step": 1478
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4065420560747666e-06,
+      "loss": 4.1769,
+      "step": 1479
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.398753894080997e-06,
+      "loss": 3.8962,
+      "step": 1480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.3909657320872277e-06,
+      "loss": 4.0905,
+      "step": 1481
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.3831775700934583e-06,
+      "loss": 4.0966,
+      "step": 1482
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.375389408099689e-06,
+      "loss": 4.1805,
+      "step": 1483
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.367601246105919e-06,
+      "loss": 3.7803,
+      "step": 1484
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.3598130841121496e-06,
+      "loss": 4.136,
+      "step": 1485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.35202492211838e-06,
+      "loss": 4.2422,
+      "step": 1486
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.3442367601246107e-06,
+      "loss": 4.3106,
+      "step": 1487
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.3364485981308413e-06,
+      "loss": 4.2179,
+      "step": 1488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.328660436137072e-06,
+      "loss": 4.2121,
+      "step": 1489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.3208722741433024e-06,
+      "loss": 4.0003,
+      "step": 1490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.313084112149533e-06,
+      "loss": 4.4283,
+      "step": 1491
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.3052959501557635e-06,
+      "loss": 4.1457,
+      "step": 1492
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.297507788161994e-06,
+      "loss": 3.8459,
+      "step": 1493
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2897196261682247e-06,
+      "loss": 3.9624,
+      "step": 1494
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.281931464174455e-06,
+      "loss": 4.1517,
+      "step": 1495
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2741433021806854e-06,
+      "loss": 4.5228,
+      "step": 1496
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.266355140186916e-06,
+      "loss": 4.3213,
+      "step": 1497
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2585669781931465e-06,
+      "loss": 4.52,
+      "step": 1498
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.250778816199377e-06,
+      "loss": 4.0674,
+      "step": 1499
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2429906542056077e-06,
+      "loss": 4.3547,
+      "step": 1500
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 4.166304588317871,
+      "eval_runtime": 982.6549,
+      "eval_samples_per_second": 2.689,
+      "eval_steps_per_second": 0.337,
+      "eval_wer": 1.918524395081317,
+      "step": 1500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2352024922118382e-06,
+      "loss": 4.2198,
+      "step": 1501
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2274143302180688e-06,
+      "loss": 4.1961,
+      "step": 1502
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2196261682242994e-06,
+      "loss": 4.1441,
+      "step": 1503
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.21183800623053e-06,
+      "loss": 3.9946,
+      "step": 1504
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.20404984423676e-06,
+      "loss": 4.241,
+      "step": 1505
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.1962616822429906e-06,
+      "loss": 4.1064,
+      "step": 1506
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.188473520249221e-06,
+      "loss": 4.0086,
+      "step": 1507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1806853582554518e-06,
+      "loss": 4.1023,
+      "step": 1508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1728971962616823e-06,
+      "loss": 4.1471,
+      "step": 1509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.165109034267913e-06,
+      "loss": 4.1109,
+      "step": 1510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1573208722741435e-06,
+      "loss": 3.933,
+      "step": 1511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.149532710280374e-06,
+      "loss": 4.2576,
+      "step": 1512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1417445482866046e-06,
+      "loss": 3.9325,
+      "step": 1513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.133956386292835e-06,
+      "loss": 4.1004,
+      "step": 1514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1261682242990657e-06,
+      "loss": 4.066,
+      "step": 1515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.118380062305296e-06,
+      "loss": 4.3607,
+      "step": 1516
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1105919003115264e-06,
+      "loss": 4.0542,
+      "step": 1517
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.102803738317757e-06,
+      "loss": 4.2241,
+      "step": 1518
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.0950155763239876e-06,
+      "loss": 4.3095,
+      "step": 1519
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.087227414330218e-06,
+      "loss": 4.0374,
+      "step": 1520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.0794392523364487e-06,
+      "loss": 4.0323,
+      "step": 1521
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.0716510903426793e-06,
+      "loss": 4.0204,
+      "step": 1522
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.06386292834891e-06,
+      "loss": 4.234,
+      "step": 1523
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.0560747663551404e-06,
+      "loss": 4.0068,
+      "step": 1524
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.048286604361371e-06,
+      "loss": 4.1383,
+      "step": 1525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0404984423676016e-06,
+      "loss": 3.9631,
+      "step": 1526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0327102803738317e-06,
+      "loss": 4.0699,
+      "step": 1527
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0249221183800623e-06,
+      "loss": 4.1097,
+      "step": 1528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.017133956386293e-06,
+      "loss": 3.6756,
+      "step": 1529
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0093457943925234e-06,
+      "loss": 3.9695,
+      "step": 1530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.001557632398754e-06,
+      "loss": 3.9809,
+      "step": 1531
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9937694704049845e-06,
+      "loss": 3.8912,
+      "step": 1532
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.985981308411215e-06,
+      "loss": 3.9909,
+      "step": 1533
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9781931464174457e-06,
+      "loss": 4.1181,
+      "step": 1534
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9704049844236762e-06,
+      "loss": 4.0765,
+      "step": 1535
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.962616822429907e-06,
+      "loss": 3.8671,
+      "step": 1536
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9548286604361374e-06,
+      "loss": 3.982,
+      "step": 1537
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9470404984423675e-06,
+      "loss": 4.3603,
+      "step": 1538
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.939252336448598e-06,
+      "loss": 4.0482,
+      "step": 1539
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9314641744548286e-06,
+      "loss": 4.1731,
+      "step": 1540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9236760124610592e-06,
+      "loss": 4.2364,
+      "step": 1541
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9158878504672898e-06,
+      "loss": 3.8411,
+      "step": 1542
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9080996884735203e-06,
+      "loss": 4.2928,
+      "step": 1543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.900311526479751e-06,
+      "loss": 4.1272,
+      "step": 1544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8925233644859815e-06,
+      "loss": 4.4892,
+      "step": 1545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8847352024922118e-06,
+      "loss": 3.9688,
+      "step": 1546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8769470404984424e-06,
+      "loss": 4.3333,
+      "step": 1547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.869158878504673e-06,
+      "loss": 4.2779,
+      "step": 1548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8613707165109035e-06,
+      "loss": 4.5485,
+      "step": 1549
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8535825545171341e-06,
+      "loss": 4.0159,
+      "step": 1550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8457943925233645e-06,
+      "loss": 4.2975,
+      "step": 1551
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.838006230529595e-06,
+      "loss": 4.1462,
+      "step": 1552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8302180685358256e-06,
+      "loss": 4.0861,
+      "step": 1553
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8224299065420562e-06,
+      "loss": 3.9723,
+      "step": 1554
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8146417445482867e-06,
+      "loss": 4.2419,
+      "step": 1555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8068535825545173e-06,
+      "loss": 4.4392,
+      "step": 1556
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7990654205607477e-06,
+      "loss": 4.223,
+      "step": 1557
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7912772585669782e-06,
+      "loss": 4.2847,
+      "step": 1558
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7834890965732088e-06,
+      "loss": 3.9438,
+      "step": 1559
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7757009345794394e-06,
+      "loss": 4.1991,
+      "step": 1560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.76791277258567e-06,
+      "loss": 4.1895,
+      "step": 1561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7601246105919003e-06,
+      "loss": 4.1453,
+      "step": 1562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7523364485981308e-06,
+      "loss": 4.2729,
+      "step": 1563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7445482866043614e-06,
+      "loss": 4.0038,
+      "step": 1564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.736760124610592e-06,
+      "loss": 4.2303,
+      "step": 1565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7289719626168225e-06,
+      "loss": 4.0141,
+      "step": 1566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7211838006230531e-06,
+      "loss": 3.85,
+      "step": 1567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7133956386292835e-06,
+      "loss": 4.133,
+      "step": 1568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.705607476635514e-06,
+      "loss": 4.198,
+      "step": 1569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6978193146417446e-06,
+      "loss": 4.0277,
+      "step": 1570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6900311526479752e-06,
+      "loss": 4.127,
+      "step": 1571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6822429906542057e-06,
+      "loss": 4.3323,
+      "step": 1572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.674454828660436e-06,
+      "loss": 4.2751,
+      "step": 1573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 3.9676,
+      "step": 1574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6588785046728972e-06,
+      "loss": 3.9231,
+      "step": 1575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6510903426791278e-06,
+      "loss": 4.0464,
+      "step": 1576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6433021806853584e-06,
+      "loss": 3.9374,
+      "step": 1577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6355140186915887e-06,
+      "loss": 4.1817,
+      "step": 1578
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6277258566978193e-06,
+      "loss": 4.136,
+      "step": 1579
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6199376947040499e-06,
+      "loss": 4.3139,
+      "step": 1580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6121495327102804e-06,
+      "loss": 4.2275,
+      "step": 1581
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.604361370716511e-06,
+      "loss": 4.2978,
+      "step": 1582
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5965732087227416e-06,
+      "loss": 4.1318,
+      "step": 1583
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.588785046728972e-06,
+      "loss": 4.0645,
+      "step": 1584
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5809968847352025e-06,
+      "loss": 4.0312,
+      "step": 1585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.573208722741433e-06,
+      "loss": 4.0709,
+      "step": 1586
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5654205607476636e-06,
+      "loss": 4.0713,
+      "step": 1587
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5576323987538942e-06,
+      "loss": 4.322,
+      "step": 1588
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5498442367601245e-06,
+      "loss": 4.0131,
+      "step": 1589
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.542056074766355e-06,
+      "loss": 3.9825,
+      "step": 1590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5342679127725857e-06,
+      "loss": 4.1651,
+      "step": 1591
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5264797507788162e-06,
+      "loss": 4.0182,
+      "step": 1592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5186915887850468e-06,
+      "loss": 4.0397,
+      "step": 1593
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5109034267912774e-06,
+      "loss": 4.2164,
+      "step": 1594
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5031152647975077e-06,
+      "loss": 4.1475,
+      "step": 1595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4953271028037383e-06,
+      "loss": 4.3975,
+      "step": 1596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4875389408099689e-06,
+      "loss": 4.2711,
+      "step": 1597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4797507788161994e-06,
+      "loss": 4.4522,
+      "step": 1598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.47196261682243e-06,
+      "loss": 5.1022,
+      "step": 1599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4641744548286604e-06,
+      "loss": 4.7542,
+      "step": 1600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.456386292834891e-06,
+      "loss": 4.0741,
+      "step": 1601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4485981308411215e-06,
+      "loss": 3.9343,
+      "step": 1602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.440809968847352e-06,
+      "loss": 4.0492,
+      "step": 1603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4330218068535826e-06,
+      "loss": 4.1779,
+      "step": 1604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4252336448598132e-06,
+      "loss": 3.9255,
+      "step": 1605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4174454828660435e-06,
+      "loss": 4.1838,
+      "step": 1606
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4096573208722741e-06,
+      "loss": 4.0761,
+      "step": 1607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4018691588785047e-06,
+      "loss": 3.9834,
+      "step": 1608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3940809968847352e-06,
+      "loss": 4.301,
+      "step": 1609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3862928348909658e-06,
+      "loss": 3.9533,
+      "step": 1610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3785046728971962e-06,
+      "loss": 4.311,
+      "step": 1611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3707165109034267e-06,
+      "loss": 4.0088,
+      "step": 1612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3629283489096573e-06,
+      "loss": 4.0759,
+      "step": 1613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3551401869158879e-06,
+      "loss": 3.9159,
+      "step": 1614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3473520249221184e-06,
+      "loss": 4.1505,
+      "step": 1615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3395638629283488e-06,
+      "loss": 4.024,
+      "step": 1616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3317757009345794e-06,
+      "loss": 4.1419,
+      "step": 1617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.32398753894081e-06,
+      "loss": 4.0914,
+      "step": 1618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3161993769470405e-06,
+      "loss": 4.1033,
+      "step": 1619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.308411214953271e-06,
+      "loss": 4.1666,
+      "step": 1620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3006230529595016e-06,
+      "loss": 4.1567,
+      "step": 1621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.292834890965732e-06,
+      "loss": 3.9118,
+      "step": 1622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2850467289719625e-06,
+      "loss": 3.9593,
+      "step": 1623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2772585669781931e-06,
+      "loss": 3.9724,
+      "step": 1624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2694704049844237e-06,
+      "loss": 4.1008,
+      "step": 1625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2616822429906543e-06,
+      "loss": 4.2112,
+      "step": 1626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2538940809968846e-06,
+      "loss": 4.1622,
+      "step": 1627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2461059190031154e-06,
+      "loss": 3.9802,
+      "step": 1628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.238317757009346e-06,
+      "loss": 4.2427,
+      "step": 1629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2305295950155765e-06,
+      "loss": 3.9442,
+      "step": 1630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2227414330218069e-06,
+      "loss": 4.1924,
+      "step": 1631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2149532710280374e-06,
+      "loss": 4.159,
+      "step": 1632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.207165109034268e-06,
+      "loss": 4.1241,
+      "step": 1633
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1993769470404986e-06,
+      "loss": 4.0827,
+      "step": 1634
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1915887850467291e-06,
+      "loss": 4.1648,
+      "step": 1635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1838006230529595e-06,
+      "loss": 4.0088,
+      "step": 1636
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.17601246105919e-06,
+      "loss": 4.1058,
+      "step": 1637
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1682242990654206e-06,
+      "loss": 4.1987,
+      "step": 1638
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1604361370716512e-06,
+      "loss": 4.0025,
+      "step": 1639
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1526479750778818e-06,
+      "loss": 4.1505,
+      "step": 1640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1448598130841123e-06,
+      "loss": 4.2192,
+      "step": 1641
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1370716510903427e-06,
+      "loss": 4.2231,
+      "step": 1642
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1292834890965733e-06,
+      "loss": 4.0601,
+      "step": 1643
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1214953271028038e-06,
+      "loss": 4.2302,
+      "step": 1644
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1137071651090344e-06,
+      "loss": 4.3653,
+      "step": 1645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.105919003115265e-06,
+      "loss": 4.5605,
+      "step": 1646
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0981308411214953e-06,
+      "loss": 3.974,
+      "step": 1647
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0903426791277259e-06,
+      "loss": 4.6767,
+      "step": 1648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0825545171339565e-06,
+      "loss": 4.4507,
+      "step": 1649
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.074766355140187e-06,
+      "loss": 4.1013,
+      "step": 1650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0669781931464176e-06,
+      "loss": 4.2395,
+      "step": 1651
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.059190031152648e-06,
+      "loss": 4.1179,
+      "step": 1652
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0514018691588785e-06,
+      "loss": 4.0829,
+      "step": 1653
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.043613707165109e-06,
+      "loss": 4.0633,
+      "step": 1654
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0358255451713396e-06,
+      "loss": 3.9353,
+      "step": 1655
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0280373831775702e-06,
+      "loss": 4.0386,
+      "step": 1656
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0202492211838008e-06,
+      "loss": 4.2134,
+      "step": 1657
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0124610591900311e-06,
+      "loss": 4.1343,
+      "step": 1658
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0046728971962617e-06,
+      "loss": 4.0052,
+      "step": 1659
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.968847352024923e-07,
+      "loss": 4.1671,
+      "step": 1660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.890965732087228e-07,
+      "loss": 4.2345,
+      "step": 1661
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.813084112149534e-07,
+      "loss": 4.2383,
+      "step": 1662
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.735202492211838e-07,
+      "loss": 3.8432,
+      "step": 1663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.657320872274143e-07,
+      "loss": 4.1784,
+      "step": 1664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.579439252336449e-07,
+      "loss": 3.9555,
+      "step": 1665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.501557632398755e-07,
+      "loss": 4.0736,
+      "step": 1666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.423676012461059e-07,
+      "loss": 3.9806,
+      "step": 1667
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.345794392523365e-07,
+      "loss": 4.0233,
+      "step": 1668
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.267912772585671e-07,
+      "loss": 4.1004,
+      "step": 1669
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.190031152647975e-07,
+      "loss": 4.0175,
+      "step": 1670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.112149532710281e-07,
+      "loss": 3.9954,
+      "step": 1671
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.034267912772586e-07,
+      "loss": 4.1891,
+      "step": 1672
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.956386292834891e-07,
+      "loss": 4.0017,
+      "step": 1673
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.878504672897197e-07,
+      "loss": 4.152,
+      "step": 1674
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.800623052959501e-07,
+      "loss": 4.3383,
+      "step": 1675
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.722741433021807e-07,
+      "loss": 4.0458,
+      "step": 1676
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.644859813084113e-07,
+      "loss": 3.9002,
+      "step": 1677
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.566978193146417e-07,
+      "loss": 4.1325,
+      "step": 1678
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.489096573208723e-07,
+      "loss": 4.2374,
+      "step": 1679
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.411214953271029e-07,
+      "loss": 4.2532,
+      "step": 1680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 3.8241,
+      "step": 1681
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.255451713395639e-07,
+      "loss": 3.8529,
+      "step": 1682
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.177570093457944e-07,
+      "loss": 4.344,
+      "step": 1683
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.099688473520249e-07,
+      "loss": 3.9101,
+      "step": 1684
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.021806853582555e-07,
+      "loss": 4.2004,
+      "step": 1685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.94392523364486e-07,
+      "loss": 4.1703,
+      "step": 1686
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.866043613707165e-07,
+      "loss": 4.0057,
+      "step": 1687
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.788161993769471e-07,
+      "loss": 4.2725,
+      "step": 1688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.710280373831776e-07,
+      "loss": 4.1855,
+      "step": 1689
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.632398753894081e-07,
+      "loss": 3.9922,
+      "step": 1690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.554517133956387e-07,
+      "loss": 4.3805,
+      "step": 1691
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.476635514018691e-07,
+      "loss": 4.1965,
+      "step": 1692
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.398753894080997e-07,
+      "loss": 4.4861,
+      "step": 1693
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.320872274143302e-07,
+      "loss": 4.4656,
+      "step": 1694
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.242990654205607e-07,
+      "loss": 4.4248,
+      "step": 1695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.165109034267913e-07,
+      "loss": 4.3967,
+      "step": 1696
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.087227414330218e-07,
+      "loss": 4.2427,
+      "step": 1697
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.009345794392523e-07,
+      "loss": 4.62,
+      "step": 1698
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.931464174454829e-07,
+      "loss": 4.6369,
+      "step": 1699
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.853582554517134e-07,
+      "loss": 4.5237,
+      "step": 1700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.775700934579439e-07,
+      "loss": 4.2145,
+      "step": 1701
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.697819314641744e-07,
+      "loss": 3.9877,
+      "step": 1702
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.61993769470405e-07,
+      "loss": 3.9597,
+      "step": 1703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.542056074766355e-07,
+      "loss": 3.9723,
+      "step": 1704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.46417445482866e-07,
+      "loss": 4.005,
+      "step": 1705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.386292834890966e-07,
+      "loss": 3.9488,
+      "step": 1706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.308411214953271e-07,
+      "loss": 3.9929,
+      "step": 1707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.230529595015577e-07,
+      "loss": 4.1335,
+      "step": 1708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.152647975077883e-07,
+      "loss": 4.284,
+      "step": 1709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.074766355140187e-07,
+      "loss": 4.2583,
+      "step": 1710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.996884735202493e-07,
+      "loss": 3.9819,
+      "step": 1711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.919003115264798e-07,
+      "loss": 4.169,
+      "step": 1712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.841121495327103e-07,
+      "loss": 4.3626,
+      "step": 1713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.763239875389409e-07,
+      "loss": 4.0682,
+      "step": 1714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.685358255451713e-07,
+      "loss": 4.0979,
+      "step": 1715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.607476635514019e-07,
+      "loss": 4.1014,
+      "step": 1716
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.529595015576325e-07,
+      "loss": 4.4454,
+      "step": 1717
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.451713395638629e-07,
+      "loss": 4.389,
+      "step": 1718
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.373831775700935e-07,
+      "loss": 3.8065,
+      "step": 1719
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.29595015576324e-07,
+      "loss": 3.9721,
+      "step": 1720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.218068535825545e-07,
+      "loss": 4.0922,
+      "step": 1721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.140186915887851e-07,
+      "loss": 4.0803,
+      "step": 1722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.062305295950156e-07,
+      "loss": 4.2579,
+      "step": 1723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.984423676012461e-07,
+      "loss": 4.2053,
+      "step": 1724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.906542056074767e-07,
+      "loss": 4.1518,
+      "step": 1725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.828660436137072e-07,
+      "loss": 4.1188,
+      "step": 1726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7507788161993773e-07,
+      "loss": 4.1883,
+      "step": 1727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6728971962616824e-07,
+      "loss": 4.3013,
+      "step": 1728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5950155763239876e-07,
+      "loss": 4.3411,
+      "step": 1729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.517133956386293e-07,
+      "loss": 4.1567,
+      "step": 1730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4392523364485984e-07,
+      "loss": 3.8544,
+      "step": 1731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3613707165109035e-07,
+      "loss": 4.0856,
+      "step": 1732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2834890965732087e-07,
+      "loss": 4.0963,
+      "step": 1733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2056074766355143e-07,
+      "loss": 4.2605,
+      "step": 1734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1277258566978195e-07,
+      "loss": 4.529,
+      "step": 1735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0498442367601246e-07,
+      "loss": 4.5911,
+      "step": 1736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.97196261682243e-07,
+      "loss": 3.8847,
+      "step": 1737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8940809968847354e-07,
+      "loss": 4.1024,
+      "step": 1738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8161993769470406e-07,
+      "loss": 4.119,
+      "step": 1739
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7383177570093457e-07,
+      "loss": 3.9846,
+      "step": 1740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.660436137071651e-07,
+      "loss": 4.1438,
+      "step": 1741
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5825545171339565e-07,
+      "loss": 4.1723,
+      "step": 1742
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5046728971962617e-07,
+      "loss": 3.9846,
+      "step": 1743
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.426791277258567e-07,
+      "loss": 3.9873,
+      "step": 1744
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.348909657320872e-07,
+      "loss": 4.1639,
+      "step": 1745
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2710280373831776e-07,
+      "loss": 4.3494,
+      "step": 1746
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.193146417445483e-07,
+      "loss": 4.2833,
+      "step": 1747
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1152647975077885e-07,
+      "loss": 4.1762,
+      "step": 1748
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0373831775700936e-07,
+      "loss": 4.31,
+      "step": 1749
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.959501557632399e-07,
+      "loss": 4.2182,
+      "step": 1750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8816199376947044e-07,
+      "loss": 3.9217,
+      "step": 1751
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8037383177570096e-07,
+      "loss": 4.0883,
+      "step": 1752
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7258566978193147e-07,
+      "loss": 4.3001,
+      "step": 1753
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.64797507788162e-07,
+      "loss": 4.0114,
+      "step": 1754
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5700934579439255e-07,
+      "loss": 4.2639,
+      "step": 1755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4922118380062307e-07,
+      "loss": 3.8722,
+      "step": 1756
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.414330218068536e-07,
+      "loss": 4.0116,
+      "step": 1757
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3364485981308412e-07,
+      "loss": 4.0707,
+      "step": 1758
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2585669781931466e-07,
+      "loss": 4.0759,
+      "step": 1759
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1806853582554518e-07,
+      "loss": 3.9751,
+      "step": 1760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1028037383177572e-07,
+      "loss": 4.2123,
+      "step": 1761
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0249221183800623e-07,
+      "loss": 3.9068,
+      "step": 1762
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9470404984423677e-07,
+      "loss": 4.0427,
+      "step": 1763
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8691588785046729e-07,
+      "loss": 4.2775,
+      "step": 1764
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7912772585669783e-07,
+      "loss": 3.8871,
+      "step": 1765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7133956386292834e-07,
+      "loss": 4.0579,
+      "step": 1766
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6355140186915888e-07,
+      "loss": 4.0284,
+      "step": 1767
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5576323987538942e-07,
+      "loss": 4.0003,
+      "step": 1768
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4797507788161994e-07,
+      "loss": 4.1002,
+      "step": 1769
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4018691588785048e-07,
+      "loss": 4.0588,
+      "step": 1770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.32398753894081e-07,
+      "loss": 4.0607,
+      "step": 1771
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2461059190031153e-07,
+      "loss": 4.0823,
+      "step": 1772
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1682242990654206e-07,
+      "loss": 4.1027,
+      "step": 1773
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0903426791277259e-07,
+      "loss": 4.1698,
+      "step": 1774
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0124610591900312e-07,
+      "loss": 4.0266,
+      "step": 1775
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.345794392523364e-08,
+      "loss": 4.1723,
+      "step": 1776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.566978193146417e-08,
+      "loss": 4.0087,
+      "step": 1777
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.788161993769471e-08,
+      "loss": 4.1692,
+      "step": 1778
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.009345794392524e-08,
+      "loss": 4.3147,
+      "step": 1779
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.230529595015577e-08,
+      "loss": 4.1073,
+      "step": 1780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.4517133956386294e-08,
+      "loss": 4.4933,
+      "step": 1781
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.672897196261682e-08,
+      "loss": 4.0406,
+      "step": 1782
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8940809968847356e-08,
+      "loss": 4.5157,
+      "step": 1783
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1152647975077883e-08,
+      "loss": 3.5732,
+      "step": 1784
+    },
+    {
+      "epoch": 1.0,
+      "step": 1784,
+      "total_flos": 0.0,
+      "train_loss": 4.241978131735806,
+      "train_runtime": 8769.6328,
+      "train_samples_per_second": 3.254,
+      "train_steps_per_second": 0.203
+    }
+  ],
+  "max_steps": 1784,
+  "num_train_epochs": 1,
+  "total_flos": 0.0,
+  "trial_name": null,
+  "trial_params": null
+}